Climatic and landscape changes as drivers of environmental feedback that influence rainfall frequency in the United States

Previous studies have identified regions where the occurrence of rainfall significantly increases or decreases the probability for subsequent rainfall over periods that range from a few days to several weeks. These observable phenomena are termed "rainfall feedback" (RF). To better understand the land-atmosphere interactions involved in RF, the behavior of RF patterns was analyzed using data from 1849 to 2016 at similar to 3000 sites in the contiguous United States. We also considered changes in major land-use types and applied a geographically weighted regression model technique for analyzing the predictors of RF. This approach identified non-linear and spatially non-stationary relationships between RF, climate, land use, and land type. RF patterns in certain regions of the United States are predictable by modeling variables associated with climate, season, and land use. The model outputs also demonstrate the extent to which the effect of precipitation, temperature, and land use on RF depend on season and location. Specifically, major changes were observed for land use associated with agriculture in the western United States, which had negative and positive influences on RF in summer and winter, respectively. In contrast, developed land in the eastern United States correlated with positive RF values in summer but with negative ones in winter. We discuss how changes in climate and land use would be expected to affect land-atmosphere interactions, as well as the possible role that physical mechanisms and rain-enhanced bioaerosol emissions may play in the spatiotemporal changes observed for historical patterns of rainfall frequency in the United States.