Description
A stormwater management pond is an artificial pond that is designed to collect and retain urban stormwater. They are frequently built into urban areas in North America to also retain sediments and other materials.
Auckland Wellington Christchurch
PATTLE DELAMORE PARTNERS LTD
Drury South Business Project Stormwater Management Report Peer Review
Stevenson Group Limited
solutions for your environment
Drury South Business Project Stormwater Management Report Peer Review
• Prepared for
Stevenson Group Limited
• June 2011
PATTLE DELAMORE PARTNERS LTD Level 4, PDP House 235 Broadway, Newmarket, Auckland PO Box 9528, Auckland, New Zealand Tel +9 523 6900 Fax +9 523 6901 Web Site http: / / www.pdp.co.nz Auckland Wellington Christchurch
solutions for your environment
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
i
Quality Control Sheet
TITLE
Drury South Business Park – Stormwater Management Report Peer Review
CLIENT VERSION DATE JOB REFERENCE SOURCE FILE(S)
Stevenson Group Ltd Final 27 June 2011 A02426300
A02426300R001_DSPB SW Peer Review final 270611.doc
Prepared by
SIGNATURE
Mark Pennington
Andrew Kolper
Roger Seyb
Directed, reviewed and approved by
SIGNATURE
Peter Callander Limitations: The report has been prepared for Stevenson Group Ltd, according to their instructions, for the particular objectives described in the report. The information contained in the report should not be used by anyone else or for any other purposes.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
ii
Executive Summary
The Drury South Business Park (DSBP) is a proposed 360ha industrial and commercial development approximately 3km south-southeast of the Drury Township and extends from State Highway 1 on the western boundary to the Hunua foothills in the east. The development is expected to provide employment for approximately 7000 people when fully developed. Pattle Delamore Partners have been engaged to review the DSBP Stormwater Management Report and in particular comment on the Hydrological and Hydraulic Modelling and issues relating to stormwater quality and in stream erosion. In general the hydrological and hydraulic models have been constructed to an acceptable standard and would be expected to yield acceptable results. There is some uncertainty remaining on model calibration and validation, which is largely due to the lack of calibration data available for use. Model scenarios have been simulated for various design rainfall events for nominated Average Recurrence Intervals (ARI) events. The results from these simulations will probably differ from actual rainfall events of similar ARI. This is a common occurrence with the use of a design event methodology, which is an industry standard approach. The model results have shown some sensitivity to timing of peak flows from various contributing tributaries. Peak flows can be offset from one another in a “design” event, but under actual rainfall conditions or flow event of the same frequency this may not be the case. This can be due to variations in hydrological and hydraulic parameters across the catchment (rather than them being assumed constant in the design event approach). In response to questions, a model run using the 1988 real event scaled up to a 100 year ARI event has been carried out to check differences between the results of the TP108 100 year ARI design event and a real storm event. This showed a slight decrease in flood levels within the site area at the confluence of the Hingaia and Maketu streams indicating that the superposition of peaks does not appear to be an issue using either the design event or a real event within the Hingaia Stream. There was an increase in flood levels in the Drury township area using the real event - the reason for this is unknown and should be checked further. The LiDAR terrain data used in the modelling was quoted as being of accuracy to 0.25m in urban areas and 0.5m in rural areas. The use of LiDAR data is appropriate. However, where significant out-of-bank flooding has occurred it is likely that model accuracy could be no better than the accuracy of the terrain data. All model results should, therefore, be treated as having a range of accuracy, and an appropriate freeboard allowance applied. A check of the LiDAR data by Beca against a previous survey in the central part of the site indicates negligible difference between the LiDAR and surveyed data. There is existing flooding in the Drury township for rainfall events less than 100 year ARI. It is unknown whether the proposed development would result in changes to the
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
iii
frequency or extent of these flooding events as they have not been assessed. It is expected that provided flood storage is available to mitigate the effects of land-use changes for the 100 year ARI rainfall event, it should be possible to ensure that the flood level for a smaller (more frequent) rainfall event should not increase. The flood level for the pre-development land-use patterns may however be reached more frequently. Assessment of flooding effects in rainfall events less than the 100 year ARI is recommended and a design philosophy proposed not to increase the level or frequency of downstream flooding. Stormwater quality management in Auckland follows a Best Practicable Option, rather than an effects based, approach. The approach proposed includes source control, industrial contaminant management and stormwater treatment. This is appropriate. The appropriate design guideline for stormwater quality treatment is the former Auckland Regional Council’s Stormwater Management Design Guideline Manual, (TP10, 2003). Stormwater treatment will be in the form of six wetlands which will cater for the vast majority of the site and a bio-filtration swale for those areas not able to be treated through the wetland process. In terms of treatment, the wetlands are appropriately sized in accordance with TP10 and generally also provide a surface area greater than 2% of the catchment area (which will provide a better performance than a deeper wetland of equivalent volume). It is recommended that the project design criteria philosophy include that the wetland surface area be at least 2% of the catchment area. In addition to the proposed treatment methods, a District Plan Change has been proposed to restrict the types of roofing materials within the development. The Hingaia ICMP contaminant load modelling has assumed this plan change to be operational. This plan change is therefore an integral part of the development in terms of stormwater quality management and needs to be maintained as part of the project design philosophy. TP10 also provides an approach to manage instream channel erosion downstream of a discharge point by providing extended detention of runoff from the 34.5mm storm. The proposed stormwater management approach includes providing the required extended detention volume in accordance with TP10 for four out of the six sub-catchments. It is recommended that the design philosophy include managing the potential for downstream channel erosion. This could include varying the amount of extended detention provided in individual sub-catchments provided it is confirmed that the erosive effect on the downstream channel of providing extended detention in a sub-set of catchments rather than all sub-catchments is at least as effective.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
iv
Table of Contents
SECTION
PAGE
Executive Summary 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 2.0 2.1 2.2 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 4.0 4.1 4.2 4.3 4.4 5.0 Introduction Background Scope Terms of Reference Site Visit Relevant Reports Referenced Initial Queries and Comments Review Structure Stormwater Management Objectives Overall Stormwater Objectives Design Philosophy Hydrological Model Review Catchments and Sub-Catchments Rainfall Soils Loss Model Routing Model Third Party Models Calibration and Validation Hydraulic Model Review Boundary Conditions 1-D Component 2-D Component Calibration and Validation Flooding in response to rainfall events of ARI less than 100 years 6.0 6.1 6.2 7.0 7.1 7.2 7.3 7.4 Stormwater Quality Review On Site Stormwater Quality Extended Detention Volumes Conclusions Flood Assessment Flooding during events of ARI less than 100 years Stormwater Quality Extended Detention
ii 1 1 1 1 2 2 2 3 3 3 3 4 4 4 5 5 6 6 7 8 8 8 10 11
11 12 12 14 15 15 16 16 16
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
v
8.0 8.1 8.2 8.3 8.4
Recommendations Hydrologic and Hydraulic Model Flooding during events less than 100 year ARI Stormwater Quality Extended Detention
16 16 17 17 17
Appendices
Appendix A: Site Visit Record Appendix B: Questions to, and responses from, DHI Appendix C: Record of Teleconference held on 28 April 2011
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
1
1.0
1.1
Introduction
Background
Pattle Delamore Partners Ltd. (PDP) has been engaged by Stevenson Group Limited to undertake a desktop Peer Review of the Drury South Business Park (DSBP) Stormwater Management Report which was prepared by Beca Infrastructure Ltd (Beca) in support of a Plan Change Application, Metropolitan Urban Limit Change Application and a Network Discharge Consent Application for the proposed DSBP. The Beca Stormwater Management Report incorporates Mike Flood modelling carried out by DHI for Stevenson Group Ltd. The objective of the review was to comment on the appropriateness of the report in terms of flood plain management, hydrological changes, stormwater quality and stream erosion management, within the framework that is described below.
1.2
Scope
The scope of the review was to: ? Assess the hydrological and hydraulic modelling assumptions, parameters and results; ? Assess the approach used in identifying and assessing the stormwater quantity and quality issues and the appropriateness of key assumptions; ? Assess the assumptions and conclusions drawn as well as the proposed avoidance / mitigation measures proposed; and, ? Assess the coverage of relevant council regulations and issues against the purpose of the report. This review does not specifically addressed stormwater network, overland flow or stream re-alignment issues. The scope is outlined in greater detail in the brief forwarded to Beca on 1 April 2011.
1.3
Terms of Reference
The Peer Review which was undertaken is a simplified version of the Design Review as detailed in the IPENZ PN02 Practice Notes. The simplified design review entailed reviewing the Beca DSBP Stormwater Management Report as well as the DHI Modelling reports (Modelling for Hingaia Stream ICMP and Drury South Business Park, Hingaia Stream Flood Hazard Assessment) and commenting on their content. The review has been based on a review of reports and supplied responses to questions only. No model files have been supplied nor reviewed. This means that no checking has taken place to ensure that what has been stated in model reports has been accurately applied in the model files used. PDP have also not viewed any model result files, so checks for issues associated with model stability have not been undertaken. Rather, outputs taken from model result files,
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
2
such as peak water levels, longitudinal sections and flow rates as quoted in the model reports (rather than as taken from result files) are the subject of this review. This peer review does not cover any subsequent iterations of the review process (i.e. work carried out in response to recommendations after 24 th June 2011).
1.4
Site Visit
On 5 April 2011 PDP staff undertook a site visit of the area for the proposed Drury South Business Park and its surrounds. During the site visit a number of key features in terms of hydraulic performance were located and viewed, and an appreciation of the scale of the site was attained. A record of this site visit was kept, and this is provided in Appendix A.
1.5
Relevant Reports Referenced
Beca provided PDP with two separate modelling reports that were required to be reviewed under the scope of the work reported on in this document. These two modelling reports are:
?
Modelling for Hingaia Stream Integrated Catchment Management Plan , prepared for Papakura District Council / Franklin District Council by DHI Water and Environment Limited, 8 July 2010
?
Drury South Business Project Hingaia Stream Flood Hazard Assessment , prepared for Stevenson Group by DHI Water and Environment Limited, March 2011. The version supplied for review was a “final draft”, but subsequent correspondence (see Appendix B) indicated that this could be taken as a “final” report
In addition to the modelling reports, further reference material was used in the peer review process and included:
?
DSBP Stormwater Management Report , Prepared for Stevenson Group by Beca, March 2011
?
Hingaia Stream Integrated Catchment Management Plan, Prepared for Papakura District Council / Franklin District Council by Golder Associates, August 2010
?
Technical Publication 10 (TP10), Auckland Regional Council, 2003
1.6
Initial Queries and Comments
The initial phase of the review resulted in a series of questions and comments being submitted to DHI, to aid the reviewer’s understanding of the hydrological and hydraulic model build, calibration and scenarios investigation. The complete list of comments and questions, with answers as supplied by DHI, is given in Appendix B. Following distribution of this initial list of comments and questions, a teleconference between relevant parties was held. The purpose of this was to discuss the review and to establish the way forward, particularly with respect to any additional model runs that
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
3
needed to be undertaken. Following the teleconference PDP prepared a summary of this that was distributed by email. This email message is appended in Appendix C, and no comment on this summary was received from any of those present at the teleconference. As a result PDP assumed this to be a fair and accurate record. A further meeting was held on 16 th June 2011, at which review comments were further discussed and as a result of which DHI were instructed to run a scaled up version of a 1988 event through the model.
1.7
Review Structure
The review was broken up into review of each of the separate elements relating to: ? ? ? ? ? Stormwater Management Objectives. Hydrological Model Review. Hydraulic Model Review. Flooding in response to rainfall events of ARI less than 100 years. Stormwater quality review.
A convenient split between hydrological model (that used to simulate the rainfall-runoff process) and hydraulic model (that used for hydraulic computation) has been made. Recommendations in relation to the stormwater management report are made in association with each of the above topics.
2.0 Stormwater Management Objectives
2.1 Overall Stormwater Objectives
The Hingaia Stream ICMP Objectives are listed in Sections 11.2-11.6 of the ICMP and replicated in section 5.5 of the DSBP Stormwater Management Report. The DSBP has its own set of site stormwater objectives as set out in Sections 5.1-5.3 and design criteria in Section 1.4 of the DSBP Stormwater Management Report. Review Comment The overall stormwater management objectives adopted by DSBP are in general agreement with the Hingaia ICMP stormwater objectives. The design criteria adopted in Section 1.4 generally address normal stormwater management issues.
2.2
Design Philosophy
Beca has further developed the stormwater objectives into a design philosophy found in Section7.1 of the DSBP Stormwater Management Report.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
4
Review Comment The design philosophy as adopted for the DSBP development is generally based on industry best practice. Some additional design philosophy items are recommended and these are summarised at the end of this report.
3.0 Hydrological Model Review
3.1 Catchments and Sub-Catchments
The review of model catchments and sub-catchments and their respective connections to the relevant model was conducted based on the supplied drawings. Catchment delineation is stated to have been indicated erroneously in the report (see Comment 1 in Appendix B), but assurance has been given that this was corrected in the model. As mentioned in Section 1.1 PDP have not viewed nor reviewed any model files, so cannot confirm nor refute this. Sub-catchment delineation has been performed at a resolution that would fit with standard practice for the level of accuracy in results required to achieve the purpose of the modelling work. While the proposed DSBP development represents a significant change in landform, it was found that changes to the sub-catchment layout used for the ICMP Modelling (July 2010) could be applied to the “post-development” scenario as described in the subsequent report (March 2011). Review Comment There are no obvious errors, omissions or shortcomings in the catchment and subcatchment layout as reported on in the relevant documents. The catchment split has been made to a resolution that is suitable to meet the required level of detail.
3.2
Rainfall
Rainfall data from a range of recorders were used in the modelling. Most notably, recorded data from two gauges (Ngakora and Filter Station) were used in simulation of the validation (1988) event (see response to Question 4 in Appendix B). It is understood that an appropriate weighting for each of these gauges was used in simulation of this event, although this has not been independently verified. For design event simulation, rainfall data were taken from the guideline “Stormwater Runoff Modelling in the Auckland Region”, TP108 (ARC), and while this guideline is in the process of review, appropriate data were used given the time at which the modelling was undertaken. A nested “Chicago Storm” hyetograph was used for design event simulation, and this is the appropriate methodology in this case. As the selected hyetograph has 24 hour duration it is important that the catchment under analysis does not have response time or critical duration any longer than this, and this is definitely the case for the models that have been used.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
5
A response to Comment 13 in Appendix B highlights the likely difference in performance of the system (and model) in response to recorded as opposed to design rainfall. Review Comment The rainfall data used have been appropriately derived and applied.
3.3
Soils
Differences in surface soil types have been identified and taken into account in the modelling. Soil type is relevant in that it influences the rate at which rainfall is able to infiltrate, which has a large bearing on the rate of runoff generation in response to rainfall. While five different soil drainage classes were identified within the catchment, this was simplified with only two having been applied in the model. The total runoff volume for a recorded event (1988) of 40-50 year Average Recurrence Interval (ARI) was compared to that from a design event of 100-year ARI for a range of different infiltration rates (see Comments 11 and 12 in Appendix B). This was done to test the sensitivity of the model to the assumptions surrounding infiltration rates. In addition, increases to model infiltration rates were tested, although similar decreases were never applied (see Comment 21 in Appendix B). The response to Comment 19 in Appendix B indicates some comparison with applied infiltration rate from a nearby (calibrated) catchment had been made. Whilst the results of this comparison are not stated it is assumed that any wide discrepancy would have been highlighted and taken into account. This response also indicates that a further revision to the model may be made, using data from a flow monitoring site although this is not confirmed. The response to Question 7 in Appendix B also indicates that results from other hydrological studies were applied to the one in question, adding confidence to the apparent “assumptions” made. Review Comment The variation in soil type across the catchment has been taken into account, albeit in a simplified manner. Sensitivity testing has been undertaken on infiltration rate, and values have been compared to those derived from nearby catchments. The values applied appear to fall within what is regarded as a reasonable range. It appears that further work in establishment of design values for key parameters (such as infiltration rate) would aid with model confidence.
