Nowcasts of precipitation rate are derived from rapid-update mosaics of radar reflectivity that are cleaned of echoes from non-hydrometeor sources through automated and manual means. Subsequent radar mosaics are cross-correlated to determine precipitation motion vectors. Using the motion vectors, radar echoes are extrapolated forward in time, then forecasts of precipitation rate and probability are derived at each grid point by analyzing the extrapolated radar forecasts at grid points within a radius of influence of the forecast point. Precipitation typing is applied using forecast model estimates, that are then refined using the most recent surface observations.
Further, NWP forecasts are run at convective-allowing scales (4 km) each hour over North America and Europe and at coarser resolution (12km) globally, in order to provide additional forecast information and sensible weather parameters. To assure consistency with observations and to provide increased skill in approximately the first 4 forecast hours, forecasts of temperature, dew point temperature, cloud cover and wind are corrected by subtracting an error value that decays from the total error at the observation time to nil several hours into the forecast.
Nowcasts of precipitation fields are temporally blended with precipitation fields from the NWP forecasts to provide maximum forecast skill throughout the 0-6 hour period. The temporal weighting is a function of forecast time that favors extrapolated radar forecasts in the early period transitioning to favoring NWP forecasts in the later period. Further the weighting function is dynamically adjusted based on local atmospheric instability to favor an earlier transition from radar extrapolation towards NWP forecasts when the atmosphere is more convectively unstable.
Finally, government issued severe weather warnings are used to augment the 0-6 hour forecasts to assure that the very short-term precipitation forecasts are consistent with critical warning information, and to include appropriate qualifiers (such as severe) in worded forecasts driven by the 0-6 hour forecasts.
This presentation will provide additional details of the methodology used for generating 0-6 hour forecasts, verification of the 0-6 hour forecasts, and examples of uses of these forecasts.