Abstract A high-resolution (~50 km) atmospheric model from GFDL showed promise for simulating observed climatological features of mesoscale convective systems (MCSs) despite non-negligible biases. This study provides a comprehensive evaluation of precipitation and lifecycle characteristics of simulated MCSs in the continental US. The MCSs are tracked using cloud-top brightness temperature as the sole proxy, enabling independent process-based attribution of MCS biases to their cold cloud systems and precipitation. The MCS precipitation is underestimated in the central US and overestimated in the eastern US, with biases being more pronounced in summer than in spring. The pattern of MCS precipitation bias differs from that of total precipitation bias, suggesting that its improvement may not necessarily eliminate the long-standing precipitation deficit in the central US. Most MCS precipitation is explicitly resolved through large-scale cloud processes. The model well reproduces observed distributions of various MCS diagnostics and their interrelations. However, it produces spurious systems producing very weak and spatially unorganized precipitation (referred to as “dry MCS problem”), which is particularly severe in summer. The linear relation between MCS size and precipitating area is also underestimated, even when the dry MCSs are excluded. While the model accurately captures observed parabolic-shaped MCS size evolution, it misrepresents the fractional precipitating area as a monotonic decrease throughout the lifecycle. The MCS precipitation is largely overestimated at initiation, followed by a sharp decrease. This reflects overly efficient precipitation in the convective towers during upscale growth phase of MCSs, which is hypothesized to be responsible for suppressed stratiform precipitation at maturity.
Park et al. (Tue,) studied this question.