Predicting East Africa’s rainfall extremes with calibrated, hybrid physical and AI systems

Predicting East Africa’s rainfall extremes with calibrated, hybrid physical and AI systems | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Predicting East Africa’s rainfall extremes with calibrated, hybrid physical and AI systems Shruti Nath, David Koros, Fenwick Cooper, David MacLeod, Hannah Kimani, and 17 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8534913/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Accurate and reliable rainfall forecasts are crucial for Early Warning Systems (EWS). Hybrid Artificial Intelligence (AI) approaches, which integrate global physical forecasts with AI, enable low-cost ensemble generation, extending skill to medium-range lead times and offering scalable systems to inform EWS particularly under resource-constrained settings. However, assessing their practical value in representing user-relevant rainfall extremes and guiding decisions for EWS remains critical. We present a low-cost, fine-tuning method that calibrates ensemble forecasts to probabilities of exceedances for rainfall extremes. Applied to physical and hybrid AI rainfall forecasts over East Africa, case-study results show encouraging skill improvement from hybrid AI systems at higher rainfall thresholds. Sub-regional analysis shows the hybrid AI systems better reflecting the observed chances of extremes at longer lead times. Forecast skill decomposed across rainfall and probability thresholds further highlights the ranges of rainfall and risk tolerances where most actionable improvement arises, whether using satellite or rain-gauge data as reference. This tailored evaluation helps guide forecast adoption for EWS. Earth and environmental sciences/Climate sciences/Climate change/Projection and prediction Physical sciences/Mathematics and computing/Statistics Scientific community and society/Developing world Rainfall ML/AI extremes forecast value Early Warning Systems Full Text Additional Declarations There is NO Competing Interest. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8534913","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":574028352,"identity":"e0de2b07-1d59-405f-8723-df682c32448a","order_by":0,"name":"Shruti Nath","email":"data:image/png;base64,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","orcid":"https://orcid.org/0009-0007-4449-1621","institution":"University of Oxford","correspondingAuthor":true,"prefix":"","firstName":"Shruti","middleName":"","lastName":"Nath","suffix":""},{"id":574028353,"identity":"6d3f9537-3fd1-41f3-af58-f0dfb604faef","order_by":1,"name":"David Koros","email":"","orcid":"","institution":"Kenya Meteorological Department","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Koros","suffix":""},{"id":574028354,"identity":"05b1ebcf-b244-4785-862d-e1283416c73c","order_by":2,"name":"Fenwick Cooper","email":"","orcid":"","institution":"University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Fenwick","middleName":"","lastName":"Cooper","suffix":""},{"id":574028355,"identity":"5fab3dfc-eb29-4051-be39-5aad6820e061","order_by":3,"name":"David MacLeod","email":"","orcid":"","institution":"University of Cardiff","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"MacLeod","suffix":""},{"id":574028356,"identity":"74e99b65-7a51-4cf6-a27f-6e5f1d579cd9","order_by":4,"name":"Hannah Kimani","email":"","orcid":"","institution":"Kenya Meteorological Department","correspondingAuthor":false,"prefix":"","firstName":"Hannah","middleName":"","lastName":"Kimani","suffix":""},{"id":574028357,"identity":"0679a031-a913-4070-8dfa-67eb11568907","order_by":5,"name":"Zacharia Mwai","email":"","orcid":"","institution":"Kenya Meteorological Department","correspondingAuthor":false,"prefix":"","firstName":"Zacharia","middleName":"","lastName":"Mwai","suffix":""},{"id":574028358,"identity":"85b3fe67-6292-4651-9541-86b8847ebfb5","order_by":6,"name":"Christine Maswi","email":"","orcid":"","institution":"Kenya Meteorological Department","correspondingAuthor":false,"prefix":"","firstName":"Christine","middleName":"","lastName":"Maswi","suffix":""},{"id":574028359,"identity":"90b0fb9e-d4b4-424a-8bc5-60d0ecfa007a","order_by":7,"name":"Bernard Chanzu","email":"","orcid":"","institution":"Kenya