Spatio-temporal analysis of drought clusters: statistical characterization and physical mechanisms of propagation

Currently most drought studies focus on time series analysis of area-averaged quantities, or spatial analysis of droughts of a given duration. However, maps of precipitation and soil moisture anomalies from model simulations show that droughts organize into clusters that evolve in time and can expand and contract, merge with other clusters to form larger ones, and move in space, sometimes traversing continents. Previous studies have suggested that the drought signal propagates downwind through a combination of land-atmospheric feedbacks, such as precipitation recycling, and moisture advection. Thus, to fully capture the development and recovery of droughts we need a simultaneous understanding of their temporal and spatial evolution. In this work, drought clusters are identified globally in soil moisture data from the Climate Forecast System Reanalysis (CFSR) at a monthly time scale from 1979-2009, and their evolution is tracked through time and space. The cluster tracks are used to find "hotspots" throughout the world where higher densities of clusters pass through and to identify regions where clusters tend to move in the same direction (i.e. displacement patterns). Both of these were found to vary seasonally, with hotspots generally matching the world's semi-arid regions. The distributions of the clusters' properties, including their duration, severity, areal extent, and distance traveled by the centroids are also computed and compared across sub-regions. Furthermore, in order to understand the physical mechanisms behind the spatial displacement of these drought clusters, a few case studies in North America are considered. The propagation of drought is tracked in space and time through the hydrologic cycle starting in the atmosphere as moisture convergence anomalies, affecting precipitation, followed by soil moisture, and then evapotranspiration, using measures of moisture recycling. Lag correlations between these components are calculated along the clusters' tracks, showing the degree to which information over a region under drought can help predict its subsequent growth or displacement into another region.