{"title":"卫星观测时代以后全球冰雪覆盖回顾制图","authors":"Kingsley K. Kumah, Omid Zandi, Ali Behrangi","doi":"10.1029/2024EA004171","DOIUrl":null,"url":null,"abstract":"<p>Monitoring Earth's snow and ice cover is essential for diverse applications, including climate studies, hydrological forecasting, and precipitation mapping. This study develops and evaluates methodologies to extend the Global Multisensor Automated Snow and Ice Mapping System (GMASI) records prior to its July 1987 inception, reconstructing high-resolution global snow and ice cover data. Using ERA5 reanalysis variables, three machine learning (ML) approaches—ML-E (ML with ERA5 predictors only), ML-EC (ML with ERA5 and Climatology-based predictors), and ML-ECC (ML with ERA5 predictors, Climatology-based predictors, and additional Consistency Checks)—were tested alongside climatological and fractional cover-based methods. Validation against GMASI (1988–1991) shows that ML-EC and ML-ECC achieve superior alignment, with the latter offering marginal accuracy gains. Both methods demonstrated stable daily estimates, with mean percentage biases for snow and ice cover remaining below 3% during validation. Their high accuracy is further reflected in probabilities of detection (POD) exceeding 97% across key surface types. Across all methods, there was a general tendency to underestimate snow-free areas and overestimate snow-covered regions in the Northern Hemisphere, while classification challenges in the Southern Hemisphere were more pronounced over snow-free land and Antarctic sea ice. The ML-EC approach was subsequently applied to extend the GMASI record back to 1980, capturing seasonal and interannual variability consistent with GMASI-era trends. These results underscore the potential of ML techniques to extend snow and ice cover records as far back as the beginning of the reanalysis era (1940–present), providing invaluable insights for climate analysis and operational applications.</p>","PeriodicalId":54286,"journal":{"name":"Earth and Space Science","volume":"12 5","pages":""},"PeriodicalIF":2.9000,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1029/2024EA004171","citationCount":"0","resultStr":"{\"title\":\"Retrospective Mapping of Global Snow and Ice Cover Beyond the Satellite Observational Era\",\"authors\":\"Kingsley K. Kumah, Omid Zandi, Ali Behrangi\",\"doi\":\"10.1029/2024EA004171\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>Monitoring Earth's snow and ice cover is essential for diverse applications, including climate studies, hydrological forecasting, and precipitation mapping. This study develops and evaluates methodologies to extend the Global Multisensor Automated Snow and Ice Mapping System (GMASI) records prior to its July 1987 inception, reconstructing high-resolution global snow and ice cover data. Using ERA5 reanalysis variables, three machine learning (ML) approaches—ML-E (ML with ERA5 predictors only), ML-EC (ML with ERA5 and Climatology-based predictors), and ML-ECC (ML with ERA5 predictors, Climatology-based predictors, and additional Consistency Checks)—were tested alongside climatological and fractional cover-based methods. Validation against GMASI (1988–1991) shows that ML-EC and ML-ECC achieve superior alignment, with the latter offering marginal accuracy gains. Both methods demonstrated stable daily estimates, with mean percentage biases for snow and ice cover remaining below 3% during validation. Their high accuracy is further reflected in probabilities of detection (POD) exceeding 97% across key surface types. Across all methods, there was a general tendency to underestimate snow-free areas and overestimate snow-covered regions in the Northern Hemisphere, while classification challenges in the Southern Hemisphere were more pronounced over snow-free land and Antarctic sea ice. The ML-EC approach was subsequently applied to extend the GMASI record back to 1980, capturing seasonal and interannual variability consistent with GMASI-era trends. These results underscore the potential of ML techniques to extend snow and ice cover records as far back as the beginning of the reanalysis era (1940–present), providing invaluable insights for climate analysis and operational applications.</p>\",\"PeriodicalId\":54286,\"journal\":{\"name\":\"Earth and Space Science\",\"volume\":\"12 5\",\"pages\":\"\"},\"PeriodicalIF\":2.9000,\"publicationDate\":\"2025-05-26\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://onlinelibrary.wiley.com/doi/epdf/10.1029/2024EA004171\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Earth and Space Science\",\"FirstCategoryId\":\"89\",\"ListUrlMain\":\"https://onlinelibrary.wiley.com/doi/10.1029/2024EA004171\",\"RegionNum\":3,\"RegionCategory\":\"地球科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ASTRONOMY & ASTROPHYSICS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Earth and Space Science","FirstCategoryId":"89","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1029/2024EA004171","RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ASTRONOMY & ASTROPHYSICS","Score":null,"Total":0}
Retrospective Mapping of Global Snow and Ice Cover Beyond the Satellite Observational Era
Monitoring Earth's snow and ice cover is essential for diverse applications, including climate studies, hydrological forecasting, and precipitation mapping. This study develops and evaluates methodologies to extend the Global Multisensor Automated Snow and Ice Mapping System (GMASI) records prior to its July 1987 inception, reconstructing high-resolution global snow and ice cover data. Using ERA5 reanalysis variables, three machine learning (ML) approaches—ML-E (ML with ERA5 predictors only), ML-EC (ML with ERA5 and Climatology-based predictors), and ML-ECC (ML with ERA5 predictors, Climatology-based predictors, and additional Consistency Checks)—were tested alongside climatological and fractional cover-based methods. Validation against GMASI (1988–1991) shows that ML-EC and ML-ECC achieve superior alignment, with the latter offering marginal accuracy gains. Both methods demonstrated stable daily estimates, with mean percentage biases for snow and ice cover remaining below 3% during validation. Their high accuracy is further reflected in probabilities of detection (POD) exceeding 97% across key surface types. Across all methods, there was a general tendency to underestimate snow-free areas and overestimate snow-covered regions in the Northern Hemisphere, while classification challenges in the Southern Hemisphere were more pronounced over snow-free land and Antarctic sea ice. The ML-EC approach was subsequently applied to extend the GMASI record back to 1980, capturing seasonal and interannual variability consistent with GMASI-era trends. These results underscore the potential of ML techniques to extend snow and ice cover records as far back as the beginning of the reanalysis era (1940–present), providing invaluable insights for climate analysis and operational applications.
期刊介绍:
Marking AGU’s second new open access journal in the last 12 months, Earth and Space Science is the only journal that reflects the expansive range of science represented by AGU’s 62,000 members, including all of the Earth, planetary, and space sciences, and related fields in environmental science, geoengineering, space engineering, and biogeochemistry.