Meteorological Department","correspondingAuthor":false,"prefix":"","firstName":"Bernard","middleName":"","lastName":"Chanzu","suffix":""},{"id":574028360,"identity":"bb6ca854-dd2e-46f1-9ccd-b5f400104a56","order_by":8,"name":"Asaminew Teshome","email":"","orcid":"","institution":"Ethiopian Meteorological Institute","correspondingAuthor":false,"prefix":"","firstName":"Asaminew","middleName":"","lastName":"Teshome","suffix":""},{"id":574028361,"identity":"8ca5af73-8593-4f14-91c5-85b98adbe72c","order_by":9,"name":"Bekalu Tamene","email":"","orcid":"","institution":"Ethiopian Meteorological Institute","correspondingAuthor":false,"prefix":"","firstName":"Bekalu","middleName":"","lastName":"Tamene","suffix":""},{"id":574028362,"identity":"f898f203-2a54-487a-9efb-428c0cb3df75","order_by":10,"name":"Bekele Kebebe","email":"","orcid":"","institution":"Ethiopian Meteorological Institute","correspondingAuthor":false,"prefix":"","firstName":"Bekele","middleName":"","lastName":"Kebebe","suffix":""},{"id":574028363,"identity":"271dfdb1-5874-4fb3-ad29-b80c054c5796","order_by":11,"name":"Samrawit Abebe","email":"","orcid":"","institution":"Ethiopian Meteorological Institute","correspondingAuthor":false,"prefix":"","firstName":"Samrawit","middleName":"","lastName":"Abebe","suffix":""},{"id":574028364,"identity":"fd97e15a-5075-4b5c-86ee-642a631e82f6","order_by":12,"name":"Masilin Gudoshava","email":"","orcid":"","institution":"Climate Prediction and Application Centre, Inter-Governmental Authority on Development (ICPAC).","correspondingAuthor":false,"prefix":"","firstName":"Masilin","middleName":"","lastName":"Gudoshava","suffix":""},{"id":574028365,"identity":"0f0446b4-5943-4a21-b3ed-a28907cb4212","order_by":13,"name":"Ahmed Amdihun","email":"","orcid":"","institution":"Climate Prediction and Application Centre, Inter-Governmental Authority on Development (ICPAC).","correspondingAuthor":false,"prefix":"","firstName":"Ahmed","middleName":"","lastName":"Amdihun","suffix":""},{"id":574028366,"identity":"d7ac2d4d-a334-482d-9b0a-2dd2afb059cc","order_by":14,"name":"Isaac Obai","email":"","orcid":"","institution":"United Nations, World Food Programme","correspondingAuthor":false,"prefix":"","firstName":"Isaac","middleName":"","lastName":"Obai","suffix":""},{"id":574028367,"identity":"45b452ae-38e4-4cca-83f9-34a0ad32e41e","order_by":15,"name":"Maurine Ambani","email":"","orcid":"","institution":"United Nations, World Food Programme","correspondingAuthor":false,"prefix":"","firstName":"Maurine","middleName":"","lastName":"Ambani","suffix":""},{"id":574028368,"identity":"8d7296d0-2df3-4fbb-bc30-16dffdfbe043","order_by":16,"name":"Mark Arango","email":"","orcid":"","institution":"United Nations, World Food Programme","correspondingAuthor":false,"prefix":"","firstName":"Mark","middleName":"","lastName":"Arango","suffix":""},{"id":574028369,"identity":"0e60f1cd-1cf9-4ba2-8f56-e76e8bd375d2","order_by":17,"name":"Jesse Mason","email":"","orcid":"","institution":"United Nations, World Food Programme","correspondingAuthor":false,"prefix":"","firstName":"Jesse","middleName":"","lastName":"Mason","suffix":""},{"id":574028370,"identity":"c5946cad-af13-4fce-abdf-25f7a33a4672","order_by":18,"name":"Matthew Chantry","email":"","orcid":"https://orcid.org/0000-0002-1132-0961","institution":"ECMWF","correspondingAuthor":false,"prefix":"","firstName":"Matthew","middleName":"","lastName":"Chantry","suffix":""},{"id":574028371,"identity":"3e1cf4af-3d5f-4e08-9075-9021e37e0562","order_by":19,"name":"Florian Pappenberger","email":"","orcid":"https://orcid.org/0000-0003-1766-2898","institution":"ECMWF","correspondingAuthor":false,"prefix":"","firstName":"Florian","middleName":"","lastName":"Pappenberger","suffix":""},{"id":574028372,"identity":"25276b42-0dfd-46c3-bd59-1db9b77b21db","order_by":20,"name":"Antje Weisheimer","email":"","orcid":"https://orcid.org/0000-0002-7231-6974","institution":"University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Antje","middleName":"","lastName":"Weisheimer","suffix":""},{"id":574028373,"identity":"7bdcdbe1-26cc-45c9-a183-518fb89691ae","order_by":21,"name":"Tim Palmer","email":"","orcid":"https://orcid.org/0000-0002-7121-2196","institution":"University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Tim","middleName":"","lastName":"Palmer","suffix":""}],"badges":[],"createdAt":"2026-01-06 20:55:33","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8534913/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8534913/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100418194,"identity":"88da678a-6590-4d2e-bd63-33f926eb9fdb","added_by":"auto","created_at":"2026-01-16 13:25:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":18302822,"visible":true,"origin":"","legend":"","description":"","filename":"SEWAAenslogreg.