Analyzing the potential impacts of soil moisture on the observed and model-simulated Australian surface temperature variations

Based on observational and modeling analyses, this study aims to assess the potential influence of land surface conditions (soil moisture, in particular) on the Australian surface temperature variations. At first, a simple linear regression method is used to largely remove the ENSO influence from 50-yr observational surface temperature and precipitation datasets. Then, lag and partial correlations of the residuals are analyzed. The impacts of precipitation on the forthcoming surface temperature variations are largely attributed to the soil storage of precipitation water and the slow-varying soil moisture process. Results from partial correlations between precipitation and temperature variations suggest that when responding to anomalous atmospheric forcing, the land surface can introduce some slow-varying processes that can in turn affect the mean state of the atmosphere at monthly or longer scales and increase the predictability of the climate system. Following the observational analysis, results from 16 Atmospheric Model Intercomparison Project Phase 2 (AMIP2) AGCM simulations are analyzed to assess whether land surface modeling can affect the model-simulated climate variability. Lag-correlation analysis reveals that climatic memory'' of soil moisture has different features in the 16 models. Models with simple bucket-type schemes tend to have a rapid decay rate in the retention of soil moisture anomalies and show rapid feedback between land surface and the overlying atmosphere, with a much weaker influence of soil moisture conditions on surface climate variations. In contrast, most models using nonbucket schemes in which more physical processes are introduced in simulating soil water evaporation and soil water movement tend to show slow-varying soil moisture processes, affecting the model integrations at longer time scales. Different characteristics for translating soil moisture memory into climate variability and predictability are seen across the models, and more detailed studies are needed to further explore how land surface processes affect climate variability and predictability.