Probability distributions of apparent temperature from ensemble MOS

Probabilistic forecasts of weather elements inherently contain more value than their single-valued counterparts. Recently, a number of techniques have appeared that can create probabilistic forecasts using the output from ensembles of numerical weather prediction models. The Meteorological Development Laboratory has developed a technique named ensemble kernel density model output statistics (EKDMOS) that can produce a forecast probability density function (PDF) or cumulative distribution function (CDF) for weather elements using error estimation from linear regression and kernel density fitting (Glahn, et al. 2009).

The EKDMOS technique has now been used to generate forecast PDFs/CDFs of heat index (HI) and wind chill (WC). Model estimates of HI and WC were important predictors for these weather elements, and the development sample for each had to be limited to cases where the element was well-defined and useful. In geographic areas where HI and WC occurred infrequently, regions were selected and data were combined to develop regional equations.

HI attempts to reflect the combined effects of heat and humidity on the human body. Similarly, WC attempts to reflect the combined effects of cold and wind on the human body. Both HI and WC are used extensively by the National Weather Service (NWS) to inform public health officials and the general public about forecast weather hazards. For convenience, forecasts of HI, WC, and T are frequently combined into a single weather element named apparent temperature (AppT).

EKDMOS forecast CDFs of HI, WC, and T are combined to form AppT. The AppT forecasts have been mapped to the National Digital Forecast Database (NDFD) 2.5-km grid. The resultant grids will be available to NWS customers and parters as part of the National Digital Guidance Database.

Reference

Glahn, B., M. Peroutka, J. Wiedenfeld, J. Wagner, G. Zylstra, B. Schuknecht, and B. Jackson, 2009: MOS uncertainty estimates in an ensemble framework. Mon. Wea. Rev., 137, 246–268.