Title : Progress toward reducing cloud/radiation errors from 3 km/ 3 h forecasts to 30 km/ 3 4 week forecasts
Subgrid scale cloud representation and the closely related surface energy balance continue to be a central challenge from subseasonal to seasonal models down to storm scale models applied for forecast duration of only a few hours. Previously, NOAA/ESRL confirmed this issue from its 3 km model (HRRR using WRF ARW) for short range forecasting including sub grid scale cloud representation but also up to a 25 km subseasonal model (FV3 GFS) testing a common suite of scale aware physical parameterizations.
In a major physics suite component developed during 2018 to 2020 a modified representation of subgrid cloud water in the turbulence MYNN parameterization resulted in much improved agreement with radiation measurements as shown with 2018 2020 testing of the 3km HRRR model. Latest results will be shown using SURFRAD radiation and METAR ceiling observations, indicating much improved bias in downward solar radiation and in cloud location (via mean absolute error metric), as well as with 2m temperature and precipitation. This same parameterization suite including the MYNN scheme isalso evaluated with the FV3 GFS global model, again for cloud radiation errors.
At our NOAA Global Systems Laboratory, our group has also evaluated physical parameterizations and data assimilation techniques to improve accuracy of the surface energy balance. Therefore, this seminar will include an overall description of the NOAA High Resolution Rapid Refresh (HRRR) assimilation/modeling system as well as its FV3 dycore based regional andglobal models now used by NOAA for its future model development.