Implementation and evaluation of the unified parameterization in NCAR Community Atmospheric Model

用于天气预测和气候模拟的大气全球环流模型(CGM)通常将大气划分为一个网格，在该网格上，温度和压力等量具有单个值，以表示每个网格单元的平均条件。由于计算费用的原因，GCM传统上使用100公里或更远的网格间距，在这个网格间距下，不可能表示单个云甚至大型云系。在这些模式中，产生强烈降水并具有其他重要影响的积云通过对流参数化间接表示，这些对流参数化试图捕捉它们在网格单元上的聚集行为。随着计算机能力的增强，以足够高的分辨率运行GCM已成为可能，以进入“灰色区域”，在该区域中，最大的积云集合的某些部分的运动被显式地表示在模式网格上，但较小的云仍必须通过对流参数化间接地表示。这项工作的目标是发展一种参数化，随着网格间距的减小和积云活动的更大部分得到显式解决，该参数化可以自动和无缝地调整。在较粗的分辨率下，参数化表示云大小和运动的整个光谱，但在较细的分辨率下，参数化有效地关闭自身，并允许显式地模拟云。该方案统一了细分辨率云解析模式(CRM)和粗尺度云解析模式(GCM)之间对云的处理，在两个模式中实现了统一参数化，其中一个模式具有有限的区域区域，能够进行显式云模拟。进行了分辨率为1-2公里的参考模拟，并与使用统一参数化的逐渐变粗的分辨率的模拟进行了比较。这些模拟评估了统一格式在多大程度上模拟了精细分辨率模拟的行为，并随着分辨率的增加而更忠实地逼近它。这些模拟旨在测试和开发灰色区域细尺度端的统一参数化法。另一个是GCM，即社区大气模式(CAM)，使用该模式的实验探索了该地区粗尺度一侧的参数化行为。由于统一的参数化具有改进用于天气预报和预测气候变化可能后果的模式的潜力，因此该工作具有更广泛的影响。积云参数化的缺陷是天气和气候模式存在偏差的一个持久来源，并降低了它们在研究和业务应用中的价值。此外，随着天气和气候模式采用更高的分辨率，工作中解决的隐式-显式混合云表示问题将变得越来越重要。使用由研究界开发并为研究界开发的免费提供的模型CAM作为该计划的试验台，使社区能够获得研究结果。此外，该项目将支持和培训一名研究生，从而在模型开发方面支持下一代研究人员。该项目是在大气过程多尺度建模中心(CMMAP)的支持下开始的研究的继续，CMMAP是NSF的一个科学和技术中心。

Atmospheric global circulation models (CGMs) used for weather prediction and climate simulation typically divide the atmosphere into a grid on which quantities like temperature and pressure take on a single value to represent mean conditions across each grid cell. For reasons of computational expense GCMs have traditionally used grid spacings of a hundred kilometers or more, at which it is not possible to represent individual clouds or even large cloud systems. Cumulus clouds, which produce intense precipitation and have other important effects, are represented indirectly in these models by convective parameterizations, which attempt to capture their aggregate behavior over a grid cell. With increasing computer power it has become possible to run GCMs at resolutions high enough to enter the "gray zone", in which some portion of the motions of the largest cumulus ensembles is explicitly represented on the model grid, but the smaller clouds must still be represented indirectly through convective parameterization. The goal of the work performed here is to develop a parameterization which adjusts automatically and seamlessly as grid spacing decreases and a greater portion of the cumulus activity is resolved explicitly. At coarser resolutions the parameterization represents the entire spectrum of cloud sizes and motions, but at fine resolutions the parameterization effectively shuts itself off and allows the clouds to be explicitly simulated. The scheme is called the "unified parameterization" because it unifies the treatment of clouds between fine resolution cloud resolving models (CRMs) and coarse-scale GCMs.The unified parameterization is implemented in two models, one a CRM with a limited regional domain which is capable of explicit cloud simulation. Reference simulations with a resolution of 1-2km are performed, and these are compared to simulations at progressively coarser resolutions using the unified parameterization. The simulations assess the extent to which the unified scheme mimics the behavior of the fine-resolution simulation and approximates it more faithfully as resolution increases. These simulations are meant to test and develop the unified parameterization at the fine-scale end of the gray zone. The other is a GCM, the Community Atmosphere Model (CAM), and experiments with this model explore the behavior of the parameterization on the coarse-scale side of the zone.The work has broader impacts due to the potential of the unified parameterization to improve models used for weather forecasting and for anticipating the likely consequences of climate change. Deficiencies in cumulus parameterization are a persistent source of bias in weather and climate models, and diminish their value for research and operational applications. Moreover, the issue of mixed implicit-explicit cloud representation addressed in the work will become increasingly important as higher resolutions are adopted for weather and climate models. The use of CAM, a freely available model developed by and for the research community, as a testbed for the scheme enables community access to results of the research. In addition, the project will support and train a graduate student, thereby supporting the next generation of researchers in model development. The project is a continuation of research begun with the support of the Center for Multiscale Modeling of Atmospheric Processes (CMMAP), an NSF Science and Technology Center.