Performance evaluation of regional weather predictions with the Model for Prediction Across Scales – Atmosphere (MPAS-A) over the Maritime Continent

This study presents the first configuration and evaluation of a new convection-permitting regional weather prediction system “SINGV_NG(MPAS)” for the Maritime Continent based on the Model for Prediction Across Scales – Atmosphere (MPAS-A). We assess its performance across various combinations of physical parameterization options, including upper-air and near-surface forecasts, with a primary focus on precipitation as accurately predicting it remains a primary objective during the initial SINGV_NG(MPAS) evaluation. In general, SINGV_NG(MPAS) outperforms the global European Centre for Medium-Range Weather Forecasts (ECMWF) forecasts for precipitation, particularly for intense rainfall events. However, the upper-air and near-surface forecasts in this region still reflect the biases of the ECMWF model, which typically shows dry biases aloft and wet, cold biases at the surface. A systematic analysis of forecast sensitivity to cloud microphysics, subgrid-scale convection, and planetary boundary layer processes reveals that neither of the two default physics parameterization suites defined within MPAS-A performs adequately in the Southeast Asia region studied. Our results show that although the default convection-permitting suite supports the development of tropical precipitation systems more effectively than the mesoscale-reference suite, its performance is hindered by its use of the Thompson scheme. Our findings show that integrating either the Weather Research and Forecasting Single-Moment 6-Class (WSM6) or National Severe Storms Laboratory (NSSL) microphysics scheme with the remaining components of the convection-permitting suite leads to a significant improvement in precipitation forecast skill. Although at a 3-km resolution, the choice of scale-aware convection schemes is critical, as it can shift the balance between light and heavy rainfall. Since many regional models used in Southeast Asia were initially developed and fine-tuned for midlatitude conditions, this study provides valuable insights to the local research community working on the development of next-generation forecasting systems.