Two deep learning-based bias-correction pathways improve summer precipitation prediction over China

Abstract

As most global climate models suffer from large biases in simulating/predicting summer precipitation over China, it is of great importance to develop suitable bias-correction methods. This study proposes two pathways of bias-correction with deep learning (DL) models incorporated. One is the deterministic pathway (DP), in which the bias correction is directly applied to the precipitation forecasts. The other one, namely the probability pathway (PP), corrects the forecasted precipitation anomalies using a conditional probability method before being added to the observational climatology. These two pathways have been applied to correct the precipitation forecasts based on a global climate model prediction system NUIST-CFS1.0. The application of DL models in the both pathways yield higher resolution of corrected predictions than the uncorrected ones. Both pathways improve summer precipitation predictions at 4-month lead. Moreover, the DP correction shows a better performance in predicting extreme precipitation, while the PP is proficient in correcting the spatial pattern of precipitation anomalies over China. The present results highlight the importance of the application of appropriate correction strategy for different prediction purposes.

Investigating the Performance of CMIP6 Seasonal Precipitation Predictions and a Grid Based Model Heterogeneity Oriented Deep Learning Bias Correction Framework

“…Thanks to its capability to efficiently capture features in the spatial domains as well as image transformation, the application of CNNs has made significant improvement in the field of hydroclimate science in recent years such as predicting El Niño Southern Oscillation (ENSO) and examining extreme weather (Ham et al., 2019; Y. Liu et al., 2016). Besides, the CNNs have also been widely used in the field of bias correction and downscaling of GCMs (Le et al., 2023; Ling et al., 2022; Sha et al., 2020; Wang & Tian, 2022; Wang et al., 2021). However, the CNN approach in literature mostly uses spatial information (latitude and longitude) as convolutional inputs without considering the relative performance of different models (Figure 2a).…”

“…Liu et al, 2016). Besides, the CNNs have also been widely used in the field of bias correction and downscaling of GCMs (Le et al, 2023;Ling et al, 2022;Sha et al, 2020;Wang & Tian, 2022;Wang et al, 2021). However, the CNN approach in literature mostly uses spatial information (latitude and longitude) as convolutional inputs without considering the relative performance of different models (Figure 2a).…”

Significant advancement in subseasonal-to-seasonal summer precipitation ensemble forecast skills in China mainland through an innovative hybrid CSG-UNET method

“…studies (Ling et al 2022), we split the entire dataset into a training set (2001-2012), a validation set (2013-2015; to optimize the model hyperparameters), and an examination set (2016)(2017)(2018)(2019)(2020). To deal with the potential problem of relatively small sample sizes, the data augmentation is utilized to enrich the samples, which has shown promising results in previous studies Perez 2017, Sun et al 2023a).…”

Improving Subseasonal‐To‐Seasonal Prediction of Summer Extreme Precipitation Over Southern China Based on a Deep Learning Method

“…Recently, the data‐driven deep learning methods have been introduced to postprocess the numerical model outputs and to generate more skillful forecasts via grasping nonlinear relationships between data and features with high accuracy and high transferability (Ling et al., 2022; Rasp & Lerch, 2018). Among them, the U‐Net, taking advantages of capturing multi‐scale spatial patterns and reducing both temporal and spatial errors (Sun et al., 2022; Y. Zhu et al., 2022), has been demonstrated effective to improve the S2S probabilistic precipitation forecasts and shows general superiority to convolutional neural networks (Horat & Lerch, 2023; Vitart et al., 2022).…”