Rajulapati C., Tesemma Z., Shook K., Papalexiou S. and Pomeroy J.W. (2020). Climate Change in Canadian Floodplain Mapping Assessments. University of Saskatchewan

In the recent decades, precipitation patterns and corresponding streamflow responses in many cold regions catchments have changed considerably due to warming. Understanding historical changes and predicting future responses are of great importance for planning and management of water resources systems. Regional climate simulations using convention- permitting models are helpful in representing the fine-scale cloud and mesoscale processes, which are critical for understanding the physical mechanisms that cause in convective precipitation. From a hydrological perspective, these fine resolution simulations are helpful in understanding the runoff generation mechanisms, particularly for mountainous watersheds, which have high spatial variation in precipitation due to large differences in elevation over small distances.

Natural Resources Canada (NRCan) is developing the Federal Flood Mapping Guidelines Series to support the National Disaster Mitigation Program (NDMP) led by Public Safety Canada (PS). These documents are developed through consultation with practitioners and stakeholders including the Federal Flood Mapping Committee (FMC) chaired by NRCan and PS, and the 175 members of the Technical Working Group (TWG). The TWG comprises technical experts and practitioners from across Canada, and includes several sub-groups including the Climate Change Sub-Group. The Guideline Series is intended to move toward common practices in flood mapping across Canada and are published for all Canadians. NRCan published the “Case Studies on Climate Change and Floodplain Mapping” in 2018, which offered insight into incorporating climate change projections into flood mapping studies at three locations in Canada. This was followed by a publication by the National Research Council (NRC) in March 2019 titled “An Inventory of Methods for Estimating Climate Change-Informed Design Water Levels for Floodplain Mapping”, which describes current practices across Canada for incorporating climate change into flood mapping studies.

Global Water Futures (GWF) conducted a study on “Historical and Future Flow Regimes at the Bow River in Calgary” referred to as the Bow River Basin Study (BRBS) with funding support from NRCan’s Climate Change Impacts Adaptation Division (CCIAD). This project offered insight into the effects of climate change on flow in that watershed. GWF indicated that the ‘next steps’ for that work were to: prepare a case study report on how climate change may affect future flood flows that could be applied to floodplain mapping; and, detail how climate change can be downscaled and applied in large scale hydrological assessments of impacts on hydrological regimes.

The sister-study of this report, the Bow River Basin Study (BRBS), used a physically based hydrological land surface scheme along with a water management model, coupled with a high resolution convention- permitting atmospheric regional model (Weather Research and Forecasting, WRF) to understand the streamflow generating mechanisms and identify the changes in streamflow responses of the Bow and Elbow River Basins. The coupled model appears to provide a large improvement in predictability, with minimal calibration of parameters and without bias correction of forcing from the atmospheric model. The model 4 was able to provide reliable estimates of streamflows, despite the complex topography in the catchment. Using the WRF Pseudo Global Warming (PGW) scenario, estimated future streamflows simulated were then used to develop projected flow exceedance curves. The uncertainty in the simulations is extremely helpful in the risk assessment for downstream flood inundations. However, the uncertainty in streamflows cannot be assessed as the WRF- PGW dataset was only available for a single realization, because of the high computational cost.

The research presented in this report focusses instead on using the highly efficient hydrological model developed and verified in BRBS whilst assessing uncertainty using another regional climate model, the CanRCM4, where many realizations are available for different boundary conditions. Since the CanRCM4 simulations have a relatively low resolution, a novel methodology was developed to adjust regional climate model outputs using the WRF-PGW data. An ensemble of 15 CanRCM4 simulations was used to force the Bow River basin model to determine a measure of the uncertainty in the simulated streamflows, and the projected streamflow exceedance probability curves. These curves are extremely useful for risk assessment for downstream flood inundations. Given the importance of understanding how much extreme precipitation will change in urban areas of the basin, where short duration high intensity events cause flash flooding, frequency analysis of these events was carried out for Calgary and Intensity Duration Frequency (IDF) curves were developed. A ready-to-use empirical form of IDF curve has been proposed from this analysis for the City of Calgary.

The results from the WRF-PGW modelling indicated that future high flow, low frequency (exceedances less than 10%) streamflow events will decrease compared to those under the current climate condition by 4, 9 and 1.6 m3/s for the Bow River at Banff and Calgary and Elbow River at Sarcee Bridge respectively. The average of the 15 new CanRCM4-WRF-PGW results supports the above result with some greater decreases in streamflow of 9, 16 and 4 m3/s for Bow River at Banff and Calgary and Elbow River at Sarcee Bridge respectively. However, there were some CanRCM4-WRF-PGW realisations that suggested substantial increases in future low frequency streamflow from those indicated by the average CanRCM4- WRF-PGW-drive MESH model. The below average, high frequency (exceedances greater than 30%) future streamflows will increase modestly in all gauging locations by from 1 to 12.5 m3/s.

The results of the extreme precipitation analysis at Calgary indicated an increase in future extreme precipitation events of all duration and return periods. On an average an increase of 1.5 times is noted for short return periods (=2, 5), and an increase of 4 times for long return periods (=500, 1000).

Finally, this study provides a blueprint for other studies that aims in assessing the impact of future climate change on the streamflow and flood frequency analysis for major rivers that flow into urban areas and it can be taken as a pilot study for Canada. The study highlights the usefulness of multi-CanRCM4 realisations and high-resolution WRF model outputs in studies of the hydrological impacts of climate change. The flow duration curves developed from this study can be used to estimate flood frequency for floodplain mapping purposes if they are 5 used as inputs to locally developed hydraulic models of the region of interest. The methodology shown here can be applied to river basins flowing into communities of interest across Canada. As a precondition for such applications a national gridded database of high resolution (4 km) WRF model downscaled climate model outputs that have been perturbed by multiple bias-corrected regional climate model realisations should be prepared as a national forcing dataset for hydrological model applications. The MESH hydrological model evaluated here performed quite well when driven by high resolution WRF model outputs and should be examined for national application to force local hydraulic models of flood inundation elsewhere in Canada. This would be the basis for a coherent national approach to floodplain mapping that takes into account both non-stationarity due to climate change and uncertainty from climate models. Adopting this state-of-the-art approach would make Canada a global leader in assessing the risks of changing flooding due to climate change.

https://research-groups.usask.ca/hydrology/documents/reports/chrpt17_climate_change_in_canadian_floodplain_mapping_assessments_finalreport_-30april2020_final.pdf