Comparison between ARW and NMM objective forecast verification scores

The Weather Research and Forecasting (WRF) model contains two dynamic cores: the Non-hydrostatic Mesoscale Model (NMM), developed at the National Centers for Environmental Prediction (NCEP), and the Advanced Research WRF (ARW), developed at the National Center for Atmospheric Research. Each dynamic core corresponds to a set of dynamic solvers that operates on a particular grid projection, grid staggering, and vertical coordinate. The WRF model also contains a multitude of physical parameterizations, many of which can be used with both dynamic cores.

This work builds on the Developmental Testbed Center Rapid Refresh Core Test (RRCT), in which ARW and NMM were compared out to 24 h. The similarity in forecast verification scores between ARW and NMM found in the RRCT led to the concept that research results (for example, new physical parameterizations) tested with one core could be transferred to the other. Given that both dynamic cores of WRF are currently run operationally, the transfer of research results can substantially reduce the amount of testing required for operational implementation of new developments, potentially accelerating operational NWP improvements in the US.

To ascertain whether the similarity in forecast verification results would hold out to longer forecast lead times, it was deemed necessary to extend the RRCT out to 60 h forecasts. The configuration for this experiment is similar to the one used in the RRCT (13-km horizontal grid spacing over the CONUS), with some important changes (updates in all codes used, changes in initialization procedures, and different cycles). The results indicate the ARW and NMM 24-h accumulated precipitation frequency bias and equitable threat scores are indistinguishable. On the other hand, some statistically significant differences occur in the forecasts of temperature, wind and relative humidity. However, while the errors of each dynamic core often grow in time, the inter-core differences remain unchanged after an initial spin-up period.