H12A-02:
Observational Evidence for Coupling Between Soil Moisture and Temperature Extremes Across Climate Regimes from the U.S. Soil Climate Analysis Network (SCAN)

Monday, 15 December 2014: 10:40 AM
Justin Sheffield, Princeton Univ, Princeton, NJ, United States and Ruolan Xu, Princeton University, Princeton, NJ, United States
Abstract:
Soil moisture is known to play a key role in controlling surface moisture and heat fluxes, but much of the evidence for this at large scales is based on analysis of data from reanalyses, satellite remote sensing and model data. As such, the relationships are subject to the errors and biases in these data sources. This study evaluates direct observational evidence for land-atmosphere interactions using data from the Soil Climate Analysis Network (SCAN) which hosts monitoring sites for over 100 locations across the U.S. covering a range of climates and land surface contexts. The relationship between warm season soil moisture deficits and temperature extremes is analyzed for 80 sites that have sufficient data. Statistically significant negative correlations are found in many sites between the number of hot days and soil moisture, with the strength of the relationship increasing from the humid east to the drier Midwest. The signal in the inter-mountain west is somewhat mixed. To distinguish the sources of temperature extremes between local coupling and advection, we estimate advection at each site from meteorological data from the North American Land Data Assimilation (NLDAS2). Although the advection estimates are uncertain, they imply that local coupling can play a significant role in temperature extremes during dry periods. We further characterize individual heat waves according to the estimated contribution of advected heat or local coupling to their emergence and persistence. Heat wave events are generally terminated by precipitation events. The results have implications for understanding land-atmosphere coupling across climate regimes and the utility of model-derived estimates to represent observed behavior at large scales.