Proposal Title: Improving Week 3-4 Weather Prediction Through a Global Convection-Allowing Version of the Unified Forecast System (UFS)
Principal Investigator: James L. Kinter III (GMU)
Co-Investigators:
Ben Cash (GMU)
Vijay Tallapragada (NOAA/NWS/EMC)
ABSTRACT:
The proposed work follows on a previously-awarded NOAA grant (NA18NWS4680045) to address the goals of the Next-Generation Global Prediction System (NGGPS) program, specifically the Research-to-Operations (R2O) Initiative goal of increasing the resolution of key environmental models to improve the accuracy and specificity of the forecast. The proposed work will address the priorities of the Weeks 3-4 program by continuing to develop and test the feasibility and potential for improvement in sub-seasonal weather forecasts through dramatically enhancing the spatial resolution of the global atmosphere-land component in the NOAA Unified Forecast System (UFS) to convection-allowing scale. Building on work done in the 2018-2020 project, a series of 30-day retrospective ensemble forecasts that covers the period 2011 – 2018, for selected dates in boreal winter and summer cases will be made with 4 different configurations of the UFS. The FV3 dynamical core of the atmospheric component of the UFS has been tested at various grid resolutions (C384 – 25-km grid spacing; C768 – 13-km grid spacing; C1536 – 6.5-km grid spacing, C3072 – 3.25-km grid spacing) and has the capability to run in both uniform global grid and refined mesh or nested-grid modes (e.g. Harris et al. 2016).
There have been several efforts to increase the spatial resolution of global climate models, with the goal of improving subseasonal to seasonal (S2S) prediction. While there have been advances, there are limits to the benefits provided by simply increasing spatial resolution; specifically, the representation of sub-grid scale parameterizations of physical processes continues to hinder progress. There are demonstrable advantages of using convection-allowing models (CAMs) instead of models with cumulus parameterization. We will test the hypothesis that more realistic convection, particularly in the tropics, can improve global forecasts in a series of 30-day reforecast experiments employing 4 different configurations of coupled UFS: (1) global “S2S” resolution; (2) global “NWP” resolution; (2) global “convection allowing” model (CAM) resolution; and (3) “refined grid” resolution with the CAM grid in the tropics and the NWP grid outside the tropics. The difference in forecast skill for precipitation and near-surface air temperature in the contiguous United States, including the potential for outlooks of temperature and precipitation extremes, will be quantified and diagnosed in terms of the contribution from local explicit representation of cumulus and the change in teleconnections. Process-level diagnosis will be employed to assess the contribution to S2S forecast skill from air-sea and air-land interactions.
