Detection and classification of rainfall in South America using satellite images and machine learning techniques

IF 1.2 Q3 PHYSICS, MULTIDISCIPLINARY

Papers in Physics Pub Date : 2023-12-22 DOI:10.4279/pip.150006

Federico Andelsman, Sergio Masuelli, Francisco Tamarit

{"title":"Detection and classification of rainfall in South America using satellite images and machine learning techniques","authors":"Federico Andelsman, Sergio Masuelli, Francisco Tamarit","doi":"10.4279/pip.150006","DOIUrl":null,"url":null,"abstract":"The study of precipitation is one of the most intriguing areas in atmospheric sciences, with significant implications for our daily lives and climate change projections. This paper explores the estimation of rainfall trends in South American regions using convolutional neural networks (CNNs). The study focuses on the application of Cloud-Net, a CNN-based model with a format similar to an autoencoder, to obtain qualitative estimates of precipitation patterns. The employed loss functions, Categorical Cross Entropy and Categorical Focal Loss, address the challenges of classifying minority categories in unbalanced data. Regional analysis was conducted, identifying days with high rainfall intensity and the predominant intensities in 25 regions. The CNN model’s performance was compared with the XGBoost algorithm, showing excellent results for extreme rainfall categories and challenging intermediate categories. Furthermore, a comparison was made with Quantitative Precipitation Estimation (QPE) data and ground measurements from rain gauges. While the CNN model provided a valuable qualitative estimate of precipitation trends, achieving precise quantitative estimation would require an extensive data set of in-situ measurements. Overall, this research demonstrates the potential of CNNs for estimating rainfall trends and understanding precipitation patterns in South American regions. The findings offer valuable insights for further applications in meteorology and environmental studies.","PeriodicalId":19791,"journal":{"name":"Papers in Physics","volume":null,"pages":null},"PeriodicalIF":1.2000,"publicationDate":"2023-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Papers in Physics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.4279/pip.150006","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"PHYSICS, MULTIDISCIPLINARY","Score":null,"Total":0}

引用次数: 0

Abstract

The study of precipitation is one of the most intriguing areas in atmospheric sciences, with significant implications for our daily lives and climate change projections. This paper explores the estimation of rainfall trends in South American regions using convolutional neural networks (CNNs). The study focuses on the application of Cloud-Net, a CNN-based model with a format similar to an autoencoder, to obtain qualitative estimates of precipitation patterns. The employed loss functions, Categorical Cross Entropy and Categorical Focal Loss, address the challenges of classifying minority categories in unbalanced data. Regional analysis was conducted, identifying days with high rainfall intensity and the predominant intensities in 25 regions. The CNN model’s performance was compared with the XGBoost algorithm, showing excellent results for extreme rainfall categories and challenging intermediate categories. Furthermore, a comparison was made with Quantitative Precipitation Estimation (QPE) data and ground measurements from rain gauges. While the CNN model provided a valuable qualitative estimate of precipitation trends, achieving precise quantitative estimation would require an extensive data set of in-situ measurements. Overall, this research demonstrates the potential of CNNs for estimating rainfall trends and understanding precipitation patterns in South American regions. The findings offer valuable insights for further applications in meteorology and environmental studies.

查看原文本刊更多论文

利用卫星图像和机器学习技术对南美洲降雨量进行检测和分类

降水研究是大气科学中最引人入胜的领域之一，对我们的日常生活和气候变化预测有着重要影响。本文探讨了利用卷积神经网络（CNN）估算南美地区降水趋势的问题。研究重点是应用 Cloud-Net（一种基于 CNN 的模型，其格式类似于自动编码器）来获得降水模式的定性估计。所采用的损失函数--分类交叉熵和分类焦点损失--解决了在不平衡数据中对少数类别进行分类的难题。进行了区域分析，确定了 25 个地区的高降雨强度天数和主要强度。CNN 模型的性能与 XGBoost 算法进行了比较，结果表明，在极端降雨类别和具有挑战性的中间类别中，CNN 模型的性能表现出色。此外，还与定量降水估算（QPE）数据和雨量计的地面测量数据进行了比较。虽然 CNN 模型对降水趋势进行了有价值的定性估算，但要实现精确的定量估算，还需要大量的现场测量数据。总之，这项研究证明了 CNN 在估计降水趋势和了解南美地区降水模式方面的潜力。研究结果为进一步应用于气象学和环境研究提供了宝贵的见解。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Papers in Physics PHYSICS, MULTIDISCIPLINARY-

CiteScore

1.90

自引率

0.00%

发文量

期刊介绍： Papers in Physics publishes original research in all areas of physics and its interface with other subjects. The scope includes, but is not limited to, physics of particles and fields, condensed matter, relativity and gravitation, nuclear physics, physics of fluids, biophysics, econophysics, chemical physics, statistical mechanics, soft condensed matter, materials science, mathematical physics and general physics. Contributions in the areas of foundations of physics, history of physics and physics education are not considered for publication. Articles published in Papers in Physics contain substantial new results and ideas that advance the state of physics in a non-trivial way. Articles are strictly reviewed by specialists prior to publication. Papers in Physics highlights outstanding articles published in the journal through the Editors'' choice section. Papers in Physics offers two distinct editorial treatments to articles from which authors can choose. In Traditional Review, manuscripts are submitted to anonymous reviewers seeking constructive criticism and editors make a decision on whether publication is appropriate. In Open Review, manuscripts are sent to reviewers. If the paper is considered original and technically sound, the article, the reviewer''s comments and the author''s reply are published alongside the names of all involved. This way, Papers in Physics promotes the open discussion of controversies among specialists that are of help to the reader and to the transparency of the editorial process. Moreover, our reviewers receive their due recognition by publishing a recorded citable report. Papers in Physics publishes Commentaries from the reviewer(s) if major disagreements remain after exchange with the authors or if a different insight proposed is considered valuable for the readers.