Observation System Experiments with the Hourly Updating Rapid Refresh Model Using GSI Hybrid Ensemble-Variational Data Assimilation Monthly Weather Review

A set of observation system experiments (OSEs) over three seasons using the hourly updated Rapid Refresh (RAP) numerical weather prediction (NWP) assimilation-forecast system identifies the importance of the various components of the North American observing system for 3-12-h RAP forecasts. Aircraft observations emerge as the strongest-impact observation type for wind, relative humidity (RH), and temperature forecasts, permitting a 15%-30% reduction in 6-h forecast error in the troposphere and lower stratosphere. Major positive impacts are also seen from rawinsondes, GOES satellite cloud observations, and surface observations, with lesser but still significant impacts from GPS precipitable water (PW) observations, satellite atmospheric motion vectors (AMVs), and radar reflectivity observations. A separate experiment revealed that the aircraft-related RH forecast improvement was augmented by 50% due specifically to the addition of aircraft moisture observations. Additionally, observations from en route aircraft and those from ascending or descending aircraft contribute approximately equally to the overall forecast skill, with the strongest impacts in the respective layers of the observations. Initial results from these OSEs supported implementation of an improved assimilation configuration of boundary layer pseudoinnovations from surface observations, as well as allowing the assimilation of satellite AMVs over land. The breadth of these experiments over the three seasons suggests that observation impact results are applicable to general forecasting skill, not just classes of phenomena during limited time periods.

本研究依托逐时更新的快速刷新（Rapid Refresh, RAP）数值天气预报（numerical weather prediction, NWP）同化预报系统，开展了覆盖三个季节的观测系统试验（observation system experiments, OSEs），以此明确北美观测系统各组成部分对3~12小时RAP预报的重要性。试验结果表明，在风场、相对湿度（relative humidity, RH）与温度预报中，飞机观测的影响效应最为显著，可使对流层及平流层下部的6小时预报误差降低15%~30%。探空仪、地球静止业务环境卫星（GOES）云观测以及地面观测同样带来了显著的积极影响；而GPS大气可降水量（precipitable water, PW）观测、卫星大气运动矢量（atmospheric motion vectors, AMVs）以及雷达反射率观测的影响相对较弱，但仍具备统计学显著性。另一项独立试验显示，仅通过新增飞机水汽观测，即可使飞机观测对相对湿度预报的提升效果增强50%。此外，巡航阶段飞机观测与起降阶段飞机观测对整体预报技巧的贡献大致相当，且其影响最强的大气层次恰好对应观测所在的高度层。本次观测系统试验的初步结果支持两项改进方案的落地：一是优化地面观测的边界层伪创新同化配置，二是允许同化陆地区域的卫星大气运动矢量。本次试验覆盖三个季节的宽尺度设计表明，观测影响评估结果可推广至通用预报技巧提升，而非仅适用于特定时段的特定现象类别。