3.2

**Bayesian Processor of Output for Probabilistic Quantitative Precipitation Forecasting**

**Coire J. Maranzano**, University of Virginia, Charlottesville, VA; and R. Krzysztofowicz

The Bayesian Processor of Output (BPO) is a theoretically-based technique for probabilistic forecasting of weather variates. It processes output from a numerical weather prediction (NWP) model and optimally fuses it with climatic data in order to quantify uncertainty about a predictand. One version of the BPO is for continuous predictands. It is being tested by producing Probabilistic Quantitative Precipitation Forecasts (PQPFs) for a set of climatically diverse stations in the contiguous U.S. For each station, the PQPFs are produced for the same 6-h, 12-h, and 24-h periods up to 84-h ahead for which operational forecasts are produced by the AVN-MOS (Model Output Statistics technique applied to output fields from the Global Spectral Model run under the code name AVN). The inputs into the BPO are estimated as follows. The prior distribution is estimated from a climatic sample of the predictand collected during 7 years (January 1997 – December 2003); this sample is retrieved from the National Climatic Data Center. The family of the likelihood functions is estimated from a joint sample of the predictor vector and the predictand collected during 4 years (April 1997 – March 2001); this sample is retrieved from the same archive that the Meteorological Development Laboratory of the National Weather Service utilized to develop the AVN-MOS system. This talk will highlight some results from the testing: a numerical example of the estimation of the BPO, and a comparative verification of the BPO forecasts and the MOS forecasts. It will also outline the evaluation, which will be performed in three stages in order to determine: (i) improvement due to the theoretic structure of the BPO, (ii) improvement due to the use of climatic data and the optimal fusion of information, and (iii) improvement due to the optimal selection of predictors according to a criterion of informativeness.

Session 3, Bayesian Probability Forecasting

**Monday, 30 January 2006, 4:00 PM-5:00 PM**, A304** Previous paper Next paper
**