9.5 Refinement and testing of analysis nudging in MPAS-A

Bullock, Russell, Robert Gilliam, Hosein Foroutan, and Jerold Herwehe, U.S. Environmental Protection Agency

The Model for Prediction Across Scales - Atmosphere (MPAS-A) is being adapted to serve as the meteorological driver for EPA's "next-generation" air-quality model. To serve that purpose, it must be able to function in a diagnostic mode where past meteorological conditions are represented in greater detail and accuracy than can be provided by available observational data and meteorological reanalysis products. MPAS-A has been modified to allow four dimensional data assimilation (FDDA) by the nudging of temperature, humidity and wind toward target values predefined on the MPAS-A computational mesh. The technique of "analysis nudging" developed for the Penn State / NCAR Mesoscale Model – Version 4 (MM4), and later applied in the Weather Research and Forecasting model (WRF), is applied here in MPAS-A with adaptations for the unstructured Voronoi mesh used in MPAS-A. Target fields generated from MPAS-A initialization software and 1 × 1 degree National Centers for Environmental Prediction FNL (Final) Operational Global Analysis data were used to constrain MPAS-A simulations on a 92-25km variable-resolution mesh with refinement centered over the contiguous United States. Test simulations for the periods of January and July 2013, with and without FDDA, are compared to target fields at various vertical levels and to surface-level meteorological observations. The results show the ability to follow target fields with high fidelity while still maintaining conservation of mass as in the original model. The results also show model errors relative to observations continue to be constrained throughout the simulations using FDDA and even show some error reduction during the first few days that could be attributable to the finer resolution of the 92-25 km computational mesh relative to the 1 × 1 degree meteorological data used for model initialization and FDDA.