Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfall

The prediction of the Indian summer monsoon rainfall (ISMR) in the June–September (JJAS) season at long-lead times is challenging. The state-of-the-art dynamical models often fail to capture the sign and amplitude of the rainfall anomalies in the extreme rainfall seasons, limiting the overall skill...

Full description

Saved in:
Bibliographic Details
Main Authors: Kalpesh R. Patil, Takeshi Doi, J.V. Ratnam, Swadhin K. Behera
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Applied Computing and Geosciences
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2590197425000394
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849689219092447232
author Kalpesh R. Patil
Takeshi Doi
J.V. Ratnam
Swadhin K. Behera
author_facet Kalpesh R. Patil
Takeshi Doi
J.V. Ratnam
Swadhin K. Behera
author_sort Kalpesh R. Patil
collection DOAJ
description The prediction of the Indian summer monsoon rainfall (ISMR) in the June–September (JJAS) season at long-lead times is challenging. The state-of-the-art dynamical models often fail to capture the sign and amplitude of the rainfall anomalies in the extreme rainfall seasons, limiting the overall skill of the models. We attempted to address this issue using a deep learning model based on convolutional neural networks (CNN). An ensemble of JJAS rainfall predictions using the CNN model with a unique custom function showed high skills in predicting ISMR at a long-lead time of 12 months. The predictions had an anomaly correlation coefficient (ACC) exceeding 0.5 at all the lead times from 2 to 17 months. The CNN model predictions could capture the sign and phase of the extreme rainfall events in the study period realistically. Analysis of saliency-based heatmaps indicated the high skill to be due to the model capturing the leading modes of climate variability, such as the Indian Ocean Dipole and El Niño-Southern Oscillation, realistically. The ensemble of CNN ISMR predictions can supplement the predictions of the forecasting centers.
format Article
id doaj-art-233caf13f7cf46638c087f9ec5fdf63c
institution DOAJ
issn 2590-1974
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Applied Computing and Geosciences
spelling doaj-art-233caf13f7cf46638c087f9ec5fdf63c2025-08-20T03:21:43ZengElsevierApplied Computing and Geosciences2590-19742025-06-012610025710.1016/j.acags.2025.100257Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfallKalpesh R. Patil0Takeshi Doi1J.V. Ratnam2Swadhin K. Behera3Corresponding author. 3173-25 Showa-machi, Kanazawa-ku, Yokohama, Kanagawa, 236-0001, Japan.; Application Laboratory, VAiG, Japan Agency for Marine-Earth Science and Technology, Yokohama, JapanApplication Laboratory, VAiG, Japan Agency for Marine-Earth Science and Technology, Yokohama, JapanApplication Laboratory, VAiG, Japan Agency for Marine-Earth Science and Technology, Yokohama, JapanApplication Laboratory, VAiG, Japan Agency for Marine-Earth Science and Technology, Yokohama, JapanThe prediction of the Indian summer monsoon rainfall (ISMR) in the June–September (JJAS) season at long-lead times is challenging. The state-of-the-art dynamical models often fail to capture the sign and amplitude of the rainfall anomalies in the extreme rainfall seasons, limiting the overall skill of the models. We attempted to address this issue using a deep learning model based on convolutional neural networks (CNN). An ensemble of JJAS rainfall predictions using the CNN model with a unique custom function showed high skills in predicting ISMR at a long-lead time of 12 months. The predictions had an anomaly correlation coefficient (ACC) exceeding 0.5 at all the lead times from 2 to 17 months. The CNN model predictions could capture the sign and phase of the extreme rainfall events in the study period realistically. Analysis of saliency-based heatmaps indicated the high skill to be due to the model capturing the leading modes of climate variability, such as the Indian Ocean Dipole and El Niño-Southern Oscillation, realistically. The ensemble of CNN ISMR predictions can supplement the predictions of the forecasting centers.http://www.sciencedirect.com/science/article/pii/S2590197425000394Rainfed agricultureIndia meteorological departmentCost function
spellingShingle Kalpesh R. Patil
Takeshi Doi
J.V. Ratnam
Swadhin K. Behera
Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfall
Applied Computing and Geosciences
Rainfed agriculture
India meteorological department
Cost function
title Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfall
title_full Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfall
title_fullStr Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfall
title_full_unstemmed Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfall
title_short Enhancing Indian summer monsoon prediction: Deep learning approach for skillful long-lead forecasts of rainfall
title_sort enhancing indian summer monsoon prediction deep learning approach for skillful long lead forecasts of rainfall
topic Rainfed agriculture
India meteorological department
Cost function
url http://www.sciencedirect.com/science/article/pii/S2590197425000394
work_keys_str_mv AT kalpeshrpatil enhancingindiansummermonsoonpredictiondeeplearningapproachforskillfullongleadforecastsofrainfall
AT takeshidoi enhancingindiansummermonsoonpredictiondeeplearningapproachforskillfullongleadforecastsofrainfall
AT jvratnam enhancingindiansummermonsoonpredictiondeeplearningapproachforskillfullongleadforecastsofrainfall
AT swadhinkbehera enhancingindiansummermonsoonpredictiondeeplearningapproachforskillfullongleadforecastsofrainfall