3.4
Loss Model
The hydrological model used is known as “Model B” in DHI software terms. The loss model component of this is based on an initial loss due to factors such as depression storage and initial wetting, and an Hortonian infiltration rate decay with time. This provides relatively few parameters to be used for model calibration, many of which can be physically measured.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
6
The key to understanding the applicability of the overall hydrological method applied lies in calibration and validation (refer Section 3.7). Variations for parameters within the loss model are not expected to be significant for infrequent (very large) rainfall events as the proportion of infiltration to total rainfall depth is relatively small. The effect on runoff is however expected to be more pronounced for more frequent rainfall events. Review Comment Whilst the loss model applied is a standard approach, this does not fit with that suggested in the guideline TP108 document that was used for rainfall (this uses the SCS Curve Number loss model). In this way two different hydrological methods have been integrated, which raises the question of applicability. It is recommended that the loss model be re-assessed and then calibrated and verified for more frequent (smaller) rainfall events.
3.5
Routing Model
The routing model applied is a part of the DHI “Model B” approach, which utilises the Kinematic Wave routing model. This is based on physically measurable data (slope, length, area) together with some derived parameters (e.g. Manning’s n ) that have been well-established in the literature. Review Comment Whilst the routing model applied is a standard approach, this does not fit with that suggested in the guideline TP108 document that was used for rainfall (this uses the SCS Unit Hydrograph routing model). In this way two different hydrological methods have been integrated, which raises the question of applicability. The values applied to the relevant parameters in the routing model seem to have been appropriately derived (see the response to Comment 7 in Appendix B). The key to understanding the applicability of the overall hydrological method applied lies in calibration and validation, and this point is discussed in Section 3.7.
3.6
Third Party Models
The Ngakora Stream and Slippery Creek are understood to have been modelled by a third party, with results from this having been used by DHI in the Hingaia Stream modelling. No independent checks of the methodology nor input data applied to these other models have been carried out, and DHI have indicated a lack of detail on this (see response to questions 6 and 20 in Appendix B). PDP have not reviewed any report pertaining specifically to this model. However, as these two waterways are located somewhat downstream of the specific area of relevance to the proposed Drury South Business Park development, there is likely to be relatively low sensitivity of proposed mitigation measures to accuracy of these models.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
7
Review Comment The effects of this part of the model potentially not having been built to the same set of assumptions and underlying methodology are likely to be small.
3.7
Calibration and Validation
Hydrological model validation appears not to have been undertaken in isolation from the hydraulic model, but rather both were essentially validated together. A lack of measured data has prevented full calibration from being able to be done. Validation was done by comparison of modelled against measured/estimated levels and flows at various points in the stream network. The 1988 event was selected for this. The hydrological model used was adjusted to reflect land use in the catchment at the time of the event (see response to Question 13 in Appendix B). The hydraulic model was adjusted (channel dimensions) until modelled and observed data exhibited a good match. Review Comment As stated in Comment 9 in Appendix B, the calibration approach seems to assume accuracy in the hydrological model, with channel dimensions being adjusted to achieve a good fit with observed data. This is because it is known that channel dimensions during the 1988 event were different from those used in the model, although it is not known how different these were. This approach is not standard (nor is it endorsed as part of this review) for model calibration, although the results were able to demonstrate a reasonable fit. In response to questions, a further model run was carried out by DHI using the 1988 rainfall event. The flow at the downstream end of Hingaia Stream from the 1988 event was scaled up to match that from the 100 year design rainfall event. The resulting flood levels were slightly lower in the upper catchment and the confluence of the Hingaia and Maketu Streams in the centre of the site and slightly higher in the lower catchment in Drury township. The results are also reported by DHI to show flatter hydrographs - which could increase the likelihood of superposition of peaks (or near-peaks) within the system, Within the Hingaia Stream catchment, the superposition issue is most likely to be an issue at the confluence of the Maketu and Hingaia Streams. The drop in flood levels at this confluence indicates that the design event is slightly conservative, at least with respect to the real event available. There is an increase in flood levels using the real event lower down in the catchment at Drury Town. The reason for this is unexplained and should be checked further. Review Comment Superposition of peaks in the scaled 1988 event does not appear to be an issue at the confluence of the Hingaia Stream and the Maketu Stream. There is an unexplained increase in the flood levels (of 70mm) in the Drury township area using the scaled 1988 event compared to the design event. The reason for this increase needs to be investigated further.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
8
4.0 Hydraulic Model Review
4.1 Boundary Conditions
The downstream tidal boundary was set at a steady Mean High Water Springs (MHWS) level for all simulations. This is a conservative approach because it is known that tidal fluctuations over the course of the 24-hour rainfall events used for design simulations will occur. Question 1 in Appendix B alludes to this, and in response DHI provided a longitudinal section of the lower reach of the model. From this it is clear that given the water levels involved, the bed slope of the channel and the distance from the downstream boundary to any areas of specific interest, there is unlikely to be significant model sensitivity to this downstream boundary water level. Inflow boundaries were derived from the hydrological modelling process. Some boundary inflows occur as point inflows and others are distributed, according to the physical system as it exists. In the July 2010 report it was not clear whether or not an inflow from the Ngakora Stream was used for all simulations, but the response to Question 9 in Appendix B indicates that this was the case. The actual boundary time series used for the Ngakora Stream inflow was derived by a third party and, as mentioned in Section 3.6, this has not been reviewed by PDP. The same applies to the Slippery Creek inflow and associated model. For the start condition of the model it was assumed that the wetland areas were initially dry. This fits with a classic “design storm” approach which is generally viewed independently of antecedent conditions. Review Comment The boundaries generated by hydrological modelling appear to have been applied appropriately to the hydraulic model. No independent checking of model files to ensure the time series are connected at the correct locations in the hydraulic model has been carried out, although it is assumed that this would have formed a distinct part of the internal review process by DHI. The downstream tidal boundary used is appropriate, although the source information of the actual level applied (stated as MHWS) has not been independently verified.
4.2
4.2.1
1-D Component
Discretisation
The one-dimensional model has been discretised into branches and reaches. The computational grid has been set within these to 10m in most cases (as reported by DHI). This provides a level of detail in both computation and in results interpretation that suits (if not exceeds) the minimum needs for the model results. Review Comment Model discretisation is appropriate for the investigation undertaken.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
9
4.2.2
Structures
Physical checks of structures to ensure their correct representation in the model have not been carried out as part of this review process. The bridge modelling approach used is a widely accepted approach and is suitable for the purpose to which it was applied. Culvert structures are understood to have been used for all open channel – closed conduit transitions, and this correctly ensures that energy losses occur at entrance and exit. The response to Question 10 in Appendix B reveals that appropriate values for entrance and exit loss coefficients have been used (while these do not, theoretically, remain constant over a range of flows and upstream and downstream depths, this is an acceptable approximation). Review Comment Representation of structures is suitable for the purpose of the model.
4.2.3
Channels and Pipes
Only the low flow channels have been represented in the 1D model, as all overland flows have been taken into account in the 2D component. This is considered appropriate. In the absence of detailed calibration data, selection of roughness to be applied to the model reaches is subject to judgement and by reference to standard texts. The values selected appear to fit within what would be considered “reasonable” ranges for the surface types encountered. Smaller pipes have been omitted from the model (see Comment 6 in Appendix B). This is a simplification that has little bearing on results of extreme flood events, but may have a small local influence in less severe events. Comment 8 in Appendix B describes some possible “double-counting” of volume above the bank level within the 1D portion of the model. The response from DHI indicates the effect of this to be negligible. Review Comment The response to Comment 6 in Appendix B indicates a willingness to recalibrate the model using a smaller event, and this forms one of the recommendations of this review report. It is agreed that any potential “double-counting” of flood volume above the 1D cross sections is negligible in terms of overall results.
4.2.4
Cross Sections
Channel cross sections have been derived from survey where possible and from (adjusted) LiDAR data elsewhere. This is an appropriate approach, giving greater weight to the surveyed cross sections than the LiDAR data wherever the two overlap. The report correctly states that LiDAR cannot be relied upon for incised channels, for heavily wooded areas and for areas under water at the time of the survey.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
10
Review Comment The way in which cross sections were derived is appropriate for the terrain.
4.2.5
Hydrodynamic Parameters
For the Mike11 (1D) component of the model the response to Question 16 in Appendix B indicates that “default” values for hydrodynamic parameters were used, except for the coefficient “Delta”. This was increased from a default value of 0.5 to a value of 0.8 to aid with model stability. It is noted that the Mike 11 Hydrodynamic Reference Manual produced by DHI states that “Changing the value of DELTA from its default value should only be carried out when “short term” fluctuations are unimportant, as in the case of a wave with period of several days or longer”. This appears to conflict with what was done in the model, however this is relatively common practice. Review Comment While this appears to conflict with the literature produced by DHI for this software the practicalities of attempting to run a model such as this with Delta set to 0.5 may be excessively onerous. It is accepted that some accuracy may be lost due to damping out potential oscillations (as changing the value of Delta can do), but in terms of overall results this is likely to be insignificant given other uncertainties that also exist. Account of these uncertainties is usually made by application of a freeboard allowance, as has been done in this case.
4.3
4.3.1
2-D Component
Discretisation
It is understood that the terrain used for the 2D component of the model was derived from a LiDAR survey of the subject area. A 5mx5m grid was used which is understood to have been supplied to DHI. The vertical accuracy of the LiDAR data is stated to be to 0.5m in rural areas and 0.25m in urban areas. The distinction between rural and urban in this context is not clear. The response to Question 2 in Appendix B indicates that raw LiDAR point spacing is not known, nor if this was suitably dense to support the chosen grid resolution. Where there is a lack of suitable survey calibration data, there may be a general translation of the terrain data above or below the local survey benchmarks. In response to a question on this, Beca checked the terrain data against a survey they have previously carried in the central area of the site and found a negligible difference between the two datasets. Review Comment Experience with other LiDAR data sets indicates that a 5mx5m grid should easily be able to be supported. However mention in the July 2010 report is made of a 1mx1m grid, and this would require significantly more detailed LiDAR survey in order to not provide artificial
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
11
accuracy. The process followed by DHI in setting up the 5mx5m grid appears to be robust. Further spot checks on terrain data versus local survey benchmarks should be carried out through the design process, or, checks carried out at construction stage to ensure that the required flood storage volume below existing ground level has been provided.
4.3.2
Links
The 1D and 2D components of the model are connected using lateral links. No detail of these was provided in the reports that were reviewed. The response to Questions 11 and 12 in Appendix B indicates that there was a methodical approach applied to setting spill levels and in ensuring that the data made sense. Review Comment As the model files were not reviewed the locations and levels associated with these links were not independently verified.
4.3.3
Parameters
Model parameters were not specified in the reports, but the response to Question 16 in Appendix B has indicated that appropriate values were used. Review Comment As the model files were not reviewed numerical values applied to various parameters were not checked.
4.4
Calibration and Validation
As mentioned in Section 3.7 no suitably detailed calibration data were available at the time of the model build. The 1988 event was used for validation, using relatively sparse observations of water levels. Review Comment As stated in Comment 9 in Appendix B, the calibration approach seems to assume accuracy in the hydrological model, with channel dimensions being adjusted to achieve a good fit with observed data. This is because it is known that channel dimensions during the 1988 event were different from those used in the model, although it is not known how different these were. This approach is not standard for model calibration, although the results were able to demonstrate a reasonable fit.
5.0 Flooding in response to rainfall events of ARI less than 100 years
The DSPB Stormwater Management Report presents results of the modelling to show that the post-development scenario flooding in response to a 100-year design rainfall event is no worse than for the current level of development in the subject area. No assessment has been undertaken to determine the extent of flooding during more frequent (i.e. lower
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
12
ARI) events. Possible effects in response to more frequent events could include changing flooding levels / extent and /or increasing the frequency of flooding at a given flood level. There is existing flooding above floor levels in the downstream Drury Township under current development intensity in response to rainfall events of ARI less than 100 years and some buildings may be affected in events less than 10 years. The effects of the proposed DSBP on existing flooding levels and frequency should be assessed. This would typically be carried out for the design event associated with sizing stormwater network infrastructure (10 years) and, given flooding may occur for smaller events, a smaller event such as the 2 year ARI. Review Comment The effect of the proposed mitigation measures on the downstream environment is unknown for flood events of ARI less than 100 years. It is recommended that changes resulting from the proposed development be assessed for the 10 year and 2 year ARI events (with and without climate change allowances) for the existing and Maximum Probable Development scenarios. As noted in Section 3.4, the loss model used may have a greater effect on more frequent (smaller) events and the overall model, therefore, needs calibration for these events. The design philosophy should include that there be no increase in the frequency or degree of flooding of properties in response to both 100 year and 10 year ARI rainfall events. If this is not possible off-site mitigation should be provided.
6.0 Stormwater Quality Review
6.1
6.1.1
On Site Stormwater Quality
General
The DSBP Stormwater report details options for stormwater point source treatment such as swales, rain gardens, sand filters, proprietary devices and others and gives an overall treatment train approach for stormwater quality management including: • • • Source control of metals from roofs; Industrial and trade process contaminant management; Stormwater quality treatment devices.
The contaminants of concern for the catchment are identified within the Hingaia ICMP are sediments and other contaminants. No site specific contaminants of concern are identified, but these are expected to be sediment, metals, hydrocarbons and industry specific contaminants. Review comment An overall treatment train approach for managing stormwater contaminants is good practice and the one proposed should address the expected contaminants of concern.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
13
6.1.2
Treatment of Metals
In addition to the point source treatment and wetlands proposed, a plan change has been proposed to prevent the use of bare metal and galvanised roofs. This is expected to significantly reduce the load of metals in stormwater from the site. We note that the contaminant load modelling carried out for the Hingaia ICMP (Stormwater Management Report Section 7.5.3) assumed that the industrial roofs will be those typically available today – which could therefore include unpainted zinc-alum roofs. If the roof areas within the DSBP were zinc-alum the expected contaminant yield would be higher than that for pre-painted steel. The plan change proposed to control roof material type to pre-painted steel roofing therefore means that the contaminant load from the DSBP roofs is less than that may have been calculated in the ICMP modelling. Review Comment The reduction of metal contaminant loads at source by avoiding exposed metal type roofing materials is considered an appropriate design philosophy. An alternative to this may be that additional on-site treatment is provided for roof run-off from any exposed metal roofing.
6.1.3
Industrial and Trade Processes
In addition to the wetland approach to stormwater treatment, medium and high risk sites as defined in the Auckland Regional Plan: Air, Land, Water (ALW Plan) will require specific (ITP) consents for the management, treatment and discharge of stormwater. These sites would utilise the specific point source treatment methods described within the DSBP Stormwater report and Environmental Management Plan requirements set out in the ALW Plan. Review Comment The design philosophy should include that specific on-site stormwater treatment measures will need to be designed to comply with the objectives of the Hingaia Stream ICMP, the DSBP Stormwater Management Report and the rules in the ALW Plan. Environmental Management Plans will need to meet the requirements of the ALW Plan.
6.1.4
Treatment Sizing
The on site stormwater treatment proposed for the business park was assessed on its adherence to the former Auckland Regional Council’s Stormwater Management Devices Design Guideline, Technical Publication No. 10, 2003 (TP10). The Best Practicable Option (BPO) approach has been used to select the treatment method. The BPO, based on an analysis of the treatment choices for the site, was the construction of a number of wetlands to treat the runoff from discrete catchments within the greater site areas.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
14
In addition, a small sub-catchment (Area 7) will utilise an engineered biofiltration swale to treat the stormwater runoff from the site. No allowance for extended detention or attenuation of larger storm events has been provided in the swale. The TP10 design methodology is based on the removal of total suspended solids and, by association, metals attached to those sediment particles. The removal of dissolved metals is promoted through the use of vegetative practices (such as wetlands). Wetlands are a preferred method within TP10 to target the removal of sediment and metals. Within wetlands, metal removal is promoted by longer vegetative contact and there is greater opportunity for this within systems with large shallow vegetated areas with low through flow velocity and minimal short-circuiting. Beca report that the surface area to catchment area ratio of the individual wetlands proposed ranges between 1.75% and 4.0%, with the average more than 2% of the catchment area to the wetlands. This is beneficial and will help to optimise the wetland performance for the required storage volume. Review Comment From the information provided within the DSBP Stormwater Management report, the water quality volumes (WQV) which will be provided for each of the six catchments (as set out in report) are in general accordance with the requirements of TP10. Wetlands typically perform better when there are more extensive shallow areas than deep areas – as is the case for the wetlands proposed. It is recommended that the design philosophy include that the surface area of the wetlands be at least 2% of the catchment area draining to each wetland.
