Verification and Calibration of Neighborhood and Object-Based Probabilistic Precipitation Forecasts from a Multimodel Convection-Allowing Ensemble

• DOI: 10.1175/mwr-d-11-00356.1
• Publication Date: 2012-04-05T19:19:08Z
• Authors: Aaron Johnson; Xuguang Wang
• Keywords: Maxima and minima; Probabilistic Forecasting

Abstract Neighborhood and object-based probabilistic precipitation forecasts from a convection-allowing ensemble are verified calibrated. Calibration methods include logistic regression, one- two-parameter reliability-based calibration, cumulative distribution function (CDF)-based bias adjustment. Newly proposed for the occurrence of forecast object derived percentage members with matching object. Verification calibration single- multimodel subensembles performed to explore effect using multiple models. The uncalibrated neighborhood-based have skill minima during afternoon convective maximum. generally improves skill, especially minima, resulting in positive skill. In general all perform similarly, slight advantage regression (one-parameter reliability based) 1-h (6 h) accumulations. are, general, less skillful than forecasts. Object-based also results at lead times. For is significantly different among methods, performing best CDF-based adjustment worst. both neighborhood forecasts, impact 10 or 25 days training data small most significant method. An Advanced Research Weather Forecasting Model (ARW-WRF) subensemble more an WRF Nonhydrostatic Mesoscale (NMM) subensemble. difference reduced by calibration. only shows beyond 1-day time no