Multi‐Grid Nesting Ability to Represent Convections Across the Gray Zone

Abstract This study investigated the multi‐grid nesting ability of a limited area model to effectively represent convections across the gray zone, the resolution around 1–10 km where both cumulus parameterization and explicit convection are problematic. It evaluated the sensitivity of Meiyu rainfall forecasts in Jiangsu, China to model configurations of grid nesting and convection treatment. These configurations consisted of grid spacings from 30, 15, 9, 5, 3 to 1 km, single or double or triple nested grids, and the traditional Kain‐Fritsch (KF) or scale‐aware Grell‐Freitas cumulus parameterization or the explicit convection in the outer domain O. In single nesting O, coarse grids (>3–5 km) required parameterization to represent organized cumuli, while explicitly resolving convections in finer grids were necessary to improve forecasts. In double nesting O using cumulus parameterization at 30–9 km with the inner domain I using explicit convection at 1 km, the nesting ratio could be as large as 30 without significantly impacting I forecasts. This suggests a pragmatic approach to avoid the challenge in representing convections across the gray zone. Using Grell‐Freitas may improve mean O rainfall distributions, but this was not true for I forecasts due to counter errors in space and time, which were larger than using KF and at coarser grids. Triple nesting with a middle 3‐ or 5‐km grid was unnecessary and could even degrade I forecasts. Nesting O using KF to parameterize cumuli at 15 km with I explicitly resolving convections at 1 km achieved the best overall rainfall forecast in Jiangsu.