6.2
Extended Detention Volumes
TP10 recommends the use of an extended detention volume (EDV), in addition to WQV requirements, to manage downstream channel erosion. The EDV requirement is to detain runoff from the 34.5mm rainfall event and release that volume over a 24 hour period. As noted in Table 7 of the DSBP Stormwater Management Report, if a wetland provides for EDV, the required WQV is reduced by one half. Two of the six wetlands (Labelled Nos. 3 & 4) do not provide sufficient volume to comply the EDV requirement as set out in TP10. As explained within Section 8.2 of the DSBP Stormwater Management Report, the reasoning behind the omission of EDV on the two wetlands was a lack of available volume and a reduction in modelled downstream velocities when runoff from the site was routed directly into the streams. Review Comment The design philosophy should include the management of downstream channel erosion. Where the Extended Detention Volume cannot be provided in accordance with TP10, checks should be carried out that downstream erosion will not be exacerbated by the omission of the EDV within some wetlands (for example the length of time that channels are exposed to erosive velocities before and after development).
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
15
7.0 Conclusions
7.1
7.1.1
Flood Assessment
Hydrological Model
In general the hydrological model has been built using generally accepted standards and principles. There has been some mixing of methods in that the Chicago Storm rainfall hyetograph as specified in TP108 was used together with DHI Model B for the loss and routing models (where TP108 suggests the SCS method). A lack of calibration data has meant that the hydrological model remains uncalibrated, although reference has been made to work done on neighbouring catchments that gives some confidence in the values for various parameters that have been applied.
7.1.2
Hydraulic Model
The hydraulic model has also been built to acceptable standards, using widely accepted data and methods of representation. While the model reports reviewed did not contain all of the information required to make this assessment, the answers to subsequent questions on this have augmented the information in the model reports to the extent that adequate review could be undertaken.
7.1.3
Results
Independent spot checks of modelled peak discharge have been carried out, confirming that reasonable values have been produced through the modelling process. Model results have indicated a reliance on offsetting of flood peaks from contributing catchments in order to keep peak flood levels below threshold levels. This off-setting effect has been shown to be present in model results for design rainfall events using the synthetic Chicago Storm hyetograph and at the confluence of the Maketu and Hingaia Streams using a scaled 1988 flood event (i.e super-position of flood peaks within the Hingaia Stream is not a problem in the scenarios assessed). Use of the scaled 1988 flood event showed a slight increase in downstream flood levels at Drury township and a decrease in flood levels upstream of the confluence of the Hingaia Stream and the quarry diversion. The increase in flood levels using the real event lower down in the catchment at Drury Town is unexplained and should be checked further. Flood levels have been quoted in both reports that were reviewed. Freeboard allowance is a key component of the approach to address potential variation in model inputs, in setting finished development levels above flood level estimates. A check on the difference between “average” LiDAR terrain levels and a survey within the central part of the site indicated negligible difference in levels. Further checks should be carried out during the design stage, or, checks should be carried out at construction
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
16
stage to confirm that the required flood storage is provided below existing ground level (i.e. cut below existing ground to achieve a volume not a given level).
7.2
Flooding during events of ARI less than 100 years
The effect of the proposed mitigation measures on the downstream environment is unknown for flood events of ARI less than 100 years. It is recommended that this be assessed for the 10 year and 2 year ARI events (with and without climate change allowances) for the existing and Maximum Probable Development scenarios. The design philosophy should include that there be no increase in the frequency or degree of flooding of properties in response to both 100 year and 10 year ARI rainfall events. If this is not possible off-site mitigation should be provided.
7.3
Stormwater Quality
The stormwater quality treatment devices have been sized in general accordance with TP10. The wetlands proposed are appropriately shallow and will therefore promote contaminant removal. The proposed plan change restricting roof materials within the DSBP is an integral part of the Hingaia ICMP and the overall approach to the management of stormwater quality from the site. The design philosophy should include that specific on-site stormwater treatment measures will need to be designed to comply with the objectives of the Hingaia Stream ICMP, the DSBP Stormwater Management Report and the rules in the ALW Plan. Environmental Management Plans will need to meet the requirements of the ALW Plan.
7.4
Extended Detention
The proposed EDV was not fully in accordance with the requirements set out in TP10. Specifically, additional extended detention in wetlands 3 and 4 would be required to comply with the TP design approach. In discussions with Beca, the need for EDV for each sub-catchment was questioned as the preliminary modelling showed a decrease in stream velocities. The design philosophy for the project should include that there is no increase in downstream channel erosion and the effect of not providing EDV in each wetland on the downstream channel should be checked (for example by checking the length of time that channels are exposed to erosive velocities before and after development).
8.0 Recommendations
8.1 Hydrologic and Hydraulic Model
There are two recommendations relating to models and their accuracy:
?
The increase in flood levels using the real event lower down in the catchment at Drury Town is unexplained and should be checked further.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
17
?
Re-assess the loss model and then calibrate and verify the model for more frequent (smaller) rainfall events
8.2
Flooding during events less than 100 year ARI
Investigation of the potential for flooding during events less than the 100 year ARI is recommended. The design philosophy should include that there be no increase in the frequency or degree of flooding of properties for a 100 year or 10 year ARI rainfall event. If this is not possible mitigation should be provided.
8.3
Stormwater Quality
The design philosophy should include that specific Industrial and Trade on-site stormwater treatment measures will need to be designed to comply with the objectives of the Hingaia Stream ICMP, the DSBP Stormwater Management Report and the rules in the ALW Plan. Environmental Management Plans will need to meet the requirements of the ALW Plan. It is recommended that the stormwater management design philosophy include that each wetland has a permanent water surface area of at least 2% of its sub-catchment area.
8.4
Extended Detention
The design philosophy for the project should include that there is no increase in downstream channel erosion. The effect of not providing the full EDV in each wetland on the downstream channel should be checked (for example by checking the length of time that channels are exposed to erosive velocities before and after development) and, if required, achieve the design philosophy by other means.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
18
Appendix A: Record of Site Visit by PDP PDP visited key points of the Business Plan area on 5th April 2011 to gain an appreciation of the existing land-use, current hydrological and hydraulic conditions and the potential effects of the development proposal on the surface water regime. The key points visited included: • • • • • Hingaia Stream at Ararimu Rd; Maketu Stream at Dale Road; Hingaia Stream at Quarry Road; Low lying areas in the Drury township; An overview of the development area from MacWhinney Drive.
Hingaia Stream at Ararimu Rd
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
19
Hingaia Stream at Quarry Rd
View overlooking central part of the Business Park from MacWhinney Road
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
20
Appendix B: Questions and Comments with Response from DHI PDP supplied DHI (copy to Beca) a list of comments and questions that arose from the initial review of the reports supplied on 19th April 2011. These questions and comments were supplied in order to seek a better understanding of the modelling work that was done and to augment that information that was contained within the supplied reports. In addition to several specific questions and comments listed, PDP supplied three overall comments on the modelling work reported on in the documents that were reviewed. These are stated in the excerpt from the email message below (only the first page of this email - containing the three comments is reproduced):
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
21
Responses to the initial list of comments and questions were supplied via email from DHI on 3 rd May 2011. This email message is reproduced below, with the text of the response to follow on subsequent pages. Note that every initially posed question or comment is numbered, with the response from DHI immediately below each. Several additional questions and comments, that followed the list initially compiled, are shown in red. Collectively these make up the complete list of questions and comments mentioned in Section 1.6.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
22
Responses to PDP’s comments and questions
Section 1: Comments 1. In Figure 1-1 the catchment boundary appears to cross a stream at two locations (approx midway along western boundary and in north-east corner). This appears anomalous. Is there an explanation for this? The stream polyline used for this drawing is incorrect (not the catchment boundary). The stream polyline was included in the data supply but does not account for culverts through the motorway which we have accounted for in our catchment delineation. 2. Section 1.5: I have not seen the GAP analysis report. I presume that the Technical modelling report is contained within the document referenced above. The gap analysis report was delivered at an earlier date. The main document contains the technical modeling details. 3. Section 1.6. Note the LiDAR accuracy. Because of this, is it possible to quote flood levels to any higher degree of accuracy? Is the LiDAR accuracy plus or minus, or does this represent the range of values?
4. Section 1.6. I note the initial assumption of a dry floodplain. Some comment on how realistic/likely this is would aid understanding. Accepting that a design storm philosophy is applied this appears realistic, but there is no comment on this in the report. The dry floodplain assumption is suitable for the design storm. We start getting into combined probabilities when considering wetting before the main storm. If using a historical event it would be suitable and possible to set appropriate initial water level conditions in the flood plain. 5. Section 1.6. Note the comment “... it is not possible to quantify the accuracy of the model.”. This needs to be borne in mind when referencing results. Agreed
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
23
6. Section 2.6. I agree with the omission of smaller stormwater pipes for significant events, but am not convinced on the results of 5-year ARI rainfall events being necessarily valid in this case, as I would anticipate that much of the local reticulation has been sized for an event of 5-year ARI. In this case I can agree that the model results will be conservative in that higher peak flood levels will be the result, although the degree of conservatism has not been quantified. Yes although much of the catchment is still rural and thus the pipe network extent is reasonably small. We are also planning on recalibrating the model with a smaller rainfall event to ensure the model is appropriate to use in smaller rainfall events. 7. Section 3.1. A minor point, but roughness appears to have been omitted as a hydrological parameter that is used with the kinematic wave (Model B). Roughness was not used as a calibration parameter in the hydrological model. General values were assigned based on the roughness characteristics of the landuse. The infiltration rates were found to be the most sensitive values in the calibration, which is typical for Model B. 8. Section 3.3.3. The way I understand it, the 1D reach can fill to bank level before the 2D model takes water. The 2D grid covers the 1D portion of the model (“... Once water enters the 2D model it is free to flow over the 1D model”). In the 1D model, will water level not rise as if constrained within “glass walls”, thereby double-counting the volume that exists over the 1D model between the banks? If so, I presume this effect is small, but I’d appreciate your comment on this. To me this would result in unconservative results, as volume is essentially spread between two different models where in reality it can only exist once. The bank markers in the 1-D channel are assigned so that only the main channel is modeled in 1-D. In addition, where the 1-D and 2D models are linked, the 2D model ground level is raised to the bank spill level (i.e. the river is filled in). So only the volume between the bank markers and the above the bank spill level is “double counted”, which turns out to be a very small amount. In addition the mass balance for the model has been checked and the error is less than 5%. 9. Section 4.1.2. This appears anomalous to me. The way I understand it, the hydraulic model was adjusted until the
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
24
model results fitted with measured data. This presumes that the hydrological model was accurate, and also that channel dimensions must have been as modelled (without knowledge of this). Surely the dimensions from stream improvement works should be modelled and not allowed to change, with other variables being tweaked to effect calibration? Interested in your opinion here. We had no data at the time on the 1988 stream dimensions. We agree that it would have been suitable to use the real data. 10. Section 4.1.3. Model results relative to measured data are compared. In many cases the differences are small, yet there is still significant uncertainty in the modelled results – the terrain used is only accurate to 0.5m in rural areas and 0.25m in urban. I suggest that “modelled results” should be quoted as a range rather than as an absolute, and ideally the range should envelope the measured data. This was discussed at the meeting on April 28th. Although a range of results may be more correct it is simpler to present a single value with the understanding that it includes uncertainty, by using a freeboard. 11. Section 4.2.1. I struggle to understand why the volumes generated using the TP108 rainfall profile should compare with those for the 1988 event. There is a disconnect between “design” and “actual” performance. Agreed but we wanted to check to see if volumes were of a realistic magnitude by comparing to a real event. 12. Table 4-5. The results from the December 1988 event are compared to those for a 100-year design rainfall event (TP108). This comparison appears meaningless as the 1988 event is quoted (Section 4.1) as being of 50-year ARI. Furthermore, comparisons are only made for Catchment 51, which by eye seems to be either largest or second largest and not necessarily representative of all Hingaia Stream catchments. Perhaps it would have been more suitable to use a 50 year event for this comparison however the table 4-5 is illustrating the differences in sensitivity to the infiltration rates for the different storm shapes, real versus design. Because of size, catchment 51 gave the biggest differences and thus made it easy to see a trend appear.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
25
13. Section 4.2.1. Below Table 4-5 it is stated that to calibrate infiltration rates it would be necessary to find an historic rainfall pattern with similar shape to the TP108 (Chicago) profile. Why could the actual rainfall data (hyetograph) from the 1988 event not be applied to the model? Am I missing something? The calibration was done using the 1988 rainfall. However we are just stating here that the design and real storms are not alike at all and that the response to a design storm would be different to that of a real storm. 14. There are two tables numbered “Table B-3”, one on page B-1 and the other on page B-3. I presume the table on page B-1 is Table B-1. Yes 15. Section 4.2.1. The model is said to have been run for 3 different sets of start/end infiltration rates, however none of those quoted match those presented in Table B-1 (labelled as B-3, see note above). Were these applied to all catchments, regardless of soil type? If so, could you justify this, please? For the higher infiltration soils we used start/end infiltration of 12.5/1.5 mm/hr and for the lower infiltration soils we used start/end infiltration of 7.5/1.2 mm/hr. Perhaps it is a bit unclear that the rates stated in the table are the range of rates used over the catchment. 16. Section 4.2.2. The 1988 flood event is quoted as having ARI of between 50 and 100 years, yet Section 2.8.1 states the ARI as being 40-50 years. Which is it? 40-50 as stated in the Snelder report. 17. I think that a differentiation is required between rainfall ARI and flood ARI. If the frequency analysis is performed on rainfall data then the floods being simulated are those that occur in response to the n -year rainfall ARI. If the discharge records (of which there are none) are analysed then the flood may be assigned this probability. I know this is a fine point. Yes, however as you say we have no flow record. Perhaps it just needs to be made clearer that we are modeling the 100 year Rainfall ARI and not necessarily the 100yr Flood ARI. 18. Section 4.3 also mentioned that the 1988 event was simulated to be of ARI between 50 and 100 years, while previous references indicated it to be of 40-50 year ARI.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
26
Refer to answer on comment 16 19. Section 4.3. The “mid-range” of infiltration rates has been quoted as being used, but the range itself was purely arbitrary, making “mid-range” equally arbitrary. Is it not possible to apply calibrated parameters from similar soils in a nearby catchment to the subject area? Given that modelled and recorded 1988 events appear to be of different ARI, this possibly has relevance. We did compare infiltration rates to a flood study we did on the nearby Papakura Stream catchment and with literature. We plan to use the Ngakaroa Stream flow site to help give a better idea of the infiltration for some of the soil (mainly the hilly areas of the catchment). 20. Section 4.3. I note that modelled peak water levels are quoted as being within 300mm of observed. If model uncertainty is taken into account (eg 0.5m accuracy in terrain), this represents a good fit. I suggest that model results be quoted as plus or minus, and ideally these extremes would envelope the observed values. True. 21. Section 4.4.1. I note that sensitivity to infiltration rate was tested by application of an increase to the values used. Could/should a decrease also have been applied? I would expect that the response to changes in infiltration rate would not be linear, and that doing both increase and decrease would provide greater clarity. I am interested in your comments. We agree in hindsight that it would be useful to also test the sensitivity to the decrease in infiltration rates. I believe the omission was probably due to a limit on the number of sensitivity tests we were to run. 22. Section 5.3.1. All floodmap figures appear to have a blue colouring (in the legend) for flood depths “below 0.0m”. To me this means that the area is dry, yet is plotted with a blue colour. Could you explain this, please? I also note that these maps appear to have been produced at a discrete time step, rather than having been plotted at maximum water levels across the entire simulation. I accept that the flooded areas being shown in each are relatively small, but can you confirm that the maximum occurs at the time step selected in each case?
