The Long-Term Trends of Global Land Precipitation in GFDL's CM4 and ESM4 Climate Models

Historical precipitation and temperature trends and variations over global land regions are compared with sim-ulations of two climate models focusing on grid points with substantial observational coverage from the early twentieth cen-tury. Potential mechanisms for the differences between modeled and observed trends are investigated using subsets of historical forcings, including ones using only anthropogenic greenhouse gases or aerosols, and simulations forced with the ob-served sea surface temperature and sea ice distribution. For century-scale (1915-2014) precipitation trends, underestimated in-creasing or unrealistic decreasing trends are found in the models over the extratropical Northern Hemisphere. The temporal evolution of key discrepancies between the observations and simulations indicates that 1) for averages over 15 & DEG;-45 & DEG;N, while there is not a significant trend in observations, both models simulate reduced precipitation from 1940 to 2014, and 2) for 45 & DEG;-80 & DEG;N observations suggest sizable precipitation increases while models do not show a significant increase, particularly during ;1950-80. The timing of differences between models and observations suggests a key role for aerosols in these dry trend biases over the extratropical Northern Hemisphere. Additionally, 3) for 15 & DEG;S-15 & DEG;N the observed multidecadal decrease over tropical west Africa (1950-80) is only roughly captured by simulations forced with observed sea surface temperature; additionally, 4) in the all-forcing runs, the model with higher global climate sensitivity simulates increasing trends of temperature and precipitation over lands north of 45 & DEG;N that are significantly stronger than the lower-sensitivity model and more consistent with the observed increases. Thus, underestimated greenhouse gas-induced warming}particularly in the lower sensitivity model}may be another important factor, besides aerosols, contributing to the modeled biases in precipitation trends.

Automated summary

This study compares historical trends in precipitation and temperature over global land regions with simulations from two climate models. The models underestimate precipitation trends in the extratropical Northern Hemisphere and the differences between models and observations may be due to the effects of aerosols and greenhouse gases. Additionally, the higher sensitivity model simulates stronger temperature and precipitation increases north of 45°N, suggesting that underestimated greenhouse gas-induced warming may also contribute to the modeled biases in precipitation trends.