Monthly High-Resolution Historical Climate Data for North America Since 1901

Abstract

Interpolated grids of historical climate variables are widely used in climate change impact and adaptation research. Here, we contribute monthly historical time series grids since 1901 for our data product ClimateNA, which integrates historical data and future projections to generate high-resolution gridded data and point estimates for North America. The historical climate grids in this study are based on interpolations of monthly anomalies (change factors) with thin-plate splines, but a novel aspect is that we rely on high-quality 1961–1990 normal estimates from ClimateNA to serve as reference for the change factor calculations instead of the reference being derived from station data itself. This allowed us to utilise records from 66,282 climate stations for interpolations, regardless of their temporal coverage. Another aspect that deviates from standard practice is that we reduce overfitting by optimising thin-plate splines at a 0.5° grid level instead of fitting weather station observations directly. The high-resolution grids generated with this approach compared favourably with other time series products, such as Daymet and advanced multi-source products, such as MSWEP, in statistical and mapped visual comparisons, and provide additional historical coverage since the beginning of the 20th century.