Assimilation of Doppler Radar Data with an Ensemble-based Variational Method for Storm-scale Numerical Weather Prediction

点击翻译按钮获取中文摘要

Most currently employed convective-scale data assimilation (DA) schemes were developed primarily for application to larger-scale atmospheric flows and weather phenomena, in which sharply contrasting balances and constraints are relevant. Moreover, at convective scales Doppler radar is the only widely available means of providing extensive observations of sufficiently high spatial and temporal resolution needed to facilitate dynamic prediction of high-impact weather phenomena such as severe thunderstorms. As such, the effective assimilation of Doppler radar data into convection-resolving models is of increasing importance, yet underutilized. Building upon their previous work on convective-scale DA, these researchers will explore new approaches to optimally assimilate operationally-collected WSR-88D Doppler radar data available from a newly-upgraded national network employing dual-polarization technology, and in particular will: (i) Determine how to best use reflectivity observations in addition to radial velocity data; (ii) examine the usefulness of the background tendency information in a storm-scale DA system; and (iii) implement an efficient ensemble-based hybrid three-dimensional variational/Ensemble Kalman Filter (3DVAR/EnKF) framework that incorporates existing mesoscale ensemble forecast information into a storm-scale three-dimensional variational DA system.The Intellectual Merit of this effort centers upon developing a novel DA strategy that makes optimal use of both radar reflectivity and radial velocity fields, which are uniquely suitable for specifying rapidly evolving convective-scale flows, in order to provide initial conditions for high-resolution storm-scale NWP models such as the NSF/NCAR Weather Research and Forecasting (WRF) model--a system that enjoys use in both research and operational settings. This approach seeks to improve our physical understanding of convective storm-scale dynamics and is further aimed toward improved detection and anticipation of thunderstorm-related hazards as well as more accurate quantitative precipitation forecasts needed for hydrological applications. This work may also help to solve the initial "balance problem" that is inherent in convective numerical weather prediction (NWP), but has heretofore been largely overlooked by the research community. Broader Impacts of this work include deriving maximum benefit from the considerable U.S. investment in the nationwide WSR-88D radar network by accelerating the use of these data in both operational and research-based NWP. This project will embody educational benefits through mentoring of graduate students and a postdoctoral research associate, and emerging results will be integrated into teaching materials and publications reaching broad audiences.