Quantile regression as a means of calibrating and verifying a mesoscale NWP ensemble

Since 2007 the Dugway Proving Ground (DPG) in Utah has been using an operational mesoscale ensemble NWP system for mission-critical weather forecasting. The ensemble comprises 30 members, each a different configuration of the Four-Dimensional Weather (4DWX) system jointly developed by the National Center for Atmospheric Research and the U. S. Army Test and Evaluation Command.

Many users of the weather forecasts at DPG benefit from objective probabilistic guidance, which requires that the ensemble be formally calibrated. The calibration technique that NCAR has developed is based on quantile regression (QR) under a step-wise forward selection framework. Model selection for each quantile relies on both the QR cost function and the binomial distribution, leading to ensemble forecasts with both good reliability and sharpness. In addition, a second pass is performed to re-calibrate over separate intervals of self-diagnosed forecast instability, leading to a calibrated ensemble forecast with an informative skill-spread relationship. Currently the variables being calibrated are near-surface temperature, humidity, and wind, as well as precipitation.

The presenter will describe the mesoscale ensemble, review the steps used to calibrate the ensemble forecast, and present verification statistics and figures from operational ensemble forecasts before and after calibration.