Has agricultural irrigation masked intense warming in the central United States?

Since the 1980s, the central United States and southern-central Canada have experienced a notable lack of high temperature extremes, with many temperature record highs from the 1930s Dust Bowl period still standing. By contrast, atmospheric general circulation models (AGCMs) forced with observed sea surface temperatures consistently simulate exceptional warming over the central US during this period. What accounts for this discrepancy between observed and simulated temperature trends? We use ensembles of coupled and atmosphere-only climate model experiments to disentangle the influences of remote sea surface temperatures and local land-atmosphere interactions on historical temperature change in the central United States. Tropical Pacific teleconnections strongly impact central US temperatures: coupled general circulation models, which cannot reproduce observed trends in the tropical Pacific SST gradient, produce a moderate central US warming trend that is closer to observations than AGCMs prescribed with observed SSTs. Comparing seasonal latent and sensible heat fluxes in these experiments, we describe the central role of turbulent exchanges at the land surface on temperature trends. In a heavily irrigated area whose climate is known to be sensitive to changes in soil moisture, our results point to a possible role for agricultural irrigation in alleviating historical heat extremes, and in explaining the large difference between models and observations. We highlight the importance of understanding model-data discrepancies in tropical SST patterns and local land temperatures for predicting future climate extremes in the central US.