Data assimilation and medium-range ensemble forecasting using variable-resolution global models with ~3-km horizontal grid spacing over the United States

Craig Schwartz

Mesoscale and Microscale Meteorology Lab National Center for Atmospheric Research

Tuesday, Nov 07, 2023, 2:00 pm
DSRC Room GC402

Abstract

I’ll discuss two topics that are bound together by their common use of variable-resolution global models with convection-allowing (~3-km) horizontal grid spacing over the conterminous United States (CONUS) and coarser horizontal resolution over the rest of globe.

First, I’ll discuss medium-range (3–8-day) ensemble forecasts produced with variable-resolution global configurations of both the Model for Prediction Across Scales (MPAS) and an FV3-based model. Both sets of ensemble forecasts were run in real-time during NOAA’s 2023 Hazardous Weather Testbed Spring Forecasting Experiment, marking the first time that medium-range convection-allowing ensemble forecasts have been produced over the CONUS in real-time. The MPAS and FV3-based ensembles were objectively skillful and often provided valuable guidance concerning precipitation and severe weather at 3–8-day forecast ranges.

Second, I’ll describe results from several month-long ensemble-variational (EnVar) data assimilation experiments with MPAS-JEDI that likely represented the first demonstrations of continuously cycling global data assimilation on a variable-resolution mesh with a large area of convection-allowing cell spacing. These experiments were an initial step toward realizing a longer-term goal of developing a global convection-allowing ensemble-based data assimilation and prediction system with MPAS and JEDI, which I will briefly detail.