Evaluating the ability of gridded climate datasets to capture temperature and precipitation trends and extremes

Data informs policy and drives decisions. With an increasing number of regulations requiring entities to use local climate data in their planning, it's more important than ever to understand the strengths and limitations of data we use. While they have been shown to capture long-term statistics on global or regional levels, the ability of gridded climate datasets to capture trends and extreme events is not common knowledge. Four widely used gridded datasets, ERA5, ERA5-Land, MERRA-2, and PRISM, were assessed for their ability to capture extreme heat, extreme cold, and heavy precipitation events, as well as trends in annual maximum and minimum temperatures and total precipitation, over the contiguous US (CONUS). Spatial patterns are evident in each dataset, with the largest differences between observations and the gridded data across the western US for temperature and along the Gulf Coast for heavy precipitation. In general, gridded datasets captured extreme heat better than extreme cold or heavy precipitation, and trends in annual maximum temperature better than trends in annual minimum temperatures and annual total precipitation. All dataset capture extreme heat days comparably, but PRISM generally performed best for extreme cold and the bias-adjusted MERRA-2 dataset generally performed best for heavy precipitation days.