Errors in rainfall estimation in the southern Alaska WSR-88D network

Very little research has been done regarding the use of weather radars in Alaska, not only to localize them for the unique set of environmental phenomena found in Alaska, but also to evaluate their feasibility as reliable data sources in observation-deficient areas. This study isolated specific areas where the Alaskan radar network has performed well, where its data quality is suspect, and where improvement of the system is desired. Several interviews were conducted with forecasters at all three Alaskan National Weather Service forecast offices, as well as the Alaska River Forecast Center and Alaska Region Headquarters, to learn about perspectives on the radar network. The area chosen for further investigation was the precipitation amount estimates given by various radars. Specific excessive rainfall and flooding cases recommended by local forecasters were evaluated, using two of Alaska's Doppler radars. One-hour, radar-derived precipitation estimates were compared to observed hourly rain gauge data, and WRF model simulations using reanalysis data were also used to “estimate” precipitation from these events. It can be shown that, for river forecasting and flash flooding purposes, radar-derived precipitation estimates very often, though not always, substantially improve upon model forecast estimates. However, even with this improvement, it was found that radar-derived precipitation estimates were rarely accurate within their scale of resolution as compared to rain gauge data, and the error was not consistent from site to site or from event to event. Possible reasons for these errors, particularly with regards to the Alaskan environment, are discussed, as are questions that would require future research.