1225 Identification of Forecast Biases to Improve Fire Danger Forecasts in Colorado

Wednesday, 15 January 2020
Hall B (Boston Convention and Exhibition Center)
Brandon K. Cohen, Univ. of Louisiana, Monroe, Monroe, LA; and P. T. Schlatter and L. Kriederman

Every day meteorologists at the National Weather Service (NWS) produce routine weather forecasts, and a subset of these forecasts include parameters that are then utilized to calculate the National Fire Danger Rating System (NFDRS) indices. Accurate fire danger forecasts are necessary for fire management agencies and officials to effectively make daily decisions. This project examines NWS Boulder fire danger forecasts from May 2018 through May 2019, all of which were input into the Weather Information Management System (WIMS) for the calculation of NFDRS indices. Specifically this study looks at biases and trends in dry bulb temperature, dew point temperature, relative humidity, and wind speed forecasts versus observations from 23 Remote Automatic Weather Stations (RAWS) across the NWS Boulder forecast area of responsibility. Numerical differences were calculated between forecast and observation values from each station over the time period. These differences for each parameter were used to calculate statistics including mean error, mean absolute error, and standard deviation to perform statistical analysis. On an annual basis, more instances of forecasting values greater than the observed values are noted, though we find that seasonal variability in forecast bias exists across a number of the stations. Throughout the study, some stations reveal inherent forecast biases that persist across the entire year. We suspect some of these biases and trends are perhaps a result of localized topographic features at the RAWS site, which could be better accounted for. The results of this study will be used to improve NWS Boulder fire danger forecasts through bias correction techniques and permit improved fire operations for their partners.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner