11.5
Improved Methods for Blending Extrapolation-based with High-resolution Model Forecasts

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 7 January 2015: 5:00 PM
129A (Phoenix Convention Center - West and North Buildings)
James O. Pinto, NCAR/Research Applications Laboratory, Boulder, CO; and J. Grim, D. Ahijevych, and M. Steiner

Handout (2.7 MB)

Accurate short-term prediction of the location, intensity and organization of high-topped convective storms is vital for setting up air-space flow structures for routing air traffic across the National Airspace. CoSPA, which combines extrapolation-based nowcasts generated by the MIT Lincoln Laboratory with the High Resolution Rapid Refresh (HRRR) model forecasts produced by the NOAA/ESRL Global Systems Division based on a blending technology established by the NCAR Research Applications Laboratory, was developed to supplement the Collaborative Convective Forecast Products (CCFP) today's primary convective weather guidance used in the daily strategic aviation planning. The current blending module of CoSPA, which is run at NCAR, includes a model calibration step, a model position error correction step, and a heuristic weighted averaging step. The heuristic weighted averaging step includes separate treatments of the weights for areas classified as ongoing precipitation versus those classified as initiating precipitation. The initiation areas are generally given more weight than areas classified as ongoing precipitation with the weight increase obtained using a constant multiplier. The basic assumption here is that the model performs similarly for all types of initiation in all regions of the country at all times of day.

The work presented here focuses on improving the treatment of the weights in areas classified as initiating precipitation by allowing the weights to vary with region, time of day and storm type. New areas of precipitation are identified in the model forecasts using an object-based technique. In the case of the hourly-updating HRRR, multiple forecasts are typically valid at a given time in the future allowing one to build a time-lagged ensemble. The uncertainty information contained in this time-lagged ensemble is then used to determine the magnitude of the weights in areas identified as initiating precipitation. The new treatment of storm initiation is explained in detail and demonstrated on several test cases. The skill of the new technique is then compared to that obtained with the version of CoSPA demonstrated during the summer of 2014.

This research is supported by the Federal Aviation Administration (FAA). The views expressed are those of the authors and do not necessarily represent the official policy or position of the FAA.