[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117438-en":3,"doc-seo-117438-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},117438,13056703019404,"Miles","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","INTERPRETABLE MACHINE LEARNING FOR EXTREME EVENTS DETECTION - AN APPLICATION TO DROUGHTS IN THE PO RIVER BASIN","Increasing drought frequency and intensity represent one of the most serious climate-change impacts, making monitoring and early detection essential for reducing societal risk. Conventional drought indices often miss impacts because they rely on single precursors rather than capturing their interactions. This study develops a data-driven, impact-based drought index targeting the Vegetation Health Index derived from satellites. Interpretable dimensionality reduction aggregates drivers from precipitation, temperature, snow, and lakes; filter-based feature selection removes redundancy, and linear supervised learning fits multi-task, low-sample models. Experiments cover ten Po River sub-basins while targeting scalable workflow design.","INTERPRETABLE MACHINE LEARNING FOR EXTREME EVENTS DETECTION: AN APPLICATION TO DROUGHTSIN THE PO RIVER BASIN  \nPaolo Bonetti∗  \nPolitecnico di Milano  \nMatteo Giuliani  \nPolitecnico di Milano  \nVeronica Cardigliano  \nPolitecnico di Milano  \nAlberto Maria Metelli  \nPolitecnico di Milano  \nMarcello Restelli  \nPolitecnico di Milano  \nAndrea Castelletti  \nPolitecnico di Milano  \nABSTRACT  \nThe increasing frequency and intensity of drought events—periods of significant decrease in water availability—are among the most alarming impacts of climate change. Monitoring and detecting these events is essential to mitigate their impact on our society. However, traditional drought indices often fail to accurately detect such impacts as they mostly focus on single precursors. In this study, we leverage machine learning algorithms to define a novel data-driven, impact-based drought index reproducing as target the Vegetation Health Index, a satellite signal that directly assesses the vegetation status. We first apply novel dimensionality reduction methods that allow for interpretable spatial aggregation of features related to precipitation, temperature, snow, and lakes. Then, we select the most informative and non-redundant features through filter feature selection. Finally, linear supervised learning methods are considered, given the small number of samples and the aim of preserving interpretability. The experimental setting focuses on ten sub-basins of the Po River basin, but the aim is to design a machine learning-based workflow applicable on a large scale.  \n1 INTRODUCTION  \nIn recent years, climate change has gained increasing attention both in the media and within the scientific community (Parmesan et al., 2022) . An alarming consequence of climate change and global warming is the rising frequency and intensity of droughts (Spinoni et al., 2016), defined as periods of aridity and scarcity of water compared to normal conditions (Van Loon & Van Lanen, 2012), which can have severe economic and ecological effects, impacting agriculture, water resources, tourism, ecosystems, especially in regions lacking effective mitigation and drought management plans (Dai, 2011) . In this scenario, the identification of drought-prone conditions and early detection are crucial to mitigate their consequences.  \nTraditional standardized drought indices (e.g., SPI (Guttman, 1999), SPEI (Vicente-Serrano et al., 2010)) are widely used to monitor drought conditions. However, they often fail to detect drought impacts as they focus on specific drivers (e.g., precipitation and evapotranspiration) without accounting for their complex interactions, that eventually generate the drought’s impacts. Ad hoc index formulations have also been designed to identify drought conditions in specific basins (e.g., Estrela & Vargas (2012)) . However, these indices are basin-specific, with no capability to generalize and no automatization of the process, requiring years of expert refinements. As an alternative, indices derived from satellite data are available (e.g., NDVI (Pettorelli, 2013), VHI (Bento et al., 2018)) and allow the direct observation of vegetation status in terms of colors, making them more suitable proxies of its condition. Yet, these indices can only be observed in real-time, preventing their prediction weeks to months in advance (Meehl et al., 2021) to timely prompt anticipatory operations.  \n∗[paolo.bonetti@polimi.it](paolo.bonetti@polimi.it)  \nRecent works applied Machine Learning to automatically derive data-driven drought indices, exploiting classical indices as proxies of drought conditions (e.g.,(Zaniolo et al., 2018; Feng et al., 2019; Dikshit et al., 2021)) . Building on these works, we apply a similar data-driven workflow in the Po River basin (Italy), which is the largest Italian catchment and the most populated area of the country. This region plays a central role in the national economy, and it is one of the largest agricultural areas in Europe. In this","cbCaiqomCEcXlumM","https://ap.wps.com/l/cbCaiqomCEcXlumM","pdf",2355859,1,11,"English","en",105,"# Introduction\n## Limits of traditional drought indices\n## Satellite-based proxies and predictive needs\n# Data\n## Vegetation Health Index as target\n## Study area and sub-basin setup\n# Methodology\n## Interpretable dimensionality reduction\n## Filter feature selection using conditional mutual information\n## Linear supervised (multi-task) modeling\n# Experimental setup and results\n## Ten sub-basins and scalability goal","[{\"question\":\"Why are traditional drought indices insufficient for impact detection?\",\"answer\":\"They primarily track specific drivers such as precipitation and evapotranspiration and do not model the complex interactions that produce drought impacts.\"},{\"question\":\"What is the core idea of the proposed drought index?\",\"answer\":\"It uses machine learning to build a data-driven, impact-based index that reproduces the satellite-derived Vegetation Health Index (VHI), linking drought conditions to observed vegetation status.\"},{\"question\":\"How does the workflow keep the model interpretable?\",\"answer\":\"It applies interpretable dimensionality reduction for spatial aggregation of drivers, then uses filter feature selection via conditional mutual information to keep relevant, non-redundant features before training linear supervised learning models.\"}]","INTERPRETABLE MACHINE LEARNING FOR EXTREME EVENTS DETECTION - AN APPLICATION TO DROUGHTS IN THE PO RIVER BASIN | PDF",1785675881,28,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"interpretable-machine-learning-for-extreme-events-detection-an-application-to-droughts-in-the-po-river-basin","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/interpretable-machine-learning-for-extreme-events-detection-an-application-to-droughts-in-the-po-river-basin/117438/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why are traditional drought indices insufficient for impact detection?","Question",{"text":75,"@type":76},"They primarily track specific drivers such as precipitation and evapotranspiration and do not model the complex interactions that produce drought impacts.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the core idea of the proposed drought index?",{"text":80,"@type":76},"It uses machine learning to build a data-driven, impact-based index that reproduces the satellite-derived Vegetation Health Index (VHI), linking drought conditions to observed vegetation status.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the workflow keep the model interpretable?",{"text":84,"@type":76},"It applies interpretable dimensionality reduction for spatial aggregation of drivers, then uses filter feature selection via conditional mutual information to keep relevant, non-redundant features before training linear supervised learning models.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]