Precipitation Spin-up Processes in a Global Model with a Cloud-Permitting-Scale Mesh Refinement

Quantitative precipitation forecasts from the global Model for Prediction Across Scales (MPAS) with mesh refinement from a global 15-km horizontal grid spacing to 3-km over CONUS showed good forecast skill during NOAA's Hazardous Weather Testbed Spring Experiment 2016 in May. Daily global 120-h forecasts were initialized using the 00 UTC 0.25° GFS analysis. Precipitation statistics over the Central Plains showed that the model is able to capture the average diurnal cycle and frequency distributions at various rain rate thresholds well, as compared with those from the high-resolution Multi-Radar Multi-Sensor (MRMS) precipitation analyses. The 24-48 h forecasts showed more skill than the 0-24 h forecasts due to the time needed for the model to spin up on convective precipitation processes. Here, we utilize the skillful 24-48 h forecasts to examine the causes of the precipitation spin-up. Comparisons between the 0-24 h and 24-48 h forecasts show that cold pool dynamics are typically missing in the 00 UTC initializations. A sensitivity experiment is conducted to examine the impact of these fine-scale dynamics at initial time on the precipitation spin-up time (for a ‘dry’ start). Comparisons with a globally uniform MPAS with 15-km horizontal grid spacing, which relies more on the convection parameterization scheme, will also be shown.