An assessment of land-atmosphere interactions over South America using satellites, reanalysis and two global climate models

In South America, land–atmosphere interactions have an important impact on climate, particularly the regional hydrological cycle, but detailed evaluation of these processes in global climate models has been limited. Focusing on the satellite-era period of 2003–14, we assess land–atmosphere interactions on annual to seasonal time scales over South America in satellite products, a novel reanalysis (ERA5-Land), and two global climate models: the Brazilian Global Atmospheric Model version 1.2 (BAM-1.2) and the U.K. Hadley Centre Global Environment Model version 3 (HadGEM3). We identify key features of South American land–atmosphere interactions represented in satellite and model datasets, including seasonal variation in coupling strength, large-scale spatial variation in the sensitivity of evapotranspi-ration to surface moisture, and a dipole in evaporative regime across the continent. Differences between products are also identified, with ERA5-Land, HadGEM3, and BAM-1.2 showing opposite interactions to satellites over parts of the Amazon and the Cerrado and stronger land–atmosphere coupling along the North Atlantic coast. Where models and satellites disagree on the strength and direction of land–atmosphere interactions, precipitation biases and misrepresentation of processes controlling surface soil moisture are implicated as likely drivers. These results show where improvement of model processes could reduce uncertainty in the modeled climate response to land-use change, and highlight where model biases could unrealistically amplify drying or wetting trends in future climate projections. Finally, HadGEM3 and BAM-1.2 are consistent with the median response of an ensemble of nine CMIP6 models, showing they are broadly representative of the latest generation of climate models.