Downward Convective Moisture Transport Dominated by a Few Overshooting Clouds in Marine and Continental Shallow Convection

In a previous study (Xiao et al., 2023, https://doi.org/10.1029/2022ms003526), we found that ignoring the moist convective downdrafts associated with overshooting clouds in parameterizations can lead to significant biases in the simulated depth and liquid water content of a shallow cloud layer. In this study, we seek to better quantify the properties of the clouds responsible for these moist downdrafts to help improve shallow convection parameterizations. We apply a 3-D cloud-tracking algorithm to large-eddy simulations (LESs) of marine and continental shallow convection. We find that top 1% and 2% of the tracked cloud population ranked by lifetime-mean cloud-base mass flux can explain 90%–95% of the total downward moisture transport in the upper cloud layer whereas top 10%–20% is required to explain 90%–95% of the total upward moisture transport near mean cloud base. The vertical structure of the clouds in the top 1% and 2% (the overshooting “deep mode”) is also distinctively different from that of the rest of the cloud population (the “shallow mode”). Shallow convection parameterizations need to capture accurately the properties and convective transports of the clouds in both the deep and shallow modes. To do that, our results suggest that mass-flux parameterizations need to (a) accurately predict the size and number of the deep-mode clouds and (b) explicitly represent overshooting cloud updrafts and associated moist downdrafts.