Ensemble analyses and forecasts of an operational multi-scale mesoscale ensemble data assimilation and prediction system (E-RTFDDA) for the Mountain States

Mesoscale (10-2000 km) meteorological processes over complex terrain are featured with rapid weather changes in space and time, which render them less predictable. These processes are influenced by synoptic circulations, but largely determined by regional and local terrain, land-surface heterogeneity, and associated physical properties. Physical processes such as radiative transfer, cloud and precipitation, boundary layer mixing, play a crucial role in shaping the regional weather and climate. Dynamically and thermodynamically forced terrain flows over mountain ranges are very sensitive to all these factors, which can frequently impair the ability of single-model deterministic forecasts. To consider the uncertainties induced by various aspects of the modeling system, an innovative mesoscale ensemble analysis and prediction system has been developed. The system, named as E-RTFDDA (Ensemble Real-Time Four Dimensional Data Assimilation and forecasting system), is built up on the Weather Research and Forecasting model (WRF) and the Penn-State NCAR Mesoscale Model version 5 (MM5) that are furnished with a continuous four-dimensional data assimilation (FDDA) scheme. The FDDA allows to effectively assimilate all observations into WRF and MM5 models and produces dynamically-balanced and physically-consistent 4D analyses and “spun-up” initial conditions to initialize each forecast cycles, which is particularly important over complex terrain. The ensemble realizations sample uncertainties in model initial conditions (IC), lateral boundary conditions (LBC), model physical parameterizations (PP), and the underlying land-surface (LS) properties, using advanced approaches. A 30-member E-RTFDDA system has been implemented on an Army HPC platform and started forecasting operation since September 2007. The system has three nested domains with grid sizes of 30, 10 and 3.33 km respectively. The 30-km domain contains the Western/Mountain states. The 10-km domain extends over Utah and the most part of Colorado where the Rocky Mountains are high and feature steep slopes, and the 3.33-km domain covers a 230 by 230 km region in the middle Northwest Utah where inter-mountain flows are common. The system cycles at 6-hr intervals and produces 48 h forecasts in each cycle. Weather forecasting and analysis examples and studies of local flow predictability at the different regions in the Rockies, in particular the strong wind scenarios in the 10-km domain 2 strong, will be presented.