Logan River Catchment Hydrological Study
Total Page:16
File Type:pdf, Size:1020Kb
Logan River Catchment Hydrological Study December 2014 1 Title: Logan River Catchment Hydrological Study Author: Study for: City Planning Branch Planning and Environment Directorate The City of Gold Coast File Reference: WF15/44/02 (P1) TRACKS #45737331 Version history Changed by Reviewed by & Version Comments/Change & date date 1.0 Draft 2.0 Review 3.0 Review 4.0 Peer Review 3.0 Review 4.0 Review Distribution list Name Title Directorate Branch Version 3 TRACKS-#45737331-v3- Page 2 of 175 LOGAN_RIVER_HYDROLOGICAL_STUDY_DECEMBER_2014 1. Executive Summary Overview The main objective of the study is to develop a suite of hydrological models for the City of Gold Coast (City) that are based on one type of modelling software (URBS), calibrated and verified against available data, and fully documented to a consistent standard. The calibrated model is to be used to estimate design flood discharges for design events ranging from 2 years Average Recurrence Interval (ARI) to PMF. The hydrological modelling of the Logan River catchment was undertaken using an approach and methodology consistent with the other catchments in the City area. For this study, the Logan River catchment has been developed as a single URBS model for the whole Logan River Catchment (consisting Upper Logan, Teviot Brook, Albert River and Lower Logan). The model parameters were kept global for the whole catchment, and the model configurations were kept as simple as possible. The URBS models have been configured based on current catchment land uses. Model Calibration and Verification Calibration data used in the Logan River Hydrology study 2009 (Ref. 2) were available for use in the current study. All the available data were sourced from BOM for a total of 26 flood events between 1947 and 2013. The available rainfall and stream flow data for early events are very limited and of poor quality. Therefore the selected calibration and verification events are generally the most recent events, with the exception of the January 1974 event, which is the largest on record. From the available data, five flood events (January 1974, April 1990, February 1991, January 2008 and January 2013) were selected for model calibration and a further four events (May 1980, April 1988, March 2004 and January 2012) were selected for model verification. The selected events cover a wide range of discharges across the whole Logan River catchment. The emphasis of the model calibration was to achieve the best possible fit between the predicted and recorded discharge hydrographs at key stations along the main streams of the Logan River catchment for the selected calibration events. For these stations, the calibration attempted to match the predicted and recorded flood peaks and volumes, and also the shape of the hydrographs. The calibrated model was then verified by comparing the model predictions against the discharge hydrographs recorded at various gauging stations for the selected verification events. Due to the lack of available rainfall data for most events and the lack of detailed rainfall data where data were available, the URBS model cannot be expected to accurately reproduce flood behaviour for all events and at all gauging stations. As such more calibration emphasis was placed on large events, as the accuracy of small events are impacted significantly by spatial and temporal variation in rainfall. A single set of model parameters were adopted for the model, and maintained for all calibration and verification events. The model parameters were adjusted to achieve the best calibration across all events, resulting in a compromise between model accuracy and model simplicity. It is noted that calibration of the model for gauging stations in different parts (Teviot Brook, Upper Logan, Albert River and Lower Logan) of the Logan River catchment can be improved by adopting different sets of model parameters for each part of the catchment. Further the calibration of the model for individual events can be improved by adopting a different set of model parameters for each of the different events. The adopted model parameters are given below: Version 3 TRACKS-#45737331-v3- Page 3 of 175 LOGAN_RIVER_HYDROLOGICAL_STUDY_DECEMBER_2014 Parameter Adopted Value (Channel Lag Parameter) 0.2 (Catchment Lag Parameter) 2.5 m (Catchment non-linearity Parameter) 0.75 F (Forest Factor) F*0.5 Rainfall losses were adjusted to achieve the best possible hydrograph shapes and flood volumes. A uniform initial loss and continuing loss rate were adopted for the model and each flood event. It is noted that calibration of the models for individual events can be improved by adopting a set of variable loss rates within the catchment for each of the different events. Calibration