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
27
The legend is just being misleading here, it should say greater than 0. The timestep closest to the peak was chosen but obviously the peak will occur at different times in different areas. These plots are here to show the flow patterns by using the flow vectors. The maps in the floodmaps section show the full maximum flood depths. 23. Figure 6-1. The legend indicates that flood depth is plotted but I think flood level is intended. While plotting flood level is useful to see hydraulic grade, plotting flood depth in addition may aid with understanding. Yes the legend is incorrect. Also see response to 22. 24. Section 6.2.5. The results of this simulation are surprising. I presume that significant hydraulic resistance exists in the unmodified case downstream of the proposed diversion, and that this diversion relieves this. I presume that the same tidal boundary condition was used for both cases. Long sections of each would aid with understanding of this. Yes there is significant hydraulic resistance in this area. We can provide additional long sections if necessary. 25. Section 7.2. The report states that “... considerable uncertainty exists in the model predictions”. I suggest that this point be highlighted, perhaps by quoting model predictions as falling within a range, particularly when absolute values are being derived. Noted [Mark Pennington] I note that the status of the March 2011 report is “draft” on the front cover and “final draft” in the quality assurance statement on the first page. We can release this as final now. [Mark Pennington] Section 2 in the March 2011 report states that no buildings were blocked out in the DSBP area. I presume this is because flood levels are all below proposed ground level. Is this true? Yes flood levels are below the proposed ground level, thus it makes no difference if the buildings are in the model or not. [Mark Pennington] In Figure 3-1 of the March 2011 report a larger head difference (for DSBP up- and downstream) across the structure appears to correspond to a smaller peak discharge. Could the legend be confirmed here, please? Will need to look into this further
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
28
[Mark Pennington] In the conclusions of the March 2011 report it is stated that flood risk is reduced downstream by “various aspects” of the DSBP. It should be noted that this is in response to design event rainfall and may not necessarily apply to actual rainfall. OK
Section 2: Questions 1. Section 2.2.4. Please supply a longitudinal section to show the downstream extension to the model with tide levels. This will allow the reviewer to assess effects of tidal boundary assumption. Provided 2. Section 2.3.1. Does raw LiDAR data set support 1m x 1m grid interpolation? What is the density of raw data points? Was the 5m x 5m grid interpolated off the 1m x 1m grid, or was it reinterpolated from the raw data? What interpolation method was used, and were breaklines applied? We will need to dig out the LiDAR again to confirm the point spacing. LiDAR points were built into an Arc GIS terrain which uses the entire dataset. The 1m and 5m grids were interpolated using the Natural Neighbour interpolation transferring from the terrain to raster. Significant highpoints i.e. banks and roads were traced from the 1m grid and copied into the 5m grid so that spill levels had higher accuracy. Breaklines were not used, however the resolution of the point data and the additional road/bank tracing was believed to generate a suitably accurate representation of the surface. 3. Section 2.5.1. How was chainage measured for 1D model? Is this centreline distances? Distance along thalweg? Are all cross sections normal to this line? Thalweg, and the cross sections were measured normal to this line. 4. Validation event – how was rainfall applied to the catchment? Off one gauge, fitted Chicago hyetograph? Two rainfall gauges (Ngakaroa and the Filter Station) were used and weighting based on the location of the catchments relative to the gauges.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
29
5. Section 2.8.1. Were the peak flow estimates by Snelder obtained using a nested Chicago rainfall profile? Snelder mentions that he used a design storm profile based on a pattern developed by the ACC around 1988. He does not mention whether a Chicago profile has been used. We can supply you with a copy of the Snelder report if you would like more information (page 46). 6. Section 2.9. For the Slippery Creek model (HEC-RAS), were the hydrological parameters and assumptions applied compatible to those of the current study? Same rainfall depths/profile? This is unknown, Larry Shui supplied the hydrographs to us directly. 7. Section 3.2.2. How was Horton decay applied? Was the decay exponent one derived from calibration of another catchment, or was the DHI default value applied? Where did the 5.5 hours come from? The decay exponent was based on that used in other studies, including the Papakura Stream study referenced before. 8. Table 3-1. Start and end infiltration rates for 3 different soil classes are the same. Can you confirm these, and provide justification for why they do not differ? Yes these are the values used. The values were grouped to simplify the model, so that there are two soil types, those with high infiltration and those with low infiltration. 9. Section 3.3.1. Report states that “The majority of the model runs have been updated to include the Ngakoroa Stream”. Could you elaborate on which runs do and do not include this, please? All models have now been rerun to include the Ngakoroa Stream, at the time of writing the report the DSBP options did not include the Ngakoroa Stream but have since been rerun to include this. 10. I may have missed it, but in Section 3.3.1 energy losses are quoted as having been applied at entrance and exit to closed sections. What coefficients were used? I assume these are applied to velocity head – could you confirm please?
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
30
Yes the loss is applied to the velocity head, the coefficients are 0.5 at the entrance and 1 at the exit. 11. Section 3.3.2. Second bullet point states that “... or LiDAR data has been processed incorrectly”. Could you explain how such areas may be identified, and how the effects of this might impact on results? Figure 3-2 gives a good example of this. The grid can be checked by eye to find very obvious errors or large areas where data points are missing. If areas like this are not identified they could have an impact on results especially if they are in the flooding region. 12. Section 3.3.2. Second bullet point. Where did the “correct” road spill levels come from? Have these levels been taken from the raw LiDAR, from survey or from interpolated LiDAR? What accuracy can be assigned to these levels? The road spill levels were used from the bridge survey or where this was not available from the nearest road level LiDAR point. 13. Section 4.1. Was land use in the 1988 event the same as it is currently? If not, was the hydrological model used for the 1988 event adjusted to reflect land use at that time? This point emerges again in Section 4.1.1 – ie was same percentage impervious applied for 1988 event as for the design scenarios? The land use and impervious areas for the 1988 event are different to the existing event, to reflect the land use scenario at this time. 14. Section 4.1.3. Could you please explain how the “calibration” (ie channel dimension modification) that was undertaken to get measured and modelled results to match for a small portion of the 1D model was “... transferred to the full MIKE FLOOD model...”? Were all channels widened by the same amount? Were all roughnesses decreased by the same amount? The changes made to the downstream end of the 1D model (around the motorway bridge) were copied into the full 1D model used in the MIKE FLOOD simulation. 15. Section 4 in general. What rainfall profile was applied to the 1988 event? Was this measured at a raingauge, and if so, which one? Was there any spatial variation of rainfall in this event? I note that no hydrological model calibration was undertaken (admittedly due to a lack of suitable data), but the model validation appears to have been done on the assumption that the hydrological model needed no adjustment,
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
31
with all changes being made to the hydraulic model. Could you comment on antecedent conditions for the 1988 event? Two rain gauges were used for the 1988 event, the Ngakaroa gauge (at the south end of the catchment) and the Filter Station record (north end of the catchment). The rainfall at the two sites had a different pattern. The Snelder report states that the antecedent wetness in the catchment was high due to continuous heavy rainfall before the storm. 16. Section 4.3. There is no mention of hydrodynamic parameters applied in the models. Could you please provide information on these, including eddy viscosity (flux/velocity based?), and whether or not “default” values for parameters (such as delta, theta, etc) were used? Was super-critical flow allowed in the model and were there any locations where this occurred? Eddy viscosity: velocity based, value of 1m2/s. 1D model parameters were kept at default values except delta which was increased to 0.8 for stability. Super critical flow is allowed in the model. We have not looked into super critical flow in depth since the model build stage however as I recall the Froude numbers were below 1, but they were reaching a reasonably high value at the downstream end of the Drury Town upstream of the motorway bridge. 17. Section 5.1.1. Last two paragraphs. Initially I was surprised that the model was adjusted to not reflect the case being modelled, but am I correct in assuming that these changes to model dimensions are recommended changes to be made to the final design concept to achieve an acceptable result? The DSBP design concept has been updated since this report was written. The raising of the road was made so that we could determine the highest level the flood waters would reach without overtopping the road, in order to aid design. 18. Could the actual floodmaps for Scenario 4 be plotted, please, rather than just the differences? Where differences are plotted it seems to highlight areas where notable terrain changes have taken place, making it difficult to really assess the effects of the changes. Refer to the floodmaps in the March 2011 report as this reflects the latest DSBP design model. Maximum depth is plotted as well as differences in flood level.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
32
19. Section 5.3.4. Ballard’s Bride is stated as acting as a control point. Could you supply Froude number over the simulation at this point, please? The latest simulation was not saved with Froude numbers, however a hand calculation gives a number of around 0.45 at the time of peak, at the point directly upstream of the bridge. This is for the DSBP scenario. 20. Section 6.1.1. Was the hydrology for the Ngakoroa Stream done in the same way as that for the rest of the catchment? Was the same rainfall applied? Again, the flows for the Ngakoroa Stream were supplied by Larry Shui, I don’t have more information on the modeling. 21. Section 7. Could you please confirm that hydrodynamic parameters were kept constant across all models for all simulations? Yes, except roughness was changed for the DSBP scenario runs to reflect the changes to landuse. 22. Could you please supply a model plan plot showing branch names and chainages (even just the .nwk11 file) to aid with location of reaches in the long section(s)? supplied [Mark Pennington] Section 2 in the March 2011 report, Table 2-1, gives bridge bottom and top levels for Beca Bridge as the same. Does this mean that deck underside and roadway level are the same? This was done at a request from Beca, however the point is moot given the flood level does not reach the soffit level of the bridge. Mark Pennington] Section 3 in the March 2011 report. Could you indicate start water level in the wetland, and whether or not spring flows are present that influence baseflow? Dale may be able to give more insight here, the wetland is modeled as “dry” for the start of the simulation however the ground level of the wetland area is in fact the dead storage level.
Email outlining results of scaled 1988 flood event and subsequent discussions
A02426300R001_DSBP SW Peer Review final final 270611
From: Antoinette Taylor [mailto:[email protected]] Sent: Wednesday, 22 June 2011 11:38 a.m. To: Mark Pennington Subject: RE: Drury
Hi Mark,
Did you note the same effects relating to timing of peak flows from sub-catchments for the 1988 event as for the design (TP108) events? If not then could you comment?
The timing of the peaks in the 1988 event are different to those in the design event. The 1988 event rainfall is double peaked, with the second peak being higher, overall resulting in a flatter flow hydrograph in the model. The timing of the second peak in the tributaries lines up with the Hingaia Stream flow peak at the downstream end resulting in the higher water levels.
The scaled 1988 event simulation results appear to show a flatter hydraulic grade overall, with difference appearing to increase with distance upstream. Is this observation correct?
Yes. See below, total rainfall depth is less in the 1988, especially in the upper catchments, resulting in the lower water levels. The water levels at the downstream end are higher due to the differences in timing in the events, leading to a closer synchronisation of the flow peaks.
How do the rainfall totals compare between the scaled 1988 event and a design 100 year event (the way I understand it, rainfall was scaled until peak discharge matched)?
Yes we scaled the rainfall until the peak discharge at the downstream end of Hingaia Stream was matched. Both sets of rainfall are distributed, however the total rainfall depth of the design storm was greater than the scaled up 1988 rainfall. The TP108 rainfall ranges from depths of 180mm to 225mm. The scaled 1988 rainfall ranged from 130 to 175 mm. I am out on site for the rest of the day, so I hope these answers are satisfactory. Kind Regards
Antoinette Taylor
Project Engineer
From: Mark Pennington [mailto:[email protected]] Sent: Wednesday, 22 June 2011 10:48 To: Antoinette Taylor Subject: FW: Drury Hi Antoinette I have received the appended email via several people, and just have a few questions relating to this. Did you note the same effects relating to timing of peak flows from sub-catchments for the 1988 event as for the design (TP108) events? If not then could you comment? The scaled 1988 event simulation results appear to show a flatter hydraulic grade overall, with difference appearing to increase with distance upstream. Is this observation correct?
How do the rainfall totals compare between the scaled 1988 event and a design 100 year event (the way I understand it, rainfall was scaled until peak discharge matched)? Cheers for now Mark
Mark Pennington MIPENZ CPEng IntPE(NZ) | Engineer PATTLE DELAMORE PARTNERS LTD 1st Floor ITM Building Centre, 218 Beach Road, Kaikoura | http://www.pdp.co.nz +64 3 319 3114 | +64 21 063 2112 |
[email protected]
DISCLAIMER: Any views expressed in the email may be those of the individual sender and may not necessarily reflect the views of Pattle Delamore Partners Limited. This electronic mail message together with any attachments is confidential and legally privileged between Pattle Delamore Partners Limited and the intended recipient. If you have received this message in error, please e-mail us immediately and delete the message, any attachments and any copies of the message or attachments from your system. You may not copy, disclose or use the contents in any way. All outgoing messages are swept by an Anti Virus Scan software, however, Pattle Delamore Partners Limited does not guarantee the mail message or attachments free of virus or worms.
From: Roger Seyb Sent: Monday, 20 June 2011 11:00 a.m. To: Mark Pennington Subject: FW: Drury
Roger Seyb Surface Water Engineer PATTLE DELAMORE PARTNERS LTD DDI Mob +64 9 523 6928 +64 21 540 993
From: Stephen Priestley [mailto:[email protected]] Sent: Monday, 20 June 2011 10:58 a.m. To: Roger Seyb Cc: Andrew Nell Subject: FW: Drury Roger The following is the response from DHI. I think it is a considered response. I hope it helps you to finalise your review (by Wednesday). Cheers Stephen From: Antoinette Taylor [mailto:[email protected]] Sent: Monday, 20 June 2011 10:53 a.m. To: Stephen Priestley Subject: RE: Drury
Hi Stephen,
The following is our response to the calibration comments and the model results for the scaled 1988 simulation: In the context of a ‘classic’ model validation exercise one would have a number of observed events by which to adjust model parameters to achieve a fit against those observations (calibration) and a number of observed events by which to compare model predictions with those observations and comment on the validity of the calibration (verification). The combined process is validation of the model.. Often, storm water studies of require the development of a model for which there are less than ideal observations on the performance of the system. In these cases it is not possible to carry out a ‘classic’ calibration and verification processes and any efforts to prove the validity of the model is then simply referred to as validation – without explicitly undertaking calibration and verification. This was the case here and it is necessary to therefore make a number of changes to model parameters and prove those changes all in one simulation. Our strategy in the project was to test the model – as built using available data and information against measured water levels (peak levels likely to be from debris marks with an uncertainty estimated to be of the order of +/- 100mm at best. Our testing found that the model predicted water levels in the up-stream reach of the river to be acceptable whereas the model was underpredicting water levels in the vicinity of up-stream of the bridge. From this we concluded that the model set-up and parameter choice up-stream of the Drury town were acceptable and that there was some miss-representation of the channel geometry and hydraulics in the model in the vicinity of the bridge. Following investigation we discovered that the channel in the vicinity of the up-stream of the bridge had been modified following the 1988 event. There were no actual pre-modification cross sections available but we made the assumption that the cross sections could be approximated to be of similar cross section shape to those upstream of the modified reach. We also represented the hydraulics of the confluence by using an energy loss description in the model Together, these changes to the model resulted in an agreeable match in water levels for the 1988 event. To this end the model was considered as validated against what (limited) data was available. The 1988 storm event has been run through the DSBP design model, where the 1988 rainfall was scaled up to match the 100 year flow at the downstream end. The table below shows a comparison of water levels at the DSBP site. Overall there is a decrease in levels upstream of the project area, as you move downstream the water level difference decreases and at the downstream end of the project the 1988 water levels are higher than the 100 year. Location 100yr DSBP Scaled 1988 Difference scenario Water Water Level Level 23.83 23.49 -0.34 20.5 19.08 17.34 20.19 18.83 17.16 -0.31 -0.25 -0.18
Upstream of Ararimu DS Ararimu Bridge, north end of project Upstream Beca Bridge Ponding area Upstream of Ballard’s Bridge
Confulence of quarry diversion stream and Hingaia Downstream end of Project Drury Town
15.46 10.58 6.91
15.4 10.61 6.98
-0.06 0.03 0.07
Kind Regards
Antoinette Taylor
Project Engineer
From: Stephen Priestley [mailto:[email protected]] Sent: Monday, 20 June 2011 10:46 To: Antoinette Taylor Cc: Andrew Nell Subject: Drury Antoinette As discussed on Friday, I requested the following information by 10am today. Could you please respond as this matter is urgent for our client. • • A graph showing the sensitivity of the assumed channel dimensions of the upgraded works (as per Papakura DC suggestion) to the flood levels. To model the 100 year event based on scaling up the 1988 rainfall hyetographs so as to match the Q100 flows at the downstream bridge. Then assess the difference in the levels (between the proposed Q100 levels and the levels derived from the up?scaled 1988 event) within the DSBP area.
Cheers Stephen Priestley Technical Director Beca Phone +64-9 300 9000 Fax +64-9 300 9300 DDI +64-9 300 9282 [email protected] www.beca.com Company of the Year Award 2010 // Deloitte/NZ Management Magazine Top 200 Awards
- -----------------------------------------------------------------------------------------------------------------------------------------------------
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
34
Appendix C: Record of Teleconference held on 28 April 2011
A02426300R001_DSBP SW Peer Review final final 270611
doc_884748460.pdf
A stormwater management pond is an artificial pond that is designed to collect and retain urban stormwater. They are frequently built into urban areas in North America to also retain sediments and other materials.
