Development of a Gridded Surface Weather Dataset forWildfire Analysis and Planning

Wildfire planning and climate analysis would benefit from a spatially and temporally continuous dataset of surface weather information that covers a minimum of 30 years. Currently, the United States wildfire community has been reliant upon surface weather observations collected from the Remote Automated Weather Station (RAWS) network. This network has evolved from a few hundred stations located primarily in the western US to nearly 2000 stations across the entire US today. The spatial and temporal inconsistency of RAWS data throughout its period of record makes climate analyses of this dataset across the US challenging.

The National Center for Environmental Prediction's North American Regional Reanalysis (NARR) dataset is a 32-km spatial grid that provides surface and other atmospheric, oceanic, and land data since 1979. While this dataset would suggest an ideal alternative to the RAWS for the purposes of wildfire planning and climate analysis, it does not incorporate data from RAWS network in its reanalysis. Statistical analysis has indicated that while NARR correlates well with RAWS data for some atmospheric variables (e.g., temperature), other variables have shown weak correlations, strong biases or both.

This paper presents a methodology for developing an adjusted gridded dataset that incorporates the 32-km gridded NARR data with RAWS. Methods were employed for adjusting continuous atmospheric parameters such as temperature and relative humidity and the more localized variables precipitation and wind. The final adjusted, gridded dataset indicates much stronger correlations and lower biases with the original RAWS data and can be developed and made available on a real-time monthly basis. The ultimate goal is to develop a gridded dataset consistent with RAWS that can be used for wildland fire analysis and planning.