Stochastic Physics in MPAS: Facilitating the Transition of MPAS into the UFS for Future Use in the RRFS

In preparation for the inclusion of the Model for Prediction Across Scales (MPAS) within the Unified Forecast System (UFS), the Developmental Testbed Center has begun work implementing and testing a stochastic physics capability in MPAS. Work is underway to 1) call the UFS stochastic physics pattern generation code from within stand-alone MPAS, and 2) pass the pattern back to MPAS and use it to apply the ad-hoc schemes—SPPT, SHUM, and SKEB—to MPAS state variables and/or physics tendencies during integration of the model. The addition of this functionality will allow these three stochastic physics schemes to be run within MPAS. This capability will in turn provide critical functionality for and expedite the transition of MPAS into the UFS framework in preparation for its use within version 2 of NOAA’s Rapid Refresh Forecast System (RRFSv2).

Tests were conducted to ensure the stochastic physics pattern generator is working as expected, and a number of sensitivity tests were run for each stochastic physics scheme available within the UFS stochastic physics repository, with the aim of evaluating their impact on specific model fields, error spectra, and computational costs. Efforts were made to ensure low impact to model runtime by exploring options such as pattern update frequency, with evaluation of the impact of such options on the forecast. After these tests, week-long spring and winter retrospective ensemble simulations were run using a variety of stochastic physics combinations to assess the impact on forecast accuracy and spread generation as a function of lead time using a variety of verification metrics. Through this work, the DTC hopes to play a critical role in helping prepare MPAS for implementation into the UFS, and, ultimately, for its use as the dycore underpinning the RRFS.