Auckland Wellington Christchurch
PATTLE DELAMORE PARTNERS LTD
Drury South Business Project Stormwater Management Report Peer Review
Stevenson Group Limited
solutions for your environment
Drury South Business Project Stormwater Management Report Peer Review
• Prepared for
Stevenson Group Limited
• June 2011
PATTLE DELAMORE PARTNERS LTD Level 4, PDP House 235 Broadway, Newmarket, Auckland PO Box 9528, Auckland, New Zealand Tel +9 523 6900 Fax +9 523 6901 Web Site http: / / www.pdp.co.nz Auckland Wellington Christchurch
solutions for your environment
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
i
Quality Control Sheet
TITLE
Drury South Business Park – Stormwater Management Report Peer Review
CLIENT VERSION DATE JOB REFERENCE SOURCE FILE(S)
Stevenson Group Ltd Final 27 June 2011 A02426300
A02426300R001_DSPB SW Peer Review final 270611.doc
Prepared by
SIGNATURE
Mark Pennington
Andrew Kolper
Roger Seyb
Directed, reviewed and approved by
SIGNATURE
Peter Callander Limitations: The report has been prepared for Stevenson Group Ltd, according to their instructions, for the particular objectives described in the report. The information contained in the report should not be used by anyone else or for any other purposes.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
ii
Executive Summary
The Drury South Business Park (DSBP) is a proposed 360ha industrial and commercial development approximately 3km south-southeast of the Drury Township and extends from State Highway 1 on the western boundary to the Hunua foothills in the east. The development is expected to provide employment for approximately 7000 people when fully developed. Pattle Delamore Partners have been engaged to review the DSBP Stormwater Management Report and in particular comment on the Hydrological and Hydraulic Modelling and issues relating to stormwater quality and in stream erosion. In general the hydrological and hydraulic models have been constructed to an acceptable standard and would be expected to yield acceptable results. There is some uncertainty remaining on model calibration and validation, which is largely due to the lack of calibration data available for use. Model scenarios have been simulated for various design rainfall events for nominated Average Recurrence Intervals (ARI) events. The results from these simulations will probably differ from actual rainfall events of similar ARI. This is a common occurrence with the use of a design event methodology, which is an industry standard approach. The model results have shown some sensitivity to timing of peak flows from various contributing tributaries. Peak flows can be offset from one another in a “design” event, but under actual rainfall conditions or flow event of the same frequency this may not be the case. This can be due to variations in hydrological and hydraulic parameters across the catchment (rather than them being assumed constant in the design event approach). In response to questions, a model run using the 1988 real event scaled up to a 100 year ARI event has been carried out to check differences between the results of the TP108 100 year ARI design event and a real storm event. This showed a slight decrease in flood levels within the site area at the confluence of the Hingaia and Maketu streams indicating that the superposition of peaks does not appear to be an issue using either the design event or a real event within the Hingaia Stream. There was an increase in flood levels in the Drury township area using the real event - the reason for this is unknown and should be checked further. The LiDAR terrain data used in the modelling was quoted as being of accuracy to 0.25m in urban areas and 0.5m in rural areas. The use of LiDAR data is appropriate. However, where significant out-of-bank flooding has occurred it is likely that model accuracy could be no better than the accuracy of the terrain data. All model results should, therefore, be treated as having a range of accuracy, and an appropriate freeboard allowance applied. A check of the LiDAR data by Beca against a previous survey in the central part of the site indicates negligible difference between the LiDAR and surveyed data. There is existing flooding in the Drury township for rainfall events less than 100 year ARI. It is unknown whether the proposed development would result in changes to the
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
iii
frequency or extent of these flooding events as they have not been assessed. It is expected that provided flood storage is available to mitigate the effects of land-use changes for the 100 year ARI rainfall event, it should be possible to ensure that the flood level for a smaller (more frequent) rainfall event should not increase. The flood level for the pre-development land-use patterns may however be reached more frequently. Assessment of flooding effects in rainfall events less than the 100 year ARI is recommended and a design philosophy proposed not to increase the level or frequency of downstream flooding. Stormwater quality management in Auckland follows a Best Practicable Option, rather than an effects based, approach. The approach proposed includes source control, industrial contaminant management and stormwater treatment. This is appropriate. The appropriate design guideline for stormwater quality treatment is the former Auckland Regional Council’s Stormwater Management Design Guideline Manual, (TP10, 2003). Stormwater treatment will be in the form of six wetlands which will cater for the vast majority of the site and a bio-filtration swale for those areas not able to be treated through the wetland process. In terms of treatment, the wetlands are appropriately sized in accordance with TP10 and generally also provide a surface area greater than 2% of the catchment area (which will provide a better performance than a deeper wetland of equivalent volume). It is recommended that the project design criteria philosophy include that the wetland surface area be at least 2% of the catchment area. In addition to the proposed treatment methods, a District Plan Change has been proposed to restrict the types of roofing materials within the development. The Hingaia ICMP contaminant load modelling has assumed this plan change to be operational. This plan change is therefore an integral part of the development in terms of stormwater quality management and needs to be maintained as part of the project design philosophy. TP10 also provides an approach to manage instream channel erosion downstream of a discharge point by providing extended detention of runoff from the 34.5mm storm. The proposed stormwater management approach includes providing the required extended detention volume in accordance with TP10 for four out of the six sub-catchments. It is recommended that the design philosophy include managing the potential for downstream channel erosion. This could include varying the amount of extended detention provided in individual sub-catchments provided it is confirmed that the erosive effect on the downstream channel of providing extended detention in a sub-set of catchments rather than all sub-catchments is at least as effective.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
iv
Table of Contents
SECTION
PAGE
Executive Summary 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 2.0 2.1 2.2 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 4.0 4.1 4.2 4.3 4.4 5.0 Introduction Background Scope Terms of Reference Site Visit Relevant Reports Referenced Initial Queries and Comments Review Structure Stormwater Management Objectives Overall Stormwater Objectives Design Philosophy Hydrological Model Review Catchments and Sub-Catchments Rainfall Soils Loss Model Routing Model Third Party Models Calibration and Validation Hydraulic Model Review Boundary Conditions 1-D Component 2-D Component Calibration and Validation Flooding in response to rainfall events of ARI less than 100 years 6.0 6.1 6.2 7.0 7.1 7.2 7.3 7.4 Stormwater Quality Review On Site Stormwater Quality Extended Detention Volumes Conclusions Flood Assessment Flooding during events of ARI less than 100 years Stormwater Quality Extended Detention
ii 1 1 1 1 2 2 2 3 3 3 3 4 4 4 5 5 6 6 7 8 8 8 10 11
11 12 12 14 15 15 16 16 16
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
v
8.0 8.1 8.2 8.3 8.4
Recommendations Hydrologic and Hydraulic Model Flooding during events less than 100 year ARI Stormwater Quality Extended Detention
16 16 17 17 17
Appendices
Appendix A: Site Visit Record Appendix B: Questions to, and responses from, DHI Appendix C: Record of Teleconference held on 28 April 2011
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
1
1.0
1.1
Introduction
Background
Pattle Delamore Partners Ltd. (PDP) has been engaged by Stevenson Group Limited to undertake a desktop Peer Review of the Drury South Business Park (DSBP) Stormwater Management Report which was prepared by Beca Infrastructure Ltd (Beca) in support of a Plan Change Application, Metropolitan Urban Limit Change Application and a Network Discharge Consent Application for the proposed DSBP. The Beca Stormwater Management Report incorporates Mike Flood modelling carried out by DHI for Stevenson Group Ltd. The objective of the review was to comment on the appropriateness of the report in terms of flood plain management, hydrological changes, stormwater quality and stream erosion management, within the framework that is described below.
1.2
Scope
The scope of the review was to: ? Assess the hydrological and hydraulic modelling assumptions, parameters and results; ? Assess the approach used in identifying and assessing the stormwater quantity and quality issues and the appropriateness of key assumptions; ? Assess the assumptions and conclusions drawn as well as the proposed avoidance / mitigation measures proposed; and, ? Assess the coverage of relevant council regulations and issues against the purpose of the report. This review does not specifically addressed stormwater network, overland flow or stream re-alignment issues. The scope is outlined in greater detail in the brief forwarded to Beca on 1 April 2011.
1.3
Terms of Reference
The Peer Review which was undertaken is a simplified version of the Design Review as detailed in the IPENZ PN02 Practice Notes. The simplified design review entailed reviewing the Beca DSBP Stormwater Management Report as well as the DHI Modelling reports (Modelling for Hingaia Stream ICMP and Drury South Business Park, Hingaia Stream Flood Hazard Assessment) and commenting on their content. The review has been based on a review of reports and supplied responses to questions only. No model files have been supplied nor reviewed. This means that no checking has taken place to ensure that what has been stated in model reports has been accurately applied in the model files used. PDP have also not viewed any model result files, so checks for issues associated with model stability have not been undertaken. Rather, outputs taken from model result files,
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
2
such as peak water levels, longitudinal sections and flow rates as quoted in the model reports (rather than as taken from result files) are the subject of this review. This peer review does not cover any subsequent iterations of the review process (i.e. work carried out in response to recommendations after 24 th June 2011).
1.4
Site Visit
On 5 April 2011 PDP staff undertook a site visit of the area for the proposed Drury South Business Park and its surrounds. During the site visit a number of key features in terms of hydraulic performance were located and viewed, and an appreciation of the scale of the site was attained. A record of this site visit was kept, and this is provided in Appendix A.
1.5
Relevant Reports Referenced
Beca provided PDP with two separate modelling reports that were required to be reviewed under the scope of the work reported on in this document. These two modelling reports are:
?
Modelling for Hingaia Stream Integrated Catchment Management Plan , prepared for Papakura District Council / Franklin District Council by DHI Water and Environment Limited, 8 July 2010
?
Drury South Business Project Hingaia Stream Flood Hazard Assessment , prepared for Stevenson Group by DHI Water and Environment Limited, March 2011. The version supplied for review was a “final draft”, but subsequent correspondence (see Appendix B) indicated that this could be taken as a “final” report
In addition to the modelling reports, further reference material was used in the peer review process and included:
?
DSBP Stormwater Management Report , Prepared for Stevenson Group by Beca, March 2011
?
Hingaia Stream Integrated Catchment Management Plan, Prepared for Papakura District Council / Franklin District Council by Golder Associates, August 2010
?
Technical Publication 10 (TP10), Auckland Regional Council, 2003
1.6
Initial Queries and Comments
The initial phase of the review resulted in a series of questions and comments being submitted to DHI, to aid the reviewer’s understanding of the hydrological and hydraulic model build, calibration and scenarios investigation. The complete list of comments and questions, with answers as supplied by DHI, is given in Appendix B. Following distribution of this initial list of comments and questions, a teleconference between relevant parties was held. The purpose of this was to discuss the review and to establish the way forward, particularly with respect to any additional model runs that
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
3
needed to be undertaken. Following the teleconference PDP prepared a summary of this that was distributed by email. This email message is appended in Appendix C, and no comment on this summary was received from any of those present at the teleconference. As a result PDP assumed this to be a fair and accurate record. A further meeting was held on 16 th June 2011, at which review comments were further discussed and as a result of which DHI were instructed to run a scaled up version of a 1988 event through the model.
1.7
Review Structure
The review was broken up into review of each of the separate elements relating to: ? ? ? ? ? Stormwater Management Objectives. Hydrological Model Review. Hydraulic Model Review. Flooding in response to rainfall events of ARI less than 100 years. Stormwater quality review.
A convenient split between hydrological model (that used to simulate the rainfall-runoff process) and hydraulic model (that used for hydraulic computation) has been made. Recommendations in relation to the stormwater management report are made in association with each of the above topics.
2.0 Stormwater Management Objectives
2.1 Overall Stormwater Objectives
The Hingaia Stream ICMP Objectives are listed in Sections 11.2-11.6 of the ICMP and replicated in section 5.5 of the DSBP Stormwater Management Report. The DSBP has its own set of site stormwater objectives as set out in Sections 5.1-5.3 and design criteria in Section 1.4 of the DSBP Stormwater Management Report. Review Comment The overall stormwater management objectives adopted by DSBP are in general agreement with the Hingaia ICMP stormwater objectives. The design criteria adopted in Section 1.4 generally address normal stormwater management issues.
2.2
Design Philosophy
Beca has further developed the stormwater objectives into a design philosophy found in Section7.1 of the DSBP Stormwater Management Report.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
4
Review Comment The design philosophy as adopted for the DSBP development is generally based on industry best practice. Some additional design philosophy items are recommended and these are summarised at the end of this report.
3.0 Hydrological Model Review
3.1 Catchments and Sub-Catchments
The review of model catchments and sub-catchments and their respective connections to the relevant model was conducted based on the supplied drawings. Catchment delineation is stated to have been indicated erroneously in the report (see Comment 1 in Appendix B), but assurance has been given that this was corrected in the model. As mentioned in Section 1.1 PDP have not viewed nor reviewed any model files, so cannot confirm nor refute this. Sub-catchment delineation has been performed at a resolution that would fit with standard practice for the level of accuracy in results required to achieve the purpose of the modelling work. While the proposed DSBP development represents a significant change in landform, it was found that changes to the sub-catchment layout used for the ICMP Modelling (July 2010) could be applied to the “post-development” scenario as described in the subsequent report (March 2011). Review Comment There are no obvious errors, omissions or shortcomings in the catchment and subcatchment layout as reported on in the relevant documents. The catchment split has been made to a resolution that is suitable to meet the required level of detail.
3.2
Rainfall
Rainfall data from a range of recorders were used in the modelling. Most notably, recorded data from two gauges (Ngakora and Filter Station) were used in simulation of the validation (1988) event (see response to Question 4 in Appendix B). It is understood that an appropriate weighting for each of these gauges was used in simulation of this event, although this has not been independently verified. For design event simulation, rainfall data were taken from the guideline “Stormwater Runoff Modelling in the Auckland Region”, TP108 (ARC), and while this guideline is in the process of review, appropriate data were used given the time at which the modelling was undertaken. A nested “Chicago Storm” hyetograph was used for design event simulation, and this is the appropriate methodology in this case. As the selected hyetograph has 24 hour duration it is important that the catchment under analysis does not have response time or critical duration any longer than this, and this is definitely the case for the models that have been used.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
5
A response to Comment 13 in Appendix B highlights the likely difference in performance of the system (and model) in response to recorded as opposed to design rainfall. Review Comment The rainfall data used have been appropriately derived and applied.
3.3
Soils
Differences in surface soil types have been identified and taken into account in the modelling. Soil type is relevant in that it influences the rate at which rainfall is able to infiltrate, which has a large bearing on the rate of runoff generation in response to rainfall. While five different soil drainage classes were identified within the catchment, this was simplified with only two having been applied in the model. The total runoff volume for a recorded event (1988) of 40-50 year Average Recurrence Interval (ARI) was compared to that from a design event of 100-year ARI for a range of different infiltration rates (see Comments 11 and 12 in Appendix B). This was done to test the sensitivity of the model to the assumptions surrounding infiltration rates. In addition, increases to model infiltration rates were tested, although similar decreases were never applied (see Comment 21 in Appendix B). The response to Comment 19 in Appendix B indicates some comparison with applied infiltration rate from a nearby (calibrated) catchment had been made. Whilst the results of this comparison are not stated it is assumed that any wide discrepancy would have been highlighted and taken into account. This response also indicates that a further revision to the model may be made, using data from a flow monitoring site although this is not confirmed. The response to Question 7 in Appendix B also indicates that results from other hydrological studies were applied to the one in question, adding confidence to the apparent “assumptions” made. Review Comment The variation in soil type across the catchment has been taken into account, albeit in a simplified manner. Sensitivity testing has been undertaken on infiltration rate, and values have been compared to those derived from nearby catchments. The values applied appear to fall within what is regarded as a reasonable range. It appears that further work in establishment of design values for key parameters (such as infiltration rate) would aid with model confidence.
