Improving the reliability and accuracy of flood forecasting and warning systems

Hydrologic models, applied by engineers and hydrologists for flood forecasting and water budget modeling, are used to reproduce and predict the rainfall-runoff process to determine the response of the fluvial network (rivers, streams, etc.) to precipitation events. Changes in flow regime associated with precipitation events can result in flooding, drying up of stream flows, water quality degradation, loss of aquatic habitat, alterations to sediment transport processes, and damage to engineering infrastructure (e.g., bridge piers, drinking water intake pipes, water control structures, etc.). Impacts related to climate change, which includes the potential for more frequent and intense storm events in regions of North America (Kunkel 2003), further acerbates these challenges. Hydrologic modelers require reliable and accurate tools to predict and monitor flooding events in order to anticipate and mitigate these impacts. Acquiring and processing reliable precipitation input data for these models is essential to produce hydrologic modeling results with a high degree of confidence (McMillan et al. 2011).
In many regions of Canada rainfall is measured by rain gauge stations located throughout the watershed. This process presents technological limitations in the form of accuracy and resolution of the precipitation data. Recent advances in technology and computational power have allowed for the collection and processing of Doppler radar imagery to be calibrated and used to estimate rainfall accumulations. In contrast to traditional rain gauge data, Doppler radar data can provide near real-time areal and temporal estimation of precipitation. Doppler radar data is available in Canada from Environment Canada (EC), however the data is not readily accessible for processing and integration into hydrologic models since it is not made available in a compatible format. The goal of this research program is to develop reliable and effective means to process Doppler radar imagery data to produce more accurate hydrologic modeling and flood forecasting in Canada.
This study will use the Upper Thames River watershed (near London, Ontario) as a case study to develop a reliable, automated process for obtaining and processing Doppler radar data to be used for near real-time input into hydrologic models. The specific approach includes: 1) developing an automated system for collection of real-time radar data from EC stations; 2) calibration and verification of radar imagery data with measured point rainfall data; 3) data post processing techniques to prepare the data for display in GIS (Geographical Information System) software and incorporation into hydrologic models; and 4) testing the developed system by conducting hydrologic modeling for flood forecasting in the watershed.
This research will produce a state-of-the-art solution that will allow hydrologic modelers to produce more reliable and accurate flood forecasting and flood warning systems. This challenge is common to many jurisdictions in Canada. Results from this study will provide guidance and direction for hydrologic modelers across the country, serving to better protect the public and the environment, and reduce economic loss from flooding events.

Akhil Kumar
Faculty Supervisor: 
Andrew Binns
Partner University: