The capability of the radar to diagnose severe weather and extreme events of precipitation has been well documented throughout the years. When coupled with a hydrological model such as WATFLOOD, prediction of river flows also becomes possible. The hydrological model may also be used to perform an independent verification of the precipitation estimates. Unlike rain gauge networks, which provide only rainfall estimates at single points, streamflow data represent an “integration” of the precipitation over the drainage area. Therefore, the ability of the radar data to reproduce streamflow hydrographs (through an hydrological model) represents the ability of the radar to estimate the volume of precipitation across an area.
WATFLOOD is a distributed hydrological model that subdivides the watershed into grids. Therefore, it is ideally suited for use with gridded data sets, such as radar data. WATFLOOD uses the Grouped Response Unit methodology to account for land cover in-homogeneity. All areas of similar land cover within a grid (not necessarily contiguous) form a GRU, and runoff is calculated for each GRU separately. The runoff estimates for each GRU are summed to calculate the grid runoff, which is then routed downstream to the basin outlet. The model has been used for a variety of basins within Canada.
The authors have been using radar data from the King City Radar near Toronto, Ontario and the McGill Radar in Montreal, Quebec to estimate streamflow hydrographs for a number of drainage basins within the radar coverage area. The basins include a number of basins within Ontario and the northern United States. A digital elevation model was used to delineate the watersheds and calculate the watershed data for WATFLOOD. Temperature and streamflow data are also required to run WATFLOOD, and were downloaded separately.
This paper will present the results of using radar data in a hydrological model to estimate streamflow. In addition, the effect of using various automatic algorithms to “clean” the radar data on computed streamflow will be presented. These algorithms include: clutter removal and anomalous propagation, vertical profile of reflectivity (VPR) correction, extension of VPR correction out to 200 km, and using multiple levels of data to derive an optimal surface rainfall.
Supplementary URL: http://www.civil.uwaterloo.ca/watflood/studies/nowcasting.htm