Optimal Design of Multi-scale Ensemble Systems for Convective-Scale Probabilistic Forecasting

从气象和公共服务/社会影响的角度来看，对流尺度灾害性天气的预测非常重要。 对流尺度预报的独特挑战是，预报的准确性不仅取决于对流尺度的过程，而且还取决于支持它们的中尺度和天气尺度环境。 因此，对流尺度的可靠和准确的概率预报需要从多个尺度的误差进行适当的采样。 本研究的主要目标是确定在这种多尺度情景下集合预报系统的最佳设计，以进行对流尺度概率预报。 这个问题以前没有得到解决，现在已经成为一个紧迫的问题，因为随着计算技术的进步，对流尺度集合预报不仅是可取的，而且是完全可能的。这项研究将建立在风暴分析和预报中心（CAPS）建立的基础和初步能力上，该组织已经运行了一个有20个成员的4公里对流允许分辨率集合预报系统，加1-公里的预测，可以被认为是一个额外的成员在春季自2007年以来在美国大陆的七个相互关联的问题将进行调查，以实现研究目标. 1)在多尺度情景下，对流尺度集合的最佳初始扰动条件是什么？2)如果使用嵌套网格方法来捕获多尺度，那么外部和内部区域扰动如何通过横向边界条件（LBC）相互作用，并且从根本上讲，各种尺度扰动如何在域内和跨域内相互作用？3)耦合陆面模式的最佳扰动策略是什么？4)多模式/物理和随机物理方法在对流尺度模式误差抽样中的有效性如何？5)在初始条件、侧边界条件、大气模型（包括模型物理和动力学）和陆面模型中，取样不确定性的相对重要性和贡献是什么，它们的最佳组合是什么？6)什么是集合大小和模型分辨率之间的最佳折衷？7)对流尺度集合的有效概率预报检验和评估指标是什么？智力优点：该项目将回答多尺度情景下对流尺度概率预报集合系统优化设计的许多基本科学问题。 关于大尺度和对流尺度集合扰动如何相互作用以及相互作用如何影响对流尺度集合设计的新知识，关于在对流尺度集合预报中解释模型误差的有效方法的新知识，关于陆面和大气集合扰动相互作用的新知识，从这项研究中，将学到关于不同误差来源对对流尺度概率预报的相对重要性和影响的新知识，以及关于对流尺度概率预报最适当的客观验证方法的新知识。更广泛的影响：该项目的科学成果将为设计和加快国家努力开发和实施下一代业务中尺度集合预报系统提供指导。 它将直接处理国家在天气方面的两个主要优先事项：高影响天气预报警告和下一代预报系统，作为下一代航空运输系统的关键组成部分（NextGen，http：//www.faa.gov/about/initiatives/nextgen/）。 它还将解决天气研究最重要的目标之一-提高我们准确预测强烈危险天气的能力，这些天气对美国经济和公民的生活产生负面影响，每年造成大量金钱损失和许多生命损失。 该项目将为研究生提供对流尺度概率预报重要领域急需的教育和培训。 研究结果还将通过该小组在NOAA危险天气试验台（HWT）春季预报实验中的重要作用及其与位于科罗拉多博尔德的国家大气研究中心的发展试验台中心（DTC）的互动和合作直接进入运营。

The prediction of convective-scale hazardous weather is very important from both meteorological and public service/societal impact perspectives. The unique challenge in convective scale forecast is that the accuracy of the forecasts depends not only on processes at the convective scale but also on the mesoscale and synoptic-scale environment supporting them. Therefore, reliable and sharp probabilistic forecasts for convective scales require proper sampling of errors from multiple scales. The main goal of this research is to determine the optimal design of ensemble forecast systems under such multi-scale scenarios, for the purpose of convective-scale probabilistic forecasting. This issue has not been addressed previously and has become a pressing issue as convective-scale ensemble forecasting is not only desirable but also entirely possible with the advancement of computational technologies.The research will build on the foundation and initial capabilities established at the Center for Analysis and Prediction of Storms (CAPS), which has run a 4-km convection-allowing resolution ensemble forecasting system in real-time with 20 members, plus one 1-km forecast that can be considered an additional member during the springs since 2007 over the Continental U.S. Seven interlinked questions will be investigated in order to achieve the research goal. 1) What are the optimal initial condition perturbations for convective-scale ensemble in the multi-scale scenario? 2) If a nested-grid approach is used to capture multiple-scales, how do the outer and inner domain perturbations interact through the lateral boundary condition (LBC) and fundamentally how do the various scale perturbations interact within and across the domain? 3) What is the optimal perturbation strategy for the coupled land surface model? 4) What is the effectiveness of the multi-model/physics and stochastic physics methods in sampling model error for convective scales? 5) What is the relative importance and contributions of sampling uncertainties in the initial conditions, lateral boundary conditions, atmospheric models including model physics and dynamics, and the land surface models, and what is their optimal combination? 6) What is the best tradeoff between ensemble size and model resolution? 7) What are the effective probabilistic forecast verification and evaluation metrics for convective-scale ensemble?Intellectual merit: The project will answer many of the fundamental scientific questions concerning optimal design of ensemble system for convective scale probabilistic forecasting under the multi-scale scenario. New knowledge on how the large-scale and convective-scale ensemble perturbations interact with each other and how the interaction impacts the ensemble design for convective scales, new knowledge on the effective methods to account for model error in convective scale ensemble forecasting, new knowledge on the interaction of land surface and atmospheric ensemble perturbations, new knowledge on the relative importance and impact of different sources of errors on convective-scale probabilistic forecasting; and new knowledge on the most appropriate objective verification method for convective scale probabilistic forecasting will be learned from this study. Broader impact: The scientific findings of this project will provide guidance for the design of, and accelerate the national efforts in developing and implementing the next-generation operational mesoscale ensemble forecast systems. It will directly address two key national priorities in weather: Warn on Forecast for High Impact Weather and the Next-Generation Forecast System, as key component of the Next-Generation Air Transportation System (NextGen, http://www.faa.gov/about/initiatives/nextgen/). It will also address one of the most important goals of weather research - to improve our ability to accurately predict intense hazardous weather that negatively impacts the American economy and the lives of its citizens, causing large monetary loss and the loss of many lives each year. This project will provide much needed education and training for graduate students in the important areas of convective-scale probabilistic forecasting. The research findings will also have a direct path to operations through the group's significant role in the NOAA Hazardous Weather Testbed (HWT) Spring Forecast Experiments and its interaction and collaboration with the Developmental Testbed Center (DTC) in National Center for Atmospheric Research in Boulder, Colorado.