Diverse Planetary Boundary Layer (PBL) Algorithms within Multimodels and the Design of Unified Boundary Layer Modeling for Improved Forecasts

本研究利用高分辨率全球多模式改进了行星边界层模型。将研究多达五种不同的行星边界层算法。这些包括1)佛罗里达州立大学基于混合长度（局部闭合）的k理论，2)印度国家中期天气预报中心基于混合长度的k理论，其中稳定性依赖于Bulk Richardson数，3)对非局部方案的改进梯度方法，该方案允许沿梯度向下的通量以及对流情况，4)基于国家大气科学中心社区气候系统模式-2的非局部方案，其中地表通量与上述公式不同；5)湍流动能方法，即1.5阶闭合方案。这项研究是最近完成的工作的续集，该工作涉及利用使用单一边界层方案的单一模式、统一的单一模式（综合所有这些方案）和构建消除偏倚的边界层通量预测的多模式超系综来模拟和预测行星边界层（PBL）通量的技能。需要基准观测来验证这些预测。这些数据来自于对雨率初始化的再分析数据集的使用，以及对（对流层上空）视热源和水分汇的垂直积分。这提供了热量和水分的表面通量作为问题的残余，由于该产品更依赖于观测而不是模型，因此它是观测通量的有用代理。pi将对每种模式、统一模式和多模式超系综进行近100次预测实验。这将为他们提供高分辨率的这些字段的误差测量。接下来将对这些错误的来源进行更详细的分析。在此之前，PI进行了一项统计研究，该研究表明，可以标记导致错误的模型动力学和物理区域（包括它们在三维中的位置）。这种对PBL算法内部细节的分析将是可能的，并且可以为未来的模型改进提供见解；这是以前没有做过的。知识价值：该研究被设计为一个综合项目，其范围包括对大气行星边界层建模的许多可能贡献。pi将使用一个框架来处理PBL，其中在PBL建模中找到错误的来源是一个中心问题。数据和建模的不确定性由系综/超系综方法涵盖。方法并不局限于预测建模。它包括关于PBL物理的几个公式的基础科学问题以及发现其局限性的方法。更广泛的影响：行星边界层通量影响云生长和积云参数化的整个问题。这些反过来又会影响风暴的生命周期。本研究不仅旨在从统一的单模型和多模型（超系综）的构建中为PBL通量提供一些改进的模型，而且还扩展到对PBL物理中导致误差增长的区域进行标记。鉴于这五种不同的局域、非局域和湍流动能方案，误差标记（在三维空间）有助于进一步推进数值天气预报。两名研究生将在此资助下接受教育和培训。

This research addresses improved planetary boundary layer modeling using high resolution global multi-models. As many as five different planetary boundary layer algorithms will be examined. These include 1) Florida State University's K-theory based on mixing length (local closure), 2) India's National Center for Medium Range Weather Forecasting's K-theory with mixing length where stability dependence is strongly dependent on the Bulk Richardson number, 3) a modified gradient approach to a non-local scheme that allows for fluxes down the gradient as well as for convective situations, 4) a non-local scheme based on National Center for Atmospheric Sciences' Community Climate System Model-2, where the surface fluxes are different from the above formulation and 5) a turbulent kinetic energy approach, i.e. a 1.5 order closure scheme. The study is a sequel to recently completed work that address the skills of modeling and forecasting the skills of planetary boundary layer (PBL) fluxes using single models that use single PBL schemes, a unified single model (that synthesizes all such schemes) and a multimodel superensemble that constructs bias-removed forecasts for the PBL fluxes. Benchmark observations are needed for the validation of these forecasts. These are derived from the use of reanalysis data sets that are subjected to rain rate initialization and from a vertical integral (over the troposphere) of the apparent heat sources and moisture sinks. This provides the surface fluxes of heat and moisture as a residual of the problem, since this product is more reliant on observations rather than on models it is a useful proxy for the observed fluxes. The PIs will carry out nearly 100 forecast experiments with each model, with a unified model and with the multimodel superensemble. This will provide them with measures of the errors for these fields at high resolution. This inventory of errors will be followed up with a more detailed analysis on the sources of these errors. This follows a statistical study by the PI, that demonstrates that it is possible to tag areas of model dynamics and physics that contribute to errors (including their location in three dimensions). Such an analysis for the inner details of PBL algorithms will be possible and can provide insights for future model improvements; this has not been done previously. Intellectual Merit: The research is designed as a comprehensive project with its scope encompassing many possible contributions to modeling of the atmospheric planetary boundary layer. The PIs will use a framework for addressing the PBL where finding the source of errors in PBL modeling is a central issue. The data and modeling uncertainties are covered by an ensemble/superensemble approach. The methodology is not limited to forecast modeling. It includes basic science issues on several formulations of PBL physics and ways to find their limitations. Broader Impacts: Planetary Boundary Layer fluxes impact the entire issue of cloud growth and cumulus parameterization. Those, in-turn, impact storm life cycles. This research aims not only to provide somewhat improved models for the PBL fluxes from the construction of unified single and multimodels (the superensemble) but also extends to the tagging of the areas of the PBL physics that contribute to the growth of errors. Given five such diverse local, non-local and turbulent kinetic energy based schemes, the tagging of errors (in three dimensional space) can contribute to further advancement of numerical weather prediction. Two graduate students will be educated and trained under this grant.