Results Satisfactory calibration was achieved throughout the catchment, with the URBS models generally reproducing recorded flood discharges adequately. The model calibration for Teviot Brook is generally good, considering that a single set of global model parameters were adopted across all nine historical events. The model calibration results for large events are excellent; however the model results for smaller events show the predicted flood peak arriving earlier than the recorded peak. This issue could be addressed, if necessary, by adopting different model parameters including initial and continuing losses for smaller events. The model calibration at the Broomfleet and Wolffdene is generally good even though some inconsistencies between recorded peak discharges at Bromfleet and Wolffdene are evident. The model predicted excellent result at Broomfleet and Wolffdene, however results show early peak for smaller events. The model calibration for the Upper Logan catchment is generally excellent. Although calibration for the large events (January 1974, April 1990 and February 1991) was generally quite good at both Yarrahappini and Round Mountain gauging stations, the model predicted a lower peak at Round Mountain and a slightly higher peak at Yarrahappini for smaller events (January 2008 and January 2013). The quality of available data (both rainfall and stream flow) for the Lower Logan catchment is not as good as for the upstream of the catchment. For the January 1974, May 1980 and April 1990 flood events, the complete water level hydrographs were not recorded at the Macleans Bridge and Waterford gauges (the recorded water level hydrograph at Waterford for the January 1974 flood has been synthesised based on debris marks at the gauge site). Complete recorded water level hydrographs at both stations were available for the April 1988, February 1991, January 2008, January 2012 and January 2013 events. The two gauging stations with available data (Macleans Bridge and Waterford) are BOM flood forecasting stations, not Department of Natural Resources and Mines (DNRM) stations, and as such the quality of rating curves for these two stations is uncertain. The rating curve for Macleans Bridge appears to be acceptable; however the Waterford rating is poor, especially for higher discharges. As a result the calibration is generally good at Macleans Bridge, but somewhat poor at Waterford. The Waterford gauge is tidally affected, and may also be affected by downstream water levels (including Albert River outflows). The timing and shape of the predicted Version 3 TRACKS-#45737331-v3- Page 4 of 175 LOGAN_RIVER_HYDROLOGICAL_STUDY_DECEMBER_2014 hydrographs at Waterford is generally acceptable but the predicted peak discharges are generally higher than recorded peak flows for the larger events. Predicted peak discharges at Macleans Bridge are generally good, indicating a problem with the Waterford rating curve. Further, the recorded flood volumes at Waterford appear to be less than those recorded at Macleans Bridge. Design Flood Discharges The calibrated URBS model was used to estimate design flood discharges throughout the Logan River catchment based on design rainfall intensity – frequency – duration (IFD) data from a number of sources. Design flood discharge hydrographs were estimated for a range of storm durations up to the 120 hour event for the 2, 5, 10, 20, 50, 100, 200, 500, 1000 and 2000 year ARI events, Probable Maximum Precipitation Design Flood (PMPDF) and PMF events. The design rainfall data and associated procedures and input data (including IFD data, temporal patterns, areal reduction factors, and rainfall spatial distribution and design rainfall losses) adopted in the study are based on a comprehensive review of the latest available data and information. A comparison of the estimated peak design discharges from this study with the peak design discharges reported in three previous studies (WRM 2009 – Ref. 2, Sunwater 2007 – Ref. 5 and AWE 1997 – Ref. 4) was undertaken. A summary of findings with respect to the discharge estimates from the different studies is given below. Note that the AWE (1997) and WRM (2009) studies covered the whole Logan River catchment, whereas the Sunwater (2007) study was restricted to the Teviot Brook catchment. The URBS model estimated peak design discharges for Teviot Brook at The Overflow match closely with the estimates given by the Sunwater (2007) and WRM (2009) studies for all ARIs up to, and including, the 2000 Year ARI event. The URBS model estimated peak discharges at Bromfleet, Wolffdene and Beenleigh match the AWE (1997) estimated peak discharges reasonably; however the URSB model estimated peak discharges at these stations are higher than those estimated by the WRM (2009) study. This is likely due to differences in