Reduced Spring Precipitation Bias and Associated Physical Causes over South China in FGOALS-f3 Climate Models: Experiments with the Horizontal Resolutions

Abstract

Considerable spring precipitation occurs over South China (SC), a region that is adjacent to large-scale Asian topography and oceans. Its reasonable simulation is crucial for improving regional climate predictability. This study investigates spring precipitation biases over SC and their possible causes in atmosphere-only and coupled Flexible Global Ocean–Atmosphere–Land System finite-volume version 3 (FGOALS-f3) models with different horizontal resolutions. The performance of spring precipitation simulation over SC varies across different FGOALS-f3 model versions, with the best reproducibility in the high-resolution coupled model (25 km). In the low-resolution atmosphere-only model (100–125 km), the precipitation dry bias over SC is closely linked to overestimated surface sensible forcing over the eastern Tibetan Plateau (TP), which weakens the subtropical anticyclone over the western Pacific (SAWP) through regional circulation responses. By contrast, the high-resolution atmosphere-only model further amplifies surface thermal forcing in the Asian continents, causing intensified land-sea thermal contrast between the Southeast Asian continents and western Pacific, enhanced southerly winds and SAWP, and increased water vapor transport into SC. Meanwhile, the reduced middle-high level cold bias over 10°–30°N in the high-resolution atmosphere-only model intensifies the East Asian westerly jet and ascent over SC, leading to enhanced spring precipitation there. The high-resolution coupled model simulation not only reduces sea surface cold bias over the Bay of Bengal, thus intensifying the Indian-Burma trough and strengthening low-level water vapor transport into SC, but also enhances ascent over SC. As a result, the high-resolution coupled model better reproduces the magnitude and pattern of spring precipitation over SC than its atmosphere-only model. Compared with low-resolution models, the domain-mean spring precipitation dry bias decreases by 11.2% over SC in the high-resolution atmosphere-only model and by 35.9% in the coupled model. These results demonstrate that the high-resolution FGOALS-f3 models can improve simulations of the influencing atmospheric circulations and spring precipitation over SC.