Assimilating novel ground-based remote sensing observations into a numerical weather prediction model for improving model predictions and advancing knowledge of atmospheric boundary layer processes

虽然大气边界层在云和降水的日循环、幕面预报和恶劣天气事件中起着至关重要的作用，但其初始状态在数值天气预报（NWP）模式中是不受约束的。特别是对于目前的短程对流尺度NWP模式，关于垂直分层和边界层稳定性的真实初始化被证明是非常有价值的。地面遥感设备提供边界层的高频热力学和动量剖面观测。然而，目前还没有策略，如何利用这些数据最优地进行数值天气预报，因为商业制造商在最近几年才开发出价格合理的高质量设备。此外，边界层中的许多过程仍然没有完全被理解，特别是对于稳定的边界层，对于亚网格尺度的地形效应，重力波，热环流和亚网格尺度的表面非均质性效应。通常，大气模式对亚网格尺度边界层过程的模拟存在不足。拟议项目的目标是双重的：在第一阶段，将利用最新的数据同化（DA）和地面遥感的发展，包括制定最佳同化策略，利用最先进的地面遥感边界层观测（微波辐射计和多普勒激光雷达）来改善大气模式中边界层的初始状态和后续预测。由于数据分析社区的发展，采用集成卡尔曼滤波器进行对流尺度数据同化，提供高分辨率的流动相关协方差，并允许高频率的更新周期，因此以这种方式捕获数据分析系统中边界层典型的小尺度结构的机会现在是可能的。在第二阶段，我们建议利用这个扩展的NWP框架的巨大潜力，包括数据同化系统和连续观测的质量检查观测，作为推进基本边界层研究的新方法。将利用从数据分析系统获得的信息来纠正模型对观测结果的第一次猜测，以及分析本身作为一个自洽的三维状态在公里尺度上接近真实状态。O-B（观测值减去背景值）统计量指出了模型第一次猜测和观测值之间的系统性差异，第一次猜测的增量是系统中导致偏离真实状态的错误的指标。为了获得关于单一过程的信息，将采用Klocke和Rodwell（2013）研究中的“初始趋势方法”。

Although the atmospheric boundary layer plays a crucial role for the diurnal cycle of clouds and precipitation, screen level predictions and severe weather events, its initial state is under-constrained in numerical weather prediction (NWP) models. Especially for nowadays short-range convective-scale NWP models a realistic initialization with respect to the vertical stratification and stability of the boundary layer is expected to prove highly valuable. Ground-based remote sensing devices deliver high-frequent thermodynamic and momentum profile observations of the boundary layer. However no strategies exist yet, how to exploit those data optimally for numerical weather prediction, as the development of affordable high-quality devices by commercial manufacturers has taken place only in the recent years.Moreover many processes in the boundary layer are still not completely understood and, this holds in particular for the stable boundary layer, for subgrid-scale orographic effects, gravity waves, thermal circulations and subgrid-scale surface heterogeneity effects. Often, the simulation of subgrid-scale boundary layer processes in atmospheric models shows deficiencies.The objective of the proposed project is twofold:In a first phase the usefulness of state-of-the-art ground-based remote sensing boundary layer observations (microwave radiometer and Doppler lidar) to improve the initial state and subsequent forecast of the boundary layer in atmospheric models will be exploited, making use of recent developments in data assimilation (DA) and ground-based remote sensing, including the development of an optimal assimilation strategy. The chance to capture small-scale structures typical for the boundary layer in the DA system in such a way is now possible due to developments in the DA community to employ Ensemble Kalman Filters for convective-scale data assimilation, providing high-resolution flow-dependent covariances and allowing high-frequent update cycles.In a second phase we propose to exploit the enormous potential of this expanded NWP framework including data assimilation system and continuously observed quality-checked observations as a novel approach to advance basic boundary layer research. Both, the information obtained from the DA system to correct the model first guess towards the observations but also the analysis itself as a self-consistent three-dimensional state at the kilometer-scale close to the true state, will be exploited. The O-B (observation minus background) statistics point to systematic differences between model first guess and observations, the increments added to the first guess are an indicator of errors in the system leading to deviations from the true state. To obtain information on single processes, the “initial tendency approach” as in the study by Klocke and Rodwell (2013) will be applied.