U.S. Joint Fire Science Program


Date of this Version


Document Type



JFSP Project ID: 10-1-07-28


US government work.


This project investigates the differences between the gridded meteorological fields produced by the Real Time Mesoscale Analysis (RTMA) and observed meteorological conditions at Remote Automated Weather Stations (RAWS) for two years in the northeastern United States. National Weather Service (NWS) fire weather forecasts are produced using the National Digital Forecast Database (NDFD), which is a gridded analysis of meteorological fields generated by forecasters at NWS forecast offices nationwide. The NDFD is verified by comparing its gridded meteorological fields against the RTMA, which is an advanced modeling and data assimilation system that provides the best-available hourly gridded estimate of surface and near-surface meteorological conditions. However, for fire management activities, which critically depend on fire weather forecasts, RAWS observations are the standard observational data employed for the calculation of fire danger indices, fire behavior analyses, and for observation-based decision support. RAWS fire weather observations are collected from stations that have different siting requirements and standards of measurement than the majority of stations employed to produce the RTMA and to verify and improve the NDFD. Thus, it is necessary for fire weather forecasters and fire managers to understand how RAWS observations differ from the RTMA, and by association the NDFD, in order to make the best possible forecasts and fire management decisions with the available fire weather information. This statistical comparison between the RTMA and RAWS station data in the northeast United States over a period of two years (August 2008 through July 2010) documents differences between the two sources of meteorological information. Interpretation of the analysis is complicated by uncertainty over when RAWS observations were assimilated into the RTMA. Since the RAWS observations and the RTMA are not fully independent of each other, assessing the ability of the RTMA to capture the temporal and spatial variability of fire weather quantities and determining if systematic differences exist is not straightforward. Nevertheless, we conducted statistical analyses of the differences between RAWS and RTMA data at 237 RAWS stations in the northeast United States. We then performed a series of statistical tests and analyses to determine if systematic geographical, temporal, or topographical differences can be identified. None of these analyses were able to relate differences between RAWS and RTMA to these factors, leading to the conclusion that the major source of differences between the two datasets is local siting characteristics of the RAWS stations. These results indicate that statistical analyses of differences between RAWS stations and National Weather Service data (such as RTMA, NDFD) should be applied on a station-by-station basis and that applications of these results to the RAWS as continuous network are unlikely to yield any improvement in the quality and accuracy of the data for fire applications. The statistical differences between the two data sources are now available in both graphical and text forms at the 237 RAWS stations in northeast Unites States at http://geo.msu.edu/~raws/.