为改善东南亚低纬高原区（LLHSA）降水模拟的性能， 提高降水预报准确性， 本文采用大气环流模式WRF（4.2版）和海洋分量模式NEMO（3.4版）， 用耦合器OASIS3-MCT进行桥接， 得到区域海气耦合模式WRF-OASIS-NEMO（WON）。大气和海洋分量模式都配置成相同的Arakawa-C网格， 水平空间分辨率设为0.25°， 耦合频次设置为逐小时， 便于模拟海洋和大气环流相耦合的中尺度运动特征。为评估WON模式的模拟性能， 选取2020年8月16 -18日的强降水过程为例， 与单独WRF模式的模拟效果进行比较分析。WON和WRF模式模拟的降水大值区位于高原东北部和中西部地区， 平均日降水量约为20 mm·d^-1， 与观测事实基本相符。WON模式改善了WRF模式在高原南部降水偏多而在高原西北部降水偏少的模拟偏差。WON模式改善了降水动力条件的模拟效果， 在高原中南部气旋式环流增强， 在高原西侧反气旋式环流增强， 进而改善了WRF模式在高原南部周围降水偏少， 高原西北部降水偏多的模拟偏差。WRF和WON模式均能再现垂直螺旋度的发展特征， 即在对流层中低层为正垂直螺旋度发展， 而在对流层高层为负垂直螺旋度发展。两个模式在雨带西部400 hPa高度层附近垂直螺旋度模拟偏强， 而在600~700 hPa高度层上垂直螺旋度模拟偏弱。WON模式相对于WRF模式的改进区域主要集中在雨带中西部地区。本次强降水的水汽来源包括孟湾的西南水汽输送和中国南海的偏南水汽输送。WRF模式和WON模式均能较好地重现相关水汽通量特征。WRF模式在孟湾北部水汽辐合偏强， 而在中国南海水汽向北输送偏弱。WON模式主要改善了WRF模式在中国南海水汽输送偏弱的模拟偏差。WON模式改善降水模拟效果的主要原因是孟湾海表热通量交换导致孟湾中低层大气偏冷偏干， 大气对流活动减弱， 在孟湾北部形成的低层反气旋偏差改善了本次强降水过程动力条件和水汽条件的模拟效果。

To improve the simulation of precipitation and the accuracy of rainfall forecast over the low-latitude highland in Southeast Asia （LLHSA）， a regional air-sea coupled model is developed with the Weather Research and Forecast （WRF） Model （version 4.2） and ocean general circulation model NEMO （version 3.4）， using the coupler OASIS3-MCT.This new regional air-sea coupled model WRF-OASIS-NEMO is herein referred as WON.Both the atmospheric and oceanic components were configured into the same Arakawa-C grid with a horizontal spatial resolution of 0.25° and a coupling frequency of 1 hour， which are suitable for facilitating the mesoscale coupling between the atmosphere and ocean models.The evaluation of the WON model is based on the heavy precipitation event from August 16 to 18， 2020， where the simulation of WON model is compared with the standalone WRF model.The WON and WRF models simulated large precipitation over the northeastern LLHSA， the central and western LLHSA with daily precipitation around 20 mm·d^-1， which is generally consistent with the observation.The WON model ameliorated the underestimation bias of precipitation over the southern LLHSA and the overestimation bias of precipitation over the northwestern LLHSA and the western LLHSA in the WRF model.The WON model improved the simulation of the dynamic conditions of precipitation， with enhanced cyclonic circulation over the central and southern LLHSA and enhanced anticyclonic circulation over the western LLHSA.Hence， the WON model ameliorated underestimation of precipitation over the southern LLHSA and overestimation of precipitation over the northwestern LLHSA.Both WRF and WON models could reproduce the development characteristics of vertical helicity， that is， positive vertical helicity in the lower-mid troposphere and negative vertical helicity in the upper troposphere.However， the simulated vertical helicity is too strong near 400 hPa layer over the western rain belt， but too weak at 600~700 hPa layer.Compared with the WRF model， the WON model shows improvements mainly in the central-western part of the rain belt.The water vapor sources of this heavy rainfall include the water vapor transport from the southwest of the Bay of Bengal and the water vapor transport from the South China Sea.Both WRF model and WON model reproduced the spatial characteristics of water vapor flux.In the WRF model， the water vapor convergence is too strong over the northern Bay of Bengal， but too weak over South China Sea.The improvement of WON model is mainly over South China Sea.The simulation improvement of the WON model is mainly because the surface heat flux exchange over the Bay of Bengal caused the mid-lower levels of troposphere to become cooler and drier.The atmospheric convection was weakened， associated with a low-level anticyclonic bias over the northern Bay of Bengal.This anticyclonic bias improved the simulation of atmospheric dynamics and water vapor conditions for this heavy precipitation event.