积云对流参数化方案的集合优化及其对我国南方夏季降水的模拟

Current cumulus parameterization schemes (CUPs) are mostly based on one single closure assumption, which leads to limitations on model predictions of summer precipitation over Southern China, including mean patterns and daily rainfall frequency and intensity distributions. This project will develop an optimized ensemble cumulus parameterization (ECP) scheme which can be well applied to predict the summer precipitation in Southern China using the regional weather-climate model (CWRF). The ECP scheme incorporates five types of widely-used cumulus closure assumptions with relative weights depending upon locations. On the basis of the selected summer cases when extreme precipitation anomalies occurred over Southern China, a series of sensitivity experiments will be conducted by CWRF using the ECP scheme with different closure assumptions. This project first will identify the effects of these closure assumptions on summer precipitation predictions over different regions in Southern China, and then compare the simulated cloud-base mass fluxes, convective precipitation, and tropospheric circulations at various levels, as well as the vertical distributions of atmospheric temperature and humidity. This will help to examine the mechanism differences from these closures in describing the interactions between cumulus convection and large-scale environment, which will partly explain the reasons for the model sensitivities of existing CUPs based on these closures. After identifying appropriate closures in predicting precipitation characteristics over different regions in Southern China, we can dynamically select optimal weights for these closures over each of these regions. The weights can be derived from the algorithm for solving constrained nonlinear optimization problems with respect to appropriate objective functions. The result will reduce the deficiencies of existing CUPs based on single closure assumption, and thus be of great significance in improving the model predictive skills of summer precipitation over Southern China.

现有的积云对流参数化方案多是基于单一的闭合假设，对我国南方不同区域夏季季节平均降水量和落区、日降水强度和频率的模拟存在局限。本项目基于区域气候模式,发展适合我国南方夏季降水模拟的集合优化的对流参数化方案。该方案集合现有的五类闭合假设，并且基于格点、权重可调节。选取江淮流域和华南地区的夏季极端降水异常个例，利用不同闭合假设进行敏感性实验，定量评估不同闭合假设对南方不同区域夏季降水模拟的影响；结合云底通量、对流降水、环流场及温湿廓线的分布，探讨不同闭合假设所描述的积云和环境场相互作用机制的差异，尝试解释基于这些闭合假设的已有参数化方案模拟差异的原因；针对我国南方不同区域选择合适的闭合假设，根据恰当的目标函数，利用约束非线性优化算法求解闭合假设的最优权重，从物理过程上实现动态集合，改进基于单一闭合假设的积云对流参数化方案的缺陷，对提高我国南方地区夏季降水预报的水平有着重要的意义。

我国南方夏季降水主要集中在长江流域和华南地区，年际异常变率大，极端事件多发。目前模式普遍存在雨带北移的系统性偏差，并且严重低估极端降水的强度和频率。已有的研究指出这些问题与现有的积云对流参数化方案的缺陷有关，但较少从方案中关键的闭合假设出发，深入探讨模拟差异的原因。而且，现有的积云对流参数化方案多是基于单一的闭合假设，对不同区域降水的模拟存在局限性。在本项目的资助下，主要研究成果有：1）基于垂直速度和水汽辐合假设的集合，构建了区域气候模式中集合对流参数化方案（ECP）的控制版本，与其他7种积云对流参数化方案对照，通过36年的长期积分实验系统比较不同方案对我国南方夏季降水量以及年际变率的模拟的影响，发现ECP能较好模拟出江淮和华南两条雨带及年际变率，但低估极端日降水强度，江淮降雨总日数和雨量偏低，华南降雨总日数和雨量偏高。2）针对年际变率最大的长江上游、中下游(江淮)和华南地区的8个夏季极端降水异常年和1个气候平均年，利用ECP方案中的5大类（准平衡AS，垂直速度W，水汽辐合MC，全部不稳定能量释放KF和不稳定能量倾向TD），共16种形式的闭合假设进行敏感性实验，从闭合假设的角度探讨了已有参数化方案模拟差异的可能原因，如Grell方案普遍低估江淮与华南的降水量与极端日降水强度，与大尺度不稳定倾向假设（TD）的模拟一致；KFeta方案模拟的雨带偏南偏强，但是低估降雨日数，严重高估江淮与华南极端日降水强度，与全部不稳定能量的释放（KF）假设的表现一致。3）利用8个极端降水年与1个降水平均年，选取日降水均方根误差最小为目标函数，以约束非线性优化算法（FFSQP），基于ECP中16个闭合假设的模拟结果，取任意的8个夏季的16组模拟实验，共736天数据构建了非线性的训练模型，得到基于格点的不同闭合假设模拟的优化权重，循环预测剩余的1个夏季92天，结果有效改善了现有方案对我国南方夏季极端异常年与平均年的降水量及落区的模拟。