3.4
Loss Model
The hydrological model used is known as “Model B” in DHI software terms. The loss model component of this is based on an initial loss due to factors such as depression storage and initial wetting, and an Hortonian infiltration rate decay with time. This provides relatively few parameters to be used for model calibration, many of which can be physically measured.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
6
The key to understanding the applicability of the overall hydrological method applied lies in calibration and validation (refer Section 3.7). Variations for parameters within the loss model are not expected to be significant for infrequent (very large) rainfall events as the proportion of infiltration to total rainfall depth is relatively small. The effect on runoff is however expected to be more pronounced for more frequent rainfall events. Review Comment Whilst the loss model applied is a standard approach, this does not fit with that suggested in the guideline TP108 document that was used for rainfall (this uses the SCS Curve Number loss model). In this way two different hydrological methods have been integrated, which raises the question of applicability. It is recommended that the loss model be re-assessed and then calibrated and verified for more frequent (smaller) rainfall events.
3.5
Routing Model
The routing model applied is a part of the DHI “Model B” approach, which utilises the Kinematic Wave routing model. This is based on physically measurable data (slope, length, area) together with some derived parameters (e.g. Manning’s n ) that have been well-established in the literature. Review Comment Whilst the routing model applied is a standard approach, this does not fit with that suggested in the guideline TP108 document that was used for rainfall (this uses the SCS Unit Hydrograph routing model). In this way two different hydrological methods have been integrated, which raises the question of applicability. The values applied to the relevant parameters in the routing model seem to have been appropriately derived (see the response to Comment 7 in Appendix B). The key to understanding the applicability of the overall hydrological method applied lies in calibration and validation, and this point is discussed in Section 3.7.
3.6
Third Party Models
The Ngakora Stream and Slippery Creek are understood to have been modelled by a third party, with results from this having been used by DHI in the Hingaia Stream modelling. No independent checks of the methodology nor input data applied to these other models have been carried out, and DHI have indicated a lack of detail on this (see response to questions 6 and 20 in Appendix B). PDP have not reviewed any report pertaining specifically to this model. However, as these two waterways are located somewhat downstream of the specific area of relevance to the proposed Drury South Business Park development, there is likely to be relatively low sensitivity of proposed mitigation measures to accuracy of these models.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
7
Review Comment The effects of this part of the model potentially not having been built to the same set of assumptions and underlying methodology are likely to be small.
3.7
Calibration and Validation
Hydrological model validation appears not to have been undertaken in isolation from the hydraulic model, but rather both were essentially validated together. A lack of measured data has prevented full calibration from being able to be done. Validation was done by comparison of modelled against measured/estimated levels and flows at various points in the stream network. The 1988 event was selected for this. The hydrological model used was adjusted to reflect land use in the catchment at the time of the event (see response to Question 13 in Appendix B). The hydraulic model was adjusted (channel dimensions) until modelled and observed data exhibited a good match. Review Comment As stated in Comment 9 in Appendix B, the calibration approach seems to assume accuracy in the hydrological model, with channel dimensions being adjusted to achieve a good fit with observed data. This is because it is known that channel dimensions during the 1988 event were different from those used in the model, although it is not known how different these were. This approach is not standard (nor is it endorsed as part of this review) for model calibration, although the results were able to demonstrate a reasonable fit. In response to questions, a further model run was carried out by DHI using the 1988 rainfall event. The flow at the downstream end of Hingaia Stream from the 1988 event was scaled up to match that from the 100 year design rainfall event. The resulting flood levels were slightly lower in the upper catchment and the confluence of the Hingaia and Maketu Streams in the centre of the site and slightly higher in the lower catchment in Drury township. The results are also reported by DHI to show flatter hydrographs - which could increase the likelihood of superposition of peaks (or near-peaks) within the system, Within the Hingaia Stream catchment, the superposition issue is most likely to be an issue at the confluence of the Maketu and Hingaia Streams. The drop in flood levels at this confluence indicates that the design event is slightly conservative, at least with respect to the real event available. There is an increase in flood levels using the real event lower down in the catchment at Drury Town. The reason for this is unexplained and should be checked further. Review Comment Superposition of peaks in the scaled 1988 event does not appear to be an issue at the confluence of the Hingaia Stream and the Maketu Stream. There is an unexplained increase in the flood levels (of 70mm) in the Drury township area using the scaled 1988 event compared to the design event. The reason for this increase needs to be investigated further.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
8
4.0 Hydraulic Model Review
4.1 Boundary Conditions
The downstream tidal boundary was set at a steady Mean High Water Springs (MHWS) level for all simulations. This is a conservative approach because it is known that tidal fluctuations over the course of the 24-hour rainfall events used for design simulations will occur. Question 1 in Appendix B alludes to this, and in response DHI provided a longitudinal section of the lower reach of the model. From this it is clear that given the water levels involved, the bed slope of the channel and the distance from the downstream boundary to any areas of specific interest, there is unlikely to be significant model sensitivity to this downstream boundary water level. Inflow boundaries were derived from the hydrological modelling process. Some boundary inflows occur as point inflows and others are distributed, according to the physical system as it exists. In the July 2010 report it was not clear whether or not an inflow from the Ngakora Stream was used for all simulations, but the response to Question 9 in Appendix B indicates that this was the case. The actual boundary time series used for the Ngakora Stream inflow was derived by a third party and, as mentioned in Section 3.6, this has not been reviewed by PDP. The same applies to the Slippery Creek inflow and associated model. For the start condition of the model it was assumed that the wetland areas were initially dry. This fits with a classic “design storm” approach which is generally viewed independently of antecedent conditions. Review Comment The boundaries generated by hydrological modelling appear to have been applied appropriately to the hydraulic model. No independent checking of model files to ensure the time series are connected at the correct locations in the hydraulic model has been carried out, although it is assumed that this would have formed a distinct part of the internal review process by DHI. The downstream tidal boundary used is appropriate, although the source information of the actual level applied (stated as MHWS) has not been independently verified.
4.2
4.2.1
1-D Component
Discretisation
The one-dimensional model has been discretised into branches and reaches. The computational grid has been set within these to 10m in most cases (as reported by DHI). This provides a level of detail in both computation and in results interpretation that suits (if not exceeds) the minimum needs for the model results. Review Comment Model discretisation is appropriate for the investigation undertaken.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
9
4.2.2
Structures
Physical checks of structures to ensure their correct representation in the model have not been carried out as part of this review process. The bridge modelling approach used is a widely accepted approach and is suitable for the purpose to which it was applied. Culvert structures are understood to have been used for all open channel – closed conduit transitions, and this correctly ensures that energy losses occur at entrance and exit. The response to Question 10 in Appendix B reveals that appropriate values for entrance and exit loss coefficients have been used (while these do not, theoretically, remain constant over a range of flows and upstream and downstream depths, this is an acceptable approximation). Review Comment Representation of structures is suitable for the purpose of the model.
4.2.3
Channels and Pipes
Only the low flow channels have been represented in the 1D model, as all overland flows have been taken into account in the 2D component. This is considered appropriate. In the absence of detailed calibration data, selection of roughness to be applied to the model reaches is subject to judgement and by reference to standard texts. The values selected appear to fit within what would be considered “reasonable” ranges for the surface types encountered. Smaller pipes have been omitted from the model (see Comment 6 in Appendix B). This is a simplification that has little bearing on results of extreme flood events, but may have a small local influence in less severe events. Comment 8 in Appendix B describes some possible “double-counting” of volume above the bank level within the 1D portion of the model. The response from DHI indicates the effect of this to be negligible. Review Comment The response to Comment 6 in Appendix B indicates a willingness to recalibrate the model using a smaller event, and this forms one of the recommendations of this review report. It is agreed that any potential “double-counting” of flood volume above the 1D cross sections is negligible in terms of overall results.
4.2.4
Cross Sections
Channel cross sections have been derived from survey where possible and from (adjusted) LiDAR data elsewhere. This is an appropriate approach, giving greater weight to the surveyed cross sections than the LiDAR data wherever the two overlap. The report correctly states that LiDAR cannot be relied upon for incised channels, for heavily wooded areas and for areas under water at the time of the survey.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
10
Review Comment The way in which cross sections were derived is appropriate for the terrain.
4.2.5
Hydrodynamic Parameters
For the Mike11 (1D) component of the model the response to Question 16 in Appendix B indicates that “default” values for hydrodynamic parameters were used, except for the coefficient “Delta”. This was increased from a default value of 0.5 to a value of 0.8 to aid with model stability. It is noted that the Mike 11 Hydrodynamic Reference Manual produced by DHI states that “Changing the value of DELTA from its default value should only be carried out when “short term” fluctuations are unimportant, as in the case of a wave with period of several days or longer”. This appears to conflict with what was done in the model, however this is relatively common practice. Review Comment While this appears to conflict with the literature produced by DHI for this software the practicalities of attempting to run a model such as this with Delta set to 0.5 may be excessively onerous. It is accepted that some accuracy may be lost due to damping out potential oscillations (as changing the value of Delta can do), but in terms of overall results this is likely to be insignificant given other uncertainties that also exist. Account of these uncertainties is usually made by application of a freeboard allowance, as has been done in this case.
4.3
4.3.1
2-D Component
Discretisation
It is understood that the terrain used for the 2D component of the model was derived from a LiDAR survey of the subject area. A 5mx5m grid was used which is understood to have been supplied to DHI. The vertical accuracy of the LiDAR data is stated to be to 0.5m in rural areas and 0.25m in urban areas. The distinction between rural and urban in this context is not clear. The response to Question 2 in Appendix B indicates that raw LiDAR point spacing is not known, nor if this was suitably dense to support the chosen grid resolution. Where there is a lack of suitable survey calibration data, there may be a general translation of the terrain data above or below the local survey benchmarks. In response to a question on this, Beca checked the terrain data against a survey they have previously carried in the central area of the site and found a negligible difference between the two datasets. Review Comment Experience with other LiDAR data sets indicates that a 5mx5m grid should easily be able to be supported. However mention in the July 2010 report is made of a 1mx1m grid, and this would require significantly more detailed LiDAR survey in order to not provide artificial
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
11
accuracy. The process followed by DHI in setting up the 5mx5m grid appears to be robust. Further spot checks on terrain data versus local survey benchmarks should be carried out through the design process, or, checks carried out at construction stage to ensure that the required flood storage volume below existing ground level has been provided.
4.3.2
Links
The 1D and 2D components of the model are connected using lateral links. No detail of these was provided in the reports that were reviewed. The response to Questions 11 and 12 in Appendix B indicates that there was a methodical approach applied to setting spill levels and in ensuring that the data made sense. Review Comment As the model files were not reviewed the locations and levels associated with these links were not independently verified.
4.3.3
Parameters
Model parameters were not specified in the reports, but the response to Question 16 in Appendix B has indicated that appropriate values were used. Review Comment As the model files were not reviewed numerical values applied to various parameters were not checked.
4.4
Calibration and Validation
As mentioned in Section 3.7 no suitably detailed calibration data were available at the time of the model build. The 1988 event was used for validation, using relatively sparse observations of water levels. Review Comment As stated in Comment 9 in Appendix B, the calibration approach seems to assume accuracy in the hydrological model, with channel dimensions being adjusted to achieve a good fit with observed data. This is because it is known that channel dimensions during the 1988 event were different from those used in the model, although it is not known how different these were. This approach is not standard for model calibration, although the results were able to demonstrate a reasonable fit.
5.0 Flooding in response to rainfall events of ARI less than 100 years
The DSPB Stormwater Management Report presents results of the modelling to show that the post-development scenario flooding in response to a 100-year design rainfall event is no worse than for the current level of development in the subject area. No assessment has been undertaken to determine the extent of flooding during more frequent (i.e. lower
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
12
ARI) events. Possible effects in response to more frequent events could include changing flooding levels / extent and /or increasing the frequency of flooding at a given flood level. There is existing flooding above floor levels in the downstream Drury Township under current development intensity in response to rainfall events of ARI less than 100 years and some buildings may be affected in events less than 10 years. The effects of the proposed DSBP on existing flooding levels and frequency should be assessed. This would typically be carried out for the design event associated with sizing stormwater network infrastructure (10 years) and, given flooding may occur for smaller events, a smaller event such as the 2 year ARI. Review Comment The effect of the proposed mitigation measures on the downstream environment is unknown for flood events of ARI less than 100 years. It is recommended that changes resulting from the proposed development be assessed for the 10 year and 2 year ARI events (with and without climate change allowances) for the existing and Maximum Probable Development scenarios. As noted in Section 3.4, the loss model used may have a greater effect on more frequent (smaller) events and the overall model, therefore, needs calibration for these events. The design philosophy should include that there be no increase in the frequency or degree of flooding of properties in response to both 100 year and 10 year ARI rainfall events. If this is not possible off-site mitigation should be provided.
6.0 Stormwater Quality Review
6.1
6.1.1
On Site Stormwater Quality
General
The DSBP Stormwater report details options for stormwater point source treatment such as swales, rain gardens, sand filters, proprietary devices and others and gives an overall treatment train approach for stormwater quality management including: • • • Source control of metals from roofs; Industrial and trade process contaminant management; Stormwater quality treatment devices.
The contaminants of concern for the catchment are identified within the Hingaia ICMP are sediments and other contaminants. No site specific contaminants of concern are identified, but these are expected to be sediment, metals, hydrocarbons and industry specific contaminants. Review comment An overall treatment train approach for managing stormwater contaminants is good practice and the one proposed should address the expected contaminants of concern.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
13
6.1.2
Treatment of Metals
In addition to the point source treatment and wetlands proposed, a plan change has been proposed to prevent the use of bare metal and galvanised roofs. This is expected to significantly reduce the load of metals in stormwater from the site. We note that the contaminant load modelling carried out for the Hingaia ICMP (Stormwater Management Report Section 7.5.3) assumed that the industrial roofs will be those typically available today – which could therefore include unpainted zinc-alum roofs. If the roof areas within the DSBP were zinc-alum the expected contaminant yield would be higher than that for pre-painted steel. The plan change proposed to control roof material type to pre-painted steel roofing therefore means that the contaminant load from the DSBP roofs is less than that may have been calculated in the ICMP modelling. Review Comment The reduction of metal contaminant loads at source by avoiding exposed metal type roofing materials is considered an appropriate design philosophy. An alternative to this may be that additional on-site treatment is provided for roof run-off from any exposed metal roofing.
6.1.3
Industrial and Trade Processes
In addition to the wetland approach to stormwater treatment, medium and high risk sites as defined in the Auckland Regional Plan: Air, Land, Water (ALW Plan) will require specific (ITP) consents for the management, treatment and discharge of stormwater. These sites would utilise the specific point source treatment methods described within the DSBP Stormwater report and Environmental Management Plan requirements set out in the ALW Plan. Review Comment The design philosophy should include that specific on-site stormwater treatment measures will need to be designed to comply with the objectives of the Hingaia Stream ICMP, the DSBP Stormwater Management Report and the rules in the ALW Plan. Environmental Management Plans will need to meet the requirements of the ALW Plan.
6.1.4
Treatment Sizing
The on site stormwater treatment proposed for the business park was assessed on its adherence to the former Auckland Regional Council’s Stormwater Management Devices Design Guideline, Technical Publication No. 10, 2003 (TP10). The Best Practicable Option (BPO) approach has been used to select the treatment method. The BPO, based on an analysis of the treatment choices for the site, was the construction of a number of wetlands to treat the runoff from discrete catchments within the greater site areas.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
14
In addition, a small sub-catchment (Area 7) will utilise an engineered biofiltration swale to treat the stormwater runoff from the site. No allowance for extended detention or attenuation of larger storm events has been provided in the swale. The TP10 design methodology is based on the removal of total suspended solids and, by association, metals attached to those sediment particles. The removal of dissolved metals is promoted through the use of vegetative practices (such as wetlands). Wetlands are a preferred method within TP10 to target the removal of sediment and metals. Within wetlands, metal removal is promoted by longer vegetative contact and there is greater opportunity for this within systems with large shallow vegetated areas with low through flow velocity and minimal short-circuiting. Beca report that the surface area to catchment area ratio of the individual wetlands proposed ranges between 1.75% and 4.0%, with the average more than 2% of the catchment area to the wetlands. This is beneficial and will help to optimise the wetland performance for the required storage volume. Review Comment From the information provided within the DSBP Stormwater Management report, the water quality volumes (WQV) which will be provided for each of the six catchments (as set out in report) are in general accordance with the requirements of TP10. Wetlands typically perform better when there are more extensive shallow areas than deep areas – as is the case for the wetlands proposed. It is recommended that the design philosophy include that the surface area of the wetlands be at least 2% of the catchment area draining to each wetland.
