Assessing the impact of climate and extreme weather on society and the environment requires fine scale observational weather information with consistent coverage over the region of interest. Many such high-resolution gridded meteorological products exist (PRISM, gridMET, Daymet, Livneh, ERA5-land, etc.). These datasets use different statistical techniques to interpolate weather station data to a high-resolution grid, they sometimes merge weather station data with remote sensing and/or reanalysis products, or they are directly reanalysis products. The selection of a high-resolution gridded observational meteorological dataset is often driven by institutional or disciplinary preferences, ease of access, or because of the availability of specific variables of interest. While some evaluations and comparisons of these datasets have been conducted, they generally focus on annual or seasonal averages while neglecting many aspects of climate and weather extremes that can drive large socioeconomic impacts.
Here, we present a comprehensive evaluation and comparison of high-resolution gridded observational meteorological datasets for use in multi-sector impact analyses. First, we conduct a survey of variables and metrics of interest among the Multi-Sector Dynamics (MSD) community of practice. Then we examine how these variables and indices are represented in the different observational datasets evaluated, with a specific focus on the tails of the distributions through extreme value analysis. We further provide objective metrics that assess the skills of these observational datasets over different regions of the United States to better inform decisions on how to select these datasets for multi-sector impact analyses.