Abstract:Abstract Forecasts produced by EPSs provide the potential state of the future atmosphere and quantify uncertainty. However, the raw ensemble forecasts from a single EPS are typically characterized by underdispersive predictions, especially for precipitation that follows a right-skewed gamma distribution. In this study, censored and shifted gamma distribution ensemble model output statistics (CSG-EMOS) is performed as one of the state-of-the-art methods for probabilistic precipitation postprocessing across China. Ensemble forecasts from multiple EPSs, including the European Centre for Medium-Range Weather Forecasts, the National Centers for Environmental Prediction, and the Met Office, are collected as raw ensembles. A conditional CSG EMOS (Cond-CSG-EMOS) model is further proposed to calibrate the ensemble forecasts for heavy-precipitation events, where the standard CSG-EMOS is insufficient. The precipitation samples from the training period are divided into two categories, light- and heavy-precipitation events, according to a given precipitation threshold and prior ensemble forecast. Then individual models are, respectively, optimized for adequate parameter estimation. The results demonstrate that the Cond-CSG-EMOS is superior to the raw EPSs and the standard CSG-EMOS, especially for the calibration of heavy-precipitation events. The spatial distribution of forecast skills shows that the Cond-CSG-EMOS outperforms the others over most of the study region, particularly in North and Central China. A sensitivity testing on the precipitation threshold shows that a higher threshold leads to better outcomes for the regions that have more heavy-precipitation events, i.e., South China. Our results indicate that the proposed Cond-CSG-EMOS model is a promising approach for the statistical postprocessing of ensemble precipitation forecasts. Significance Statement Heavy-precipitation events are of highly socioeconomic relevance. But it remains a great challenge to obtain high-quality probabilistic quantitative precipitation forecasting (PQPF) from the operational ensemble prediction systems (EPSs). Statistical postprocessing is commonly used to calibrate the systematic errors of the raw EPSs forecasts. However, the non-Gaussian nature of precipitation and the imbalance between the size of light- and heavy-precipitation samples add to the challenge. This study proposes a conditional postprocessing method to improve PQPF of heavy precipitation by performing calibration separately for light and heavy precipitation, in contrast to some previous studies. Our results indicate that the conditional model mitigates the underestimation of heavy precipitation, as well as with a better calibration for the light- and moderate-precipitation.

Deep-learning-based post-processing for probabilistic precipitation forecasting

Post-processing of numerical precipitation forecast based on spatial-temporal deep learning model

Improving Global Precipitation in Numerical Weather Prediction Systems Based on Deep Learning Techniques

Deep Learning of a 200-member Ensemble with a Limited Historical Training to Improve the Prediction of Extreme Precipitation Events

Conditional Ensemble Model Output Statistics for Postprocessing of Ensemble Precipitation Forecasting

Improving Summer Precipitation Prediction in China Using Deep Learning

Deep Learning Integration of Multi-Model Forecast Precipitation Considering Long Lead Times

Deep learning precipitation prediction models combined with feature analysis

Advancing sub-seasonal to seasonal multi-model ensemble precipitation prediction in east asia: Deep learning-based post-processing for improved accuracy

Improving Monsoon Precipitation Prediction Using Combined Convolutional and Long Short Term Memory Neural Network

Deep Learning for Post-Processing Ensemble Weather Forecasts

Deep Learning for Seasonal Precipitation Prediction over China

Improve short-term precipitation forecasts using numerical weather prediction model output and machine learning

A novel hybrid artificial neural network - Parametric scheme for postprocessing medium-range precipitation forecasts

Short-term precipitation prediction using deep learning

Skillful Extended-Range Forecast of Rainfall and Extreme Events in East China Based on Deep Learning

A new scheme for probabilistic forecasting with an ensemble model based on CEEMDAN and AM-MCMC and its application in precipitation forecasting

Statistical post-processing of precipitation forecasts using circulation classifications and spatiotemporal deep neural networks

Deterministic and probabilistic evaluation of raw and post-processing monthly precipitation forecasts: a case study of China

Deep learning models for generation of precipitation maps based on numerical weather prediction

Precipitation Forecast Based on Multi-channel ConvLSTM and 3D-CNN