Using Machine Learning Techniques to Construct a Climatology of Mesoscale Convective Systems in the United States

This research uses image processing and machine learning techniques to identify, classify, and track mesoscale convective systems (MCSs) in composite radar reflectivity mosaic images. Nearly one million of these images, which cover the contiguous United States (CONUS) and span an unprecedented study period of 22 years (1996-2017), are used to generate a novel assessment of the seasonal and inter-annual variability of MCSs. Additionally, hourly precipitation data from 16 of those years (2002-2017) are used to systematically examine rainfall associated with radar-derived MCS events. The attributes and occurrence of MCSs that pass over the CONUS domain, as well as five sub-regions—North Plains, High Plains, Corn Belt, Northeast, and Mid South—are also examined. The results illustrate two preferred regions for MCS activity in the CONUS: 1) the Mid South and Gulf Coast, and 2) the Central Plains and Midwest. MCS occurrence and MCS rainfall displays a marked seasonal cycle, with most of the regions experiencing these events primarily during the warm season (May – August). Additionally, MCS rainfall was responsible for over 50% of annual and seasonal rainfall for many locations in the CONUS. Of particular importance, the majority of warm-season rainfall for regions with high agricultural land cover (Corn Belt) and important aquifer recharge properties (High Plains) is attributable to MCSs. These results confirm that MCSs are a significant aspect of the CONUS hydroclimate, and understanding how these events may change between now and the late 21^st century should be an important research priority.