November 18, 2026
Journal Article
Observed and Simulated Characteristics of Central U.S. Mesoscale Convective Systems and Their Dependence on Monthly Environmental Conditions
Abstract
Mesoscale convective systems (MCSs) are key weather systems responsible for severe flooding and damaging winds across the central U.S., yet their environmental controls (relationships between environmental predictors and MCS characteristics) remain incompletely understood. In this study, we first track MCSs in the CONUS404 dataset (4 km, over 40-year WRF simulation) using the PyFLEXTRKR algorithm to evaluate the MCS climatology and then use the Random Forest (RF) algorithm to quantify the environmental controls of MCSs at the monthly time scale, cross-validated by a high-resolution observational MCS database. The results show that CONUS404 reproduces the observed seasonal migration of MCSs but underestimates their peak summer activity over the central U.S., in line with a dry precipitation bias. Environmental predictors explain up to 50\% of the total variance in monthly MCS number and precipitation, with moisture convergence, atmospheric instability, and column water vapor emerging as the top three controlling predictors. Partial dependence analyses of paired environmental predictors reveal their nonlinear interactions in modulating MCS activity. Feature importance analyses show a clear spring–summer contrast, with spring MCSs driven mainly by the dominant predictors noted above while summer MCSs are more sensitive to vertical wind shear and relative humidity. Condition-swapping experiments using the RF representing the observed environmental control show that model biases in the simulated wind shear impact MCS number, whereas biases in moisture convergence, atmospheric instability, and low level wind shear predominantly influence MCS precipitation. These results clarify key environmental drivers of central U.S. MCSs and help anticipate future changes in MCS activity.Published: November 18, 2026