6.2
Extended Detention Volumes
TP10 recommends the use of an extended detention volume (EDV), in addition to WQV requirements, to manage downstream channel erosion. The EDV requirement is to detain runoff from the 34.5mm rainfall event and release that volume over a 24 hour period. As noted in Table 7 of the DSBP Stormwater Management Report, if a wetland provides for EDV, the required WQV is reduced by one half. Two of the six wetlands (Labelled Nos. 3 & 4) do not provide sufficient volume to comply the EDV requirement as set out in TP10. As explained within Section 8.2 of the DSBP Stormwater Management Report, the reasoning behind the omission of EDV on the two wetlands was a lack of available volume and a reduction in modelled downstream velocities when runoff from the site was routed directly into the streams. Review Comment The design philosophy should include the management of downstream channel erosion. Where the Extended Detention Volume cannot be provided in accordance with TP10, checks should be carried out that downstream erosion will not be exacerbated by the omission of the EDV within some wetlands (for example the length of time that channels are exposed to erosive velocities before and after development).
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
15
7.0 Conclusions
7.1
7.1.1
Flood Assessment
Hydrological Model
In general the hydrological model has been built using generally accepted standards and principles. There has been some mixing of methods in that the Chicago Storm rainfall hyetograph as specified in TP108 was used together with DHI Model B for the loss and routing models (where TP108 suggests the SCS method). A lack of calibration data has meant that the hydrological model remains uncalibrated, although reference has been made to work done on neighbouring catchments that gives some confidence in the values for various parameters that have been applied.
7.1.2
Hydraulic Model
The hydraulic model has also been built to acceptable standards, using widely accepted data and methods of representation. While the model reports reviewed did not contain all of the information required to make this assessment, the answers to subsequent questions on this have augmented the information in the model reports to the extent that adequate review could be undertaken.
7.1.3
Results
Independent spot checks of modelled peak discharge have been carried out, confirming that reasonable values have been produced through the modelling process. Model results have indicated a reliance on offsetting of flood peaks from contributing catchments in order to keep peak flood levels below threshold levels. This off-setting effect has been shown to be present in model results for design rainfall events using the synthetic Chicago Storm hyetograph and at the confluence of the Maketu and Hingaia Streams using a scaled 1988 flood event (i.e super-position of flood peaks within the Hingaia Stream is not a problem in the scenarios assessed). Use of the scaled 1988 flood event showed a slight increase in downstream flood levels at Drury township and a decrease in flood levels upstream of the confluence of the Hingaia Stream and the quarry diversion. The increase in flood levels using the real event lower down in the catchment at Drury Town is unexplained and should be checked further. Flood levels have been quoted in both reports that were reviewed. Freeboard allowance is a key component of the approach to address potential variation in model inputs, in setting finished development levels above flood level estimates. A check on the difference between “average” LiDAR terrain levels and a survey within the central part of the site indicated negligible difference in levels. Further checks should be carried out during the design stage, or, checks should be carried out at construction
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
16
stage to confirm that the required flood storage is provided below existing ground level (i.e. cut below existing ground to achieve a volume not a given level).
7.2
Flooding during events of ARI less than 100 years
The effect of the proposed mitigation measures on the downstream environment is unknown for flood events of ARI less than 100 years. It is recommended that this be assessed for the 10 year and 2 year ARI events (with and without climate change allowances) for the existing and Maximum Probable Development scenarios. The design philosophy should include that there be no increase in the frequency or degree of flooding of properties in response to both 100 year and 10 year ARI rainfall events. If this is not possible off-site mitigation should be provided.
7.3
Stormwater Quality
The stormwater quality treatment devices have been sized in general accordance with TP10. The wetlands proposed are appropriately shallow and will therefore promote contaminant removal. The proposed plan change restricting roof materials within the DSBP is an integral part of the Hingaia ICMP and the overall approach to the management of stormwater quality from the site. The design philosophy should include that specific on-site stormwater treatment measures will need to be designed to comply with the objectives of the Hingaia Stream ICMP, the DSBP Stormwater Management Report and the rules in the ALW Plan. Environmental Management Plans will need to meet the requirements of the ALW Plan.
7.4
Extended Detention
The proposed EDV was not fully in accordance with the requirements set out in TP10. Specifically, additional extended detention in wetlands 3 and 4 would be required to comply with the TP design approach. In discussions with Beca, the need for EDV for each sub-catchment was questioned as the preliminary modelling showed a decrease in stream velocities. The design philosophy for the project should include that there is no increase in downstream channel erosion and the effect of not providing EDV in each wetland on the downstream channel should be checked (for example by checking the length of time that channels are exposed to erosive velocities before and after development).
8.0 Recommendations
8.1 Hydrologic and Hydraulic Model
There are two recommendations relating to models and their accuracy:
?
The increase in flood levels using the real event lower down in the catchment at Drury Town is unexplained and should be checked further.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
17
?
Re-assess the loss model and then calibrate and verify the model for more frequent (smaller) rainfall events
8.2
Flooding during events less than 100 year ARI
Investigation of the potential for flooding during events less than the 100 year ARI is recommended. The design philosophy should include that there be no increase in the frequency or degree of flooding of properties for a 100 year or 10 year ARI rainfall event. If this is not possible mitigation should be provided.
8.3
Stormwater Quality
The design philosophy should include that specific Industrial and Trade on-site stormwater treatment measures will need to be designed to comply with the objectives of the Hingaia Stream ICMP, the DSBP Stormwater Management Report and the rules in the ALW Plan. Environmental Management Plans will need to meet the requirements of the ALW Plan. It is recommended that the stormwater management design philosophy include that each wetland has a permanent water surface area of at least 2% of its sub-catchment area.
8.4
Extended Detention
The design philosophy for the project should include that there is no increase in downstream channel erosion. The effect of not providing the full EDV in each wetland on the downstream channel should be checked (for example by checking the length of time that channels are exposed to erosive velocities before and after development) and, if required, achieve the design philosophy by other means.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
18
Appendix A: Record of Site Visit by PDP PDP visited key points of the Business Plan area on 5th April 2011 to gain an appreciation of the existing land-use, current hydrological and hydraulic conditions and the potential effects of the development proposal on the surface water regime. The key points visited included: • • • • • Hingaia Stream at Ararimu Rd; Maketu Stream at Dale Road; Hingaia Stream at Quarry Road; Low lying areas in the Drury township; An overview of the development area from MacWhinney Drive.
Hingaia Stream at Ararimu Rd
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
19
Hingaia Stream at Quarry Rd
View overlooking central part of the Business Park from MacWhinney Road
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
20
Appendix B: Questions and Comments with Response from DHI PDP supplied DHI (copy to Beca) a list of comments and questions that arose from the initial review of the reports supplied on 19th April 2011. These questions and comments were supplied in order to seek a better understanding of the modelling work that was done and to augment that information that was contained within the supplied reports. In addition to several specific questions and comments listed, PDP supplied three overall comments on the modelling work reported on in the documents that were reviewed. These are stated in the excerpt from the email message below (only the first page of this email - containing the three comments is reproduced):
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
21
Responses to the initial list of comments and questions were supplied via email from DHI on 3 rd May 2011. This email message is reproduced below, with the text of the response to follow on subsequent pages. Note that every initially posed question or comment is numbered, with the response from DHI immediately below each. Several additional questions and comments, that followed the list initially compiled, are shown in red. Collectively these make up the complete list of questions and comments mentioned in Section 1.6.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
22
Responses to PDP’s comments and questions
Section 1: Comments 1. In Figure 1-1 the catchment boundary appears to cross a stream at two locations (approx midway along western boundary and in north-east corner). This appears anomalous. Is there an explanation for this? The stream polyline used for this drawing is incorrect (not the catchment boundary). The stream polyline was included in the data supply but does not account for culverts through the motorway which we have accounted for in our catchment delineation. 2. Section 1.5: I have not seen the GAP analysis report. I presume that the Technical modelling report is contained within the document referenced above. The gap analysis report was delivered at an earlier date. The main document contains the technical modeling details. 3. Section 1.6. Note the LiDAR accuracy. Because of this, is it possible to quote flood levels to any higher degree of accuracy? Is the LiDAR accuracy plus or minus, or does this represent the range of values?
4. Section 1.6. I note the initial assumption of a dry floodplain. Some comment on how realistic/likely this is would aid understanding. Accepting that a design storm philosophy is applied this appears realistic, but there is no comment on this in the report. The dry floodplain assumption is suitable for the design storm. We start getting into combined probabilities when considering wetting before the main storm. If using a historical event it would be suitable and possible to set appropriate initial water level conditions in the flood plain. 5. Section 1.6. Note the comment “... it is not possible to quantify the accuracy of the model.”. This needs to be borne in mind when referencing results. Agreed
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
23
6. Section 2.6. I agree with the omission of smaller stormwater pipes for significant events, but am not convinced on the results of 5-year ARI rainfall events being necessarily valid in this case, as I would anticipate that much of the local reticulation has been sized for an event of 5-year ARI. In this case I can agree that the model results will be conservative in that higher peak flood levels will be the result, although the degree of conservatism has not been quantified. Yes although much of the catchment is still rural and thus the pipe network extent is reasonably small. We are also planning on recalibrating the model with a smaller rainfall event to ensure the model is appropriate to use in smaller rainfall events. 7. Section 3.1. A minor point, but roughness appears to have been omitted as a hydrological parameter that is used with the kinematic wave (Model B). Roughness was not used as a calibration parameter in the hydrological model. General values were assigned based on the roughness characteristics of the landuse. The infiltration rates were found to be the most sensitive values in the calibration, which is typical for Model B. 8. Section 3.3.3. The way I understand it, the 1D reach can fill to bank level before the 2D model takes water. The 2D grid covers the 1D portion of the model (“... Once water enters the 2D model it is free to flow over the 1D model”). In the 1D model, will water level not rise as if constrained within “glass walls”, thereby double-counting the volume that exists over the 1D model between the banks? If so, I presume this effect is small, but I’d appreciate your comment on this. To me this would result in unconservative results, as volume is essentially spread between two different models where in reality it can only exist once. The bank markers in the 1-D channel are assigned so that only the main channel is modeled in 1-D. In addition, where the 1-D and 2D models are linked, the 2D model ground level is raised to the bank spill level (i.e. the river is filled in). So only the volume between the bank markers and the above the bank spill level is “double counted”, which turns out to be a very small amount. In addition the mass balance for the model has been checked and the error is less than 5%. 9. Section 4.1.2. This appears anomalous to me. The way I understand it, the hydraulic model was adjusted until the
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
24
model results fitted with measured data. This presumes that the hydrological model was accurate, and also that channel dimensions must have been as modelled (without knowledge of this). Surely the dimensions from stream improvement works should be modelled and not allowed to change, with other variables being tweaked to effect calibration? Interested in your opinion here. We had no data at the time on the 1988 stream dimensions. We agree that it would have been suitable to use the real data. 10. Section 4.1.3. Model results relative to measured data are compared. In many cases the differences are small, yet there is still significant uncertainty in the modelled results – the terrain used is only accurate to 0.5m in rural areas and 0.25m in urban. I suggest that “modelled results” should be quoted as a range rather than as an absolute, and ideally the range should envelope the measured data. This was discussed at the meeting on April 28th. Although a range of results may be more correct it is simpler to present a single value with the understanding that it includes uncertainty, by using a freeboard. 11. Section 4.2.1. I struggle to understand why the volumes generated using the TP108 rainfall profile should compare with those for the 1988 event. There is a disconnect between “design” and “actual” performance. Agreed but we wanted to check to see if volumes were of a realistic magnitude by comparing to a real event. 12. Table 4-5. The results from the December 1988 event are compared to those for a 100-year design rainfall event (TP108). This comparison appears meaningless as the 1988 event is quoted (Section 4.1) as being of 50-year ARI. Furthermore, comparisons are only made for Catchment 51, which by eye seems to be either largest or second largest and not necessarily representative of all Hingaia Stream catchments. Perhaps it would have been more suitable to use a 50 year event for this comparison however the table 4-5 is illustrating the differences in sensitivity to the infiltration rates for the different storm shapes, real versus design. Because of size, catchment 51 gave the biggest differences and thus made it easy to see a trend appear.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
25
13. Section 4.2.1. Below Table 4-5 it is stated that to calibrate infiltration rates it would be necessary to find an historic rainfall pattern with similar shape to the TP108 (Chicago) profile. Why could the actual rainfall data (hyetograph) from the 1988 event not be applied to the model? Am I missing something? The calibration was done using the 1988 rainfall. However we are just stating here that the design and real storms are not alike at all and that the response to a design storm would be different to that of a real storm. 14. There are two tables numbered “Table B-3”, one on page B-1 and the other on page B-3. I presume the table on page B-1 is Table B-1. Yes 15. Section 4.2.1. The model is said to have been run for 3 different sets of start/end infiltration rates, however none of those quoted match those presented in Table B-1 (labelled as B-3, see note above). Were these applied to all catchments, regardless of soil type? If so, could you justify this, please? For the higher infiltration soils we used start/end infiltration of 12.5/1.5 mm/hr and for the lower infiltration soils we used start/end infiltration of 7.5/1.2 mm/hr. Perhaps it is a bit unclear that the rates stated in the table are the range of rates used over the catchment. 16. Section 4.2.2. The 1988 flood event is quoted as having ARI of between 50 and 100 years, yet Section 2.8.1 states the ARI as being 40-50 years. Which is it? 40-50 as stated in the Snelder report. 17. I think that a differentiation is required between rainfall ARI and flood ARI. If the frequency analysis is performed on rainfall data then the floods being simulated are those that occur in response to the n -year rainfall ARI. If the discharge records (of which there are none) are analysed then the flood may be assigned this probability. I know this is a fine point. Yes, however as you say we have no flow record. Perhaps it just needs to be made clearer that we are modeling the 100 year Rainfall ARI and not necessarily the 100yr Flood ARI. 18. Section 4.3 also mentioned that the 1988 event was simulated to be of ARI between 50 and 100 years, while previous references indicated it to be of 40-50 year ARI.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
26
Refer to answer on comment 16 19. Section 4.3. The “mid-range” of infiltration rates has been quoted as being used, but the range itself was purely arbitrary, making “mid-range” equally arbitrary. Is it not possible to apply calibrated parameters from similar soils in a nearby catchment to the subject area? Given that modelled and recorded 1988 events appear to be of different ARI, this possibly has relevance. We did compare infiltration rates to a flood study we did on the nearby Papakura Stream catchment and with literature. We plan to use the Ngakaroa Stream flow site to help give a better idea of the infiltration for some of the soil (mainly the hilly areas of the catchment). 20. Section 4.3. I note that modelled peak water levels are quoted as being within 300mm of observed. If model uncertainty is taken into account (eg 0.5m accuracy in terrain), this represents a good fit. I suggest that model results be quoted as plus or minus, and ideally these extremes would envelope the observed values. True. 21. Section 4.4.1. I note that sensitivity to infiltration rate was tested by application of an increase to the values used. Could/should a decrease also have been applied? I would expect that the response to changes in infiltration rate would not be linear, and that doing both increase and decrease would provide greater clarity. I am interested in your comments. We agree in hindsight that it would be useful to also test the sensitivity to the decrease in infiltration rates. I believe the omission was probably due to a limit on the number of sensitivity tests we were to run. 22. Section 5.3.1. All floodmap figures appear to have a blue colouring (in the legend) for flood depths “below 0.0m”. To me this means that the area is dry, yet is plotted with a blue colour. Could you explain this, please? I also note that these maps appear to have been produced at a discrete time step, rather than having been plotted at maximum water levels across the entire simulation. I accept that the flooded areas being shown in each are relatively small, but can you confirm that the maximum occurs at the time step selected in each case?
