Experiments with a 3-km Ensemble Kalman Filter Data Assimilation System over the Entire Conterminous United States

Based on the Weather Research and Forecasting (WRF) model, 80-member ensemble Kalman filter (EnKF) analyses with 3-km horizontal grid spacing were produced over a computational domain spanning the entire conterminous United States (CONUS) for 4 weeks using 1-h continuous cycling. For comparison, similarly-configured EnKF analyses with 15-km horizontal grid spacing were also produced.

At 0000 UTC, 36-h, 3-km forecasts were initialized from both EnKF mean analyses and individual analysis members. Results clearly show benefits of increased analysis resolution for short-term forecasts, as 3-km precipitation forecasts initialized from 3-km EnKF analyses were more skillful than those initialized from 15-km EnKF analyses through 12–18 h. Additionally, precipitation forecasts initialized from EnKF mean analyses were better than benchmark forecasts initialized by simply downscaling Global Forecast System (GFS) analyses onto the computational domain.

In addition to describing the promising EnKF results, this presentation will discuss technical challenges that were overcome to produce the 3-km analyses; the size of the 3-km domain was perhaps unprecedented and required substantial computational resources. Furthermore, sensitivity studies regarding assimilation of radar observations and experiments with “hybrid” variational–ensemble analyses over the large 3-km domain will be described.