Effective training strategies for deep-learning-based precipitation nowcasting and estimation

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 72
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorKo, Jihoonko
dc.contributor.authorLee, Kyuhanko
dc.contributor.authorHwang, Hyunjinko
dc.contributor.authorOh, Seok-Geunko
dc.contributor.authorSon, Seok-Wooko
dc.contributor.authorShin, Kijungko
dc.date.accessioned2022-04-13T06:46:55Z-
dc.date.available2022-04-13T06:46:55Z-
dc.date.created2022-04-05-
dc.date.created2022-04-05-
dc.date.issued2022-04-
dc.identifier.citationCOMPUTERS & GEOSCIENCES, v.161-
dc.identifier.issn0098-3004-
dc.identifier.urihttp://hdl.handle.net/10203/292550-
dc.description.abstractDeep learning has been successfully applied to precipitation nowcasting. In this work, we propose a pre-training scheme and a new loss function for improving deep-learning-based nowcasting. First, we adapt U-Net, a widely used deep-learning model, for the two problems of interest here: precipitation nowcasting and precipitation estimation from radar images. We formulate the former as a classification problem with three precipitation intervals and the latter as a regression problem. For these tasks, we propose to pre-train the model to predict radar images in the near future without requiring ground-truth precipitation, and we also propose the use of a new loss function for fine-tuning to mitigate the class imbalance problem. We demonstrate the effectiveness of our approach using radar images and precipitation datasets collected from South Korea over seven years. It is highlighted that our pre-training scheme and new loss function improve the critical success index (CSI) of nowcasting of heavy rainfall (at least 10 mm/hr) by up to 95.7% and 43.6%, respectively, at a 5-hr lead time. We also demonstrate that our approach reduces the precipitation estimation error by up to 10.7%, compared to the conventional approach, for light rainfall (between 1 and 10 mm/hr). Lastly, we report the sensitivity of our approach to different resolutions and a detailed analysis of four cases of heavy rainfall.-
dc.languageEnglish-
dc.publisherPERGAMON-ELSEVIER SCIENCE LTD-
dc.titleEffective training strategies for deep-learning-based precipitation nowcasting and estimation-
dc.typeArticle-
dc.identifier.wosid000768005400001-
dc.identifier.scopusid2-s2.0-85125012901-
dc.type.rimsART-
dc.citation.volume161-
dc.citation.publicationnameCOMPUTERS & GEOSCIENCES-
dc.identifier.doi10.1016/j.cageo.2022.105072-
dc.contributor.localauthorShin, Kijung-
dc.contributor.nonIdAuthorHwang, Hyunjin-
dc.contributor.nonIdAuthorOh, Seok-Geun-
dc.contributor.nonIdAuthorSon, Seok-Woo-
dc.description.isOpenAccessN-
dc.type.journalArticleArticle-
dc.subject.keywordAuthorPrecipitation nowcasting-
dc.subject.keywordAuthorPrecipitation estimation-
dc.subject.keywordAuthorDeep learning-
dc.subject.keywordAuthorPre-training-
dc.subject.keywordAuthorClass imbalance-
dc.subject.keywordPlusHOURLY ASSIMILATION-
dc.subject.keywordPlusFORECAST CYCLE-
Appears in Collection
AI-Journal Papers(저널논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0