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
27
The legend is just being misleading here, it should say greater than 0. The timestep closest to the peak was chosen but obviously the peak will occur at different times in different areas. These plots are here to show the flow patterns by using the flow vectors. The maps in the floodmaps section show the full maximum flood depths. 23. Figure 6-1. The legend indicates that flood depth is plotted but I think flood level is intended. While plotting flood level is useful to see hydraulic grade, plotting flood depth in addition may aid with understanding. Yes the legend is incorrect. Also see response to 22. 24. Section 6.2.5. The results of this simulation are surprising. I presume that significant hydraulic resistance exists in the unmodified case downstream of the proposed diversion, and that this diversion relieves this. I presume that the same tidal boundary condition was used for both cases. Long sections of each would aid with understanding of this. Yes there is significant hydraulic resistance in this area. We can provide additional long sections if necessary. 25. Section 7.2. The report states that “... considerable uncertainty exists in the model predictions”. I suggest that this point be highlighted, perhaps by quoting model predictions as falling within a range, particularly when absolute values are being derived. Noted [Mark Pennington] I note that the status of the March 2011 report is “draft” on the front cover and “final draft” in the quality assurance statement on the first page. We can release this as final now. [Mark Pennington] Section 2 in the March 2011 report states that no buildings were blocked out in the DSBP area. I presume this is because flood levels are all below proposed ground level. Is this true? Yes flood levels are below the proposed ground level, thus it makes no difference if the buildings are in the model or not. [Mark Pennington] In Figure 3-1 of the March 2011 report a larger head difference (for DSBP up- and downstream) across the structure appears to correspond to a smaller peak discharge. Could the legend be confirmed here, please? Will need to look into this further
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
28
[Mark Pennington] In the conclusions of the March 2011 report it is stated that flood risk is reduced downstream by “various aspects” of the DSBP. It should be noted that this is in response to design event rainfall and may not necessarily apply to actual rainfall. OK
Section 2: Questions 1. Section 2.2.4. Please supply a longitudinal section to show the downstream extension to the model with tide levels. This will allow the reviewer to assess effects of tidal boundary assumption. Provided 2. Section 2.3.1. Does raw LiDAR data set support 1m x 1m grid interpolation? What is the density of raw data points? Was the 5m x 5m grid interpolated off the 1m x 1m grid, or was it reinterpolated from the raw data? What interpolation method was used, and were breaklines applied? We will need to dig out the LiDAR again to confirm the point spacing. LiDAR points were built into an Arc GIS terrain which uses the entire dataset. The 1m and 5m grids were interpolated using the Natural Neighbour interpolation transferring from the terrain to raster. Significant highpoints i.e. banks and roads were traced from the 1m grid and copied into the 5m grid so that spill levels had higher accuracy. Breaklines were not used, however the resolution of the point data and the additional road/bank tracing was believed to generate a suitably accurate representation of the surface. 3. Section 2.5.1. How was chainage measured for 1D model? Is this centreline distances? Distance along thalweg? Are all cross sections normal to this line? Thalweg, and the cross sections were measured normal to this line. 4. Validation event – how was rainfall applied to the catchment? Off one gauge, fitted Chicago hyetograph? Two rainfall gauges (Ngakaroa and the Filter Station) were used and weighting based on the location of the catchments relative to the gauges.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
29
5. Section 2.8.1. Were the peak flow estimates by Snelder obtained using a nested Chicago rainfall profile? Snelder mentions that he used a design storm profile based on a pattern developed by the ACC around 1988. He does not mention whether a Chicago profile has been used. We can supply you with a copy of the Snelder report if you would like more information (page 46). 6. Section 2.9. For the Slippery Creek model (HEC-RAS), were the hydrological parameters and assumptions applied compatible to those of the current study? Same rainfall depths/profile? This is unknown, Larry Shui supplied the hydrographs to us directly. 7. Section 3.2.2. How was Horton decay applied? Was the decay exponent one derived from calibration of another catchment, or was the DHI default value applied? Where did the 5.5 hours come from? The decay exponent was based on that used in other studies, including the Papakura Stream study referenced before. 8. Table 3-1. Start and end infiltration rates for 3 different soil classes are the same. Can you confirm these, and provide justification for why they do not differ? Yes these are the values used. The values were grouped to simplify the model, so that there are two soil types, those with high infiltration and those with low infiltration. 9. Section 3.3.1. Report states that “The majority of the model runs have been updated to include the Ngakoroa Stream”. Could you elaborate on which runs do and do not include this, please? All models have now been rerun to include the Ngakoroa Stream, at the time of writing the report the DSBP options did not include the Ngakoroa Stream but have since been rerun to include this. 10. I may have missed it, but in Section 3.3.1 energy losses are quoted as having been applied at entrance and exit to closed sections. What coefficients were used? I assume these are applied to velocity head – could you confirm please?
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
30
Yes the loss is applied to the velocity head, the coefficients are 0.5 at the entrance and 1 at the exit. 11. Section 3.3.2. Second bullet point states that “... or LiDAR data has been processed incorrectly”. Could you explain how such areas may be identified, and how the effects of this might impact on results? Figure 3-2 gives a good example of this. The grid can be checked by eye to find very obvious errors or large areas where data points are missing. If areas like this are not identified they could have an impact on results especially if they are in the flooding region. 12. Section 3.3.2. Second bullet point. Where did the “correct” road spill levels come from? Have these levels been taken from the raw LiDAR, from survey or from interpolated LiDAR? What accuracy can be assigned to these levels? The road spill levels were used from the bridge survey or where this was not available from the nearest road level LiDAR point. 13. Section 4.1. Was land use in the 1988 event the same as it is currently? If not, was the hydrological model used for the 1988 event adjusted to reflect land use at that time? This point emerges again in Section 4.1.1 – ie was same percentage impervious applied for 1988 event as for the design scenarios? The land use and impervious areas for the 1988 event are different to the existing event, to reflect the land use scenario at this time. 14. Section 4.1.3. Could you please explain how the “calibration” (ie channel dimension modification) that was undertaken to get measured and modelled results to match for a small portion of the 1D model was “... transferred to the full MIKE FLOOD model...”? Were all channels widened by the same amount? Were all roughnesses decreased by the same amount? The changes made to the downstream end of the 1D model (around the motorway bridge) were copied into the full 1D model used in the MIKE FLOOD simulation. 15. Section 4 in general. What rainfall profile was applied to the 1988 event? Was this measured at a raingauge, and if so, which one? Was there any spatial variation of rainfall in this event? I note that no hydrological model calibration was undertaken (admittedly due to a lack of suitable data), but the model validation appears to have been done on the assumption that the hydrological model needed no adjustment,
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
31
with all changes being made to the hydraulic model. Could you comment on antecedent conditions for the 1988 event? Two rain gauges were used for the 1988 event, the Ngakaroa gauge (at the south end of the catchment) and the Filter Station record (north end of the catchment). The rainfall at the two sites had a different pattern. The Snelder report states that the antecedent wetness in the catchment was high due to continuous heavy rainfall before the storm. 16. Section 4.3. There is no mention of hydrodynamic parameters applied in the models. Could you please provide information on these, including eddy viscosity (flux/velocity based?), and whether or not “default” values for parameters (such as delta, theta, etc) were used? Was super-critical flow allowed in the model and were there any locations where this occurred? Eddy viscosity: velocity based, value of 1m2/s. 1D model parameters were kept at default values except delta which was increased to 0.8 for stability. Super critical flow is allowed in the model. We have not looked into super critical flow in depth since the model build stage however as I recall the Froude numbers were below 1, but they were reaching a reasonably high value at the downstream end of the Drury Town upstream of the motorway bridge. 17. Section 5.1.1. Last two paragraphs. Initially I was surprised that the model was adjusted to not reflect the case being modelled, but am I correct in assuming that these changes to model dimensions are recommended changes to be made to the final design concept to achieve an acceptable result? The DSBP design concept has been updated since this report was written. The raising of the road was made so that we could determine the highest level the flood waters would reach without overtopping the road, in order to aid design. 18. Could the actual floodmaps for Scenario 4 be plotted, please, rather than just the differences? Where differences are plotted it seems to highlight areas where notable terrain changes have taken place, making it difficult to really assess the effects of the changes. Refer to the floodmaps in the March 2011 report as this reflects the latest DSBP design model. Maximum depth is plotted as well as differences in flood level.
A02426300R001_DSBP SW Peer Review final final 270611
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
32
19. Section 5.3.4. Ballard’s Bride is stated as acting as a control point. Could you supply Froude number over the simulation at this point, please? The latest simulation was not saved with Froude numbers, however a hand calculation gives a number of around 0.45 at the time of peak, at the point directly upstream of the bridge. This is for the DSBP scenario. 20. Section 6.1.1. Was the hydrology for the Ngakoroa Stream done in the same way as that for the rest of the catchment? Was the same rainfall applied? Again, the flows for the Ngakoroa Stream were supplied by Larry Shui, I don’t have more information on the modeling. 21. Section 7. Could you please confirm that hydrodynamic parameters were kept constant across all models for all simulations? Yes, except roughness was changed for the DSBP scenario runs to reflect the changes to landuse. 22. Could you please supply a model plan plot showing branch names and chainages (even just the .nwk11 file) to aid with location of reaches in the long section(s)? supplied [Mark Pennington] Section 2 in the March 2011 report, Table 2-1, gives bridge bottom and top levels for Beca Bridge as the same. Does this mean that deck underside and roadway level are the same? This was done at a request from Beca, however the point is moot given the flood level does not reach the soffit level of the bridge. Mark Pennington] Section 3 in the March 2011 report. Could you indicate start water level in the wetland, and whether or not spring flows are present that influence baseflow? Dale may be able to give more insight here, the wetland is modeled as “dry” for the start of the simulation however the ground level of the wetland area is in fact the dead storage level.
Email outlining results of scaled 1988 flood event and subsequent discussions
A02426300R001_DSBP SW Peer Review final final 270611
From: Antoinette Taylor [mailto:[email protected]] Sent: Wednesday, 22 June 2011 11:38 a.m. To: Mark Pennington Subject: RE: Drury
Hi Mark,
Did you note the same effects relating to timing of peak flows from sub-catchments for the 1988 event as for the design (TP108) events? If not then could you comment?
The timing of the peaks in the 1988 event are different to those in the design event. The 1988 event rainfall is double peaked, with the second peak being higher, overall resulting in a flatter flow hydrograph in the model. The timing of the second peak in the tributaries lines up with the Hingaia Stream flow peak at the downstream end resulting in the higher water levels.
The scaled 1988 event simulation results appear to show a flatter hydraulic grade overall, with difference appearing to increase with distance upstream. Is this observation correct?
Yes. See below, total rainfall depth is less in the 1988, especially in the upper catchments, resulting in the lower water levels. The water levels at the downstream end are higher due to the differences in timing in the events, leading to a closer synchronisation of the flow peaks.
How do the rainfall totals compare between the scaled 1988 event and a design 100 year event (the way I understand it, rainfall was scaled until peak discharge matched)?
Yes we scaled the rainfall until the peak discharge at the downstream end of Hingaia Stream was matched. Both sets of rainfall are distributed, however the total rainfall depth of the design storm was greater than the scaled up 1988 rainfall. The TP108 rainfall ranges from depths of 180mm to 225mm. The scaled 1988 rainfall ranged from 130 to 175 mm. I am out on site for the rest of the day, so I hope these answers are satisfactory. Kind Regards
Antoinette Taylor
Project Engineer
From: Mark Pennington [mailto:[email protected]] Sent: Wednesday, 22 June 2011 10:48 To: Antoinette Taylor Subject: FW: Drury Hi Antoinette I have received the appended email via several people, and just have a few questions relating to this. Did you note the same effects relating to timing of peak flows from sub-catchments for the 1988 event as for the design (TP108) events? If not then could you comment? The scaled 1988 event simulation results appear to show a flatter hydraulic grade overall, with difference appearing to increase with distance upstream. Is this observation correct?
How do the rainfall totals compare between the scaled 1988 event and a design 100 year event (the way I understand it, rainfall was scaled until peak discharge matched)? Cheers for now Mark
Mark Pennington MIPENZ CPEng IntPE(NZ) | Engineer PATTLE DELAMORE PARTNERS LTD 1st Floor ITM Building Centre, 218 Beach Road, Kaikoura | http://www.pdp.co.nz +64 3 319 3114 | +64 21 063 2112 |
[email protected]
DISCLAIMER: Any views expressed in the email may be those of the individual sender and may not necessarily reflect the views of Pattle Delamore Partners Limited. This electronic mail message together with any attachments is confidential and legally privileged between Pattle Delamore Partners Limited and the intended recipient. If you have received this message in error, please e-mail us immediately and delete the message, any attachments and any copies of the message or attachments from your system. You may not copy, disclose or use the contents in any way. All outgoing messages are swept by an Anti Virus Scan software, however, Pattle Delamore Partners Limited does not guarantee the mail message or attachments free of virus or worms.
From: Roger Seyb Sent: Monday, 20 June 2011 11:00 a.m. To: Mark Pennington Subject: FW: Drury
Roger Seyb Surface Water Engineer PATTLE DELAMORE PARTNERS LTD DDI Mob +64 9 523 6928 +64 21 540 993
From: Stephen Priestley [mailto:[email protected]] Sent: Monday, 20 June 2011 10:58 a.m. To: Roger Seyb Cc: Andrew Nell Subject: FW: Drury Roger The following is the response from DHI. I think it is a considered response. I hope it helps you to finalise your review (by Wednesday). Cheers Stephen From: Antoinette Taylor [mailto:[email protected]] Sent: Monday, 20 June 2011 10:53 a.m. To: Stephen Priestley Subject: RE: Drury
Hi Stephen,
The following is our response to the calibration comments and the model results for the scaled 1988 simulation: In the context of a ‘classic’ model validation exercise one would have a number of observed events by which to adjust model parameters to achieve a fit against those observations (calibration) and a number of observed events by which to compare model predictions with those observations and comment on the validity of the calibration (verification). The combined process is validation of the model.. Often, storm water studies of require the development of a model for which there are less than ideal observations on the performance of the system. In these cases it is not possible to carry out a ‘classic’ calibration and verification processes and any efforts to prove the validity of the model is then simply referred to as validation – without explicitly undertaking calibration and verification. This was the case here and it is necessary to therefore make a number of changes to model parameters and prove those changes all in one simulation. Our strategy in the project was to test the model – as built using available data and information against measured water levels (peak levels likely to be from debris marks with an uncertainty estimated to be of the order of +/- 100mm at best. Our testing found that the model predicted water levels in the up-stream reach of the river to be acceptable whereas the model was underpredicting water levels in the vicinity of up-stream of the bridge. From this we concluded that the model set-up and parameter choice up-stream of the Drury town were acceptable and that there was some miss-representation of the channel geometry and hydraulics in the model in the vicinity of the bridge. Following investigation we discovered that the channel in the vicinity of the up-stream of the bridge had been modified following the 1988 event. There were no actual pre-modification cross sections available but we made the assumption that the cross sections could be approximated to be of similar cross section shape to those upstream of the modified reach. We also represented the hydraulics of the confluence by using an energy loss description in the model Together, these changes to the model resulted in an agreeable match in water levels for the 1988 event. To this end the model was considered as validated against what (limited) data was available. The 1988 storm event has been run through the DSBP design model, where the 1988 rainfall was scaled up to match the 100 year flow at the downstream end. The table below shows a comparison of water levels at the DSBP site. Overall there is a decrease in levels upstream of the project area, as you move downstream the water level difference decreases and at the downstream end of the project the 1988 water levels are higher than the 100 year. Location 100yr DSBP Scaled 1988 Difference scenario Water Water Level Level 23.83 23.49 -0.34 20.5 19.08 17.34 20.19 18.83 17.16 -0.31 -0.25 -0.18
Upstream of Ararimu DS Ararimu Bridge, north end of project Upstream Beca Bridge Ponding area Upstream of Ballard’s Bridge
Confulence of quarry diversion stream and Hingaia Downstream end of Project Drury Town
15.46 10.58 6.91
15.4 10.61 6.98
-0.06 0.03 0.07
Kind Regards
Antoinette Taylor
Project Engineer
From: Stephen Priestley [mailto:[email protected]] Sent: Monday, 20 June 2011 10:46 To: Antoinette Taylor Cc: Andrew Nell Subject: Drury Antoinette As discussed on Friday, I requested the following information by 10am today. Could you please respond as this matter is urgent for our client. • • A graph showing the sensitivity of the assumed channel dimensions of the upgraded works (as per Papakura DC suggestion) to the flood levels. To model the 100 year event based on scaling up the 1988 rainfall hyetographs so as to match the Q100 flows at the downstream bridge. Then assess the difference in the levels (between the proposed Q100 levels and the levels derived from the up?scaled 1988 event) within the DSBP area.
Cheers Stephen Priestley Technical Director Beca Phone +64-9 300 9000 Fax +64-9 300 9300 DDI +64-9 300 9282 [email protected] www.beca.com Company of the Year Award 2010 // Deloitte/NZ Management Magazine Top 200 Awards
- -----------------------------------------------------------------------------------------------------------------------------------------------------
PATTLE DELAMORE PARTNERS LTD Stormwater Management Report Peer Review
34
Appendix C: Record of Teleconference held on 28 April 2011
A02426300R001_DSBP SW Peer Review final final 270611
doc_884748460.pdf