Evaluation of the Model for Prediction across Scales (MPAS) in a Retrospective Application with Comparisons to WRF

The United States Environmental Protection Agency (US EPA) is leveraging recent advances in meteorological modeling to construct a next generation air quality modeling system that will allow consistent modeling from the global to local scales. The Model for Prediction Across Scales-Atmosphere (MPAS-A) has been developed by the National Center for Atmospheric Research (NCAR) in recent years as a global alternative to the regional Weather Research and Forecasting model (WRF). MPAS-A has been chosen as the initial platform for this next generation air quality modeling system. US EPA developers of the Community Multiscale Air Quality (CMAQ) modeling system have been coupling CMAQ components as subroutine calls within the MPAS-A to enable full global chemical transport modeling. As with any air quality model, meteorology is the primary driver of the chemistry and transport. We have developed physics updates in MPAS-A for retrospective applications. These include long-tested techniques like four-dimensional data assimilation that allows for long simulations of past weather with no growth in error and no re-initialization requirement. Furthermore, we have added the Pleim-Xiu land-surface model (P-X LSM), the Asymmetric Convective Model 2 (ACM2) boundary layer scheme, and the Pleim surface layer scheme, which are preferred options used for many years in WRF, as well as a few other enhancements like updated landuse and processing that refines the soil nudging inputs for the P-X LSM.

We evaluate simulations of 2016 that were produced running our EPA-enhanced MPAS-A on two different mesh structures. A 92 km global mesh seamlessly refined to 25 km over the CONUS as well as a finer 46 km global to 12 km CONUS. Since WRF has been used for many years to drive CMAQ, we include the performance of WRF in the various evaluations as a benchmark. The evaluation is comprehensive in terms of examining surface meteorology, precipitation, and shortwave radiation in comparison with surface observations, and upper-air using global rawinsonde measurements.

The evaluations show that MPAS-A generally compares well with WRF for most of 2016. Error levels in surface meteorology are at or below WRF levels during the cooler parts of the year and slightly higher during the summer. Precipitation totals are generally less than WRF and monthly observed totals during the cooler half of the year, but the differences are more mixed during the warm season. Patterns of precipitation (spatial correlation) are well replicated in both modeling systems. Solar radiation in MPAS-A is higher than WRF and measurements, indicating fewer clouds in MPAS-A versus WRF. Finally, a comparison of the 92-25 km and 46-12 km meshes shows clear improvements in surface meteorology as grid scale is reduced. The comparisons indicate the meteorology of MPAS-A is sufficient for retrospective global air quality modeling.