Censored and shifted gamma distribution based EMOS model for probabilistic quantitative precipitation forecasting

Abstract

Recently all major weather prediction centres provide forecast ensembles of different weather quantities which are obtained from multiple runs of numerical weather prediction models with various initial conditions and model parametrizations. However, ensemble forecasts often show an underdispersive character and may also be biased, so that some post-processing is needed to account for these deficiencies. Probably the most popular modern post-processing techniques are the ensemble model output statistics (EMOS) and the Bayesian model averaging (BMA) which provide estimates of the density of the predictable weather quantity. In the present work an EMOS method for calibrating ensemble forecasts of precipitation accumulation is proposed, where the predictive distribution follows a censored and shifted gamma (CSG) law with parameters depending on the ensemble members. The CSG EMOS model is tested on ensemble forecasts of 24 h precipitation accumulation of the eight-member University of Washington mesoscale ensemble and on the 11 member ensemble produced by the operational Limited Area Model Ensemble Prediction System of the Hungarian Meteorological Service. The predictive performance of the new EMOS approach is compared with the fit of the raw ensemble, the generalized extreme value (GEV) distribution based EMOS model and the gamma BMA method. According to the results, the proposed CSG EMOS model slightly outperforms the GEV EMOS approach in terms of calibration of probabilistic and accuracy of point forecasts and shows significantly better predictive skill that the raw ensemble and the BMA model.

20 pages, 5 figures, 8 tables

References in corpus (2)

• Log-normal distribution based EMOS models for probabilistic wind speed forecasting
• Forecast verification for extreme value distributions with an application to probabilistic peak wind prediction

Cited by in corpus (11)

• Combining predictive distributions for statistical post-processing of ensemble forecasts
• Distributional Regression Forests for Probabilistic Precipitation Forecasting in Complex Terrain
• Post-processing numerical weather prediction ensembles for probabilistic solar irradiance forecasting
• Forest-based methods and ensemble model output statistics for rainfall ensemble forecasting
• Statistical post-processing of hydrological forecasts using Bayesian model averaging
• Statistical post-processing of heat index ensemble forecasts: is there a royal road?
• A two-step machine learning approach to statistical post-processing of weather forecasts for power generation
• Statistical post-processing of ensemble forecasts of temperature in Santiago de Chile
• Machine learning-based probabilistic forecasting of solar irradiance in Chile
• Statistical post-processing of dual-resolution ensemble forecasts
• Post-processing of ensemble photovoltaic power forecasts with distributional and quantile regression methods