Validation of fire danger from the North American Regional Reanalysis dataset

The North American Regional Reanalysis (NARR) is an assimilated dataset at a 32-km spatial and 3-hour temporal resolution. Based upon multiple data sets including rawindondes, aircraft and surface weather stations, NARR assimilates these data into modeled output. However, one of the datasets that NARR does not include is the Remote Automated Weather Station (RAWS) network. This is a network of over 2000 currently active weather stations throughout the US operated by federal and state agencies for the primary use of wildfire applications. Therefore, many of these stations are placed in high-elevation and remote locations. RAWS is the standard data source for computing fire danger in the US. Validation of weather and fire danger from NARR compared to RAWS was analyzed for the 26-year period 1980-2005 using correlation, bias and other statistical relationships. NARR surface data was used as input to the National Fire Danger Rating System to derive daily historical Energy Release Component (ERC) values. The results are informative regarding the discrepancies and similarities that occur when mid- and high-elevation, remote location weather data are not integrated fully with an assimilated dataset, as well as assessing gridded model derived historical fire danger data.