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/9b65e081be6ab8c17e8e2e3d.pdf"},{"id":100418959,"identity":"6dffca74-9dbc-444d-869b-475fd30d59b7","added_by":"auto","created_at":"2026-01-16 13:26:27","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13980944,"visible":true,"origin":"","legend":"","description":"","filename":"CatofrelGHAF.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/47d060331ea3796d91f6f0d2.pdf"},{"id":100417794,"identity":"d6c3ead6-cc45-4d5b-99d9-d6b2f64e015c","added_by":"auto","created_at":"2026-01-16 13:24:44","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":94252,"visible":true,"origin":"","legend":"","description":"","filename":"Costlosscasestudies.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/f125082ae0d86c2d126a253a.pdf"},{"id":100418120,"identity":"96433c12-ac10-4ed3-9db1-becc2fbda06e","added_by":"auto","created_at":"2026-01-16 13:25:09","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":76116,"visible":true,"origin":"","legend":"","description":"","filename":"Costlossraingauge.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/660396f317613fa355b66aa8.pdf"},{"id":100418187,"identity":"9c772460-b802-4721-b4f1-566c0277c56d","added_by":"auto","created_at":"2026-01-16 13:25:15","extension":"pdf","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":370093,"visible":true,"origin":"","legend":"","description":"","filename":"ELRframework.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/1055b5766ded69ec75269779.pdf"},{"id":100418327,"identity":"2fb9eae9-689c-464c-b2bd-a2c7559dc7e2","added_by":"auto","created_at":"2026-01-16 13:25:40","extension":"pdf","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1138062,"visible":true,"origin":"","legend":"","description":"","filename":"casestudies.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/ec02f3a9f14c2867db457757.pdf"},{"id":100418028,"identity":"13693406-7528-4787-8025-afa316405bf1","added_by":"auto","created_at":"2026-01-16 13:24:58","extension":"pdf","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49790,"visible":true,"origin":"","legend":"","description":"","filename":"casestudycatofreliability.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/a22709be5d62645bac69fac8.pdf"},{"id":100418951,"identity":"3fd8b378-224a-4901-819f-b27b4b2d745d","added_by":"auto","created_at":"2026-01-16 13:26:27","extension":"pdf","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":42842,"visible":true,"origin":"","legend":"","description":"","filename":"casestudylogregensmean.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/a31ad28102e58ab0256c8b0a.pdf"},{"id":100418718,"identity":"404ab8d9-9265-4896-a007-317bb4d2132f","added_by":"auto","created_at":"2026-01-16 13:26:11","extension":"pdf","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40260,"visible":true,"origin":"","legend":"","description":"","filename":"casestudylogregimp.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/ce2c3f76cd807f0aa9f4ddfb.pdf"},{"id":100418492,"identity":"12fb080f-3580-4986-8eed-3b51f055df27","added_by":"auto","created_at":"2026-01-16 13:25:55","extension":"pdf","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2364644,"visible":true,"origin":"","legend":"","description":"","filename":"climatezones.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/2a57ee31d1fbfa02f49980c8.pdf"},{"id":100418355,"identity":"b82603ce-64fa-42c7-904c-bc78b6ca1620","added_by":"auto","created_at":"2026-01-16 13:25:42","extension":"pdf","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27394,"visible":true,"origin":"","legend":"","description":"","filename":"raingaugelogregimp.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/200ec965b29e7aa6cc4a5194.pdf"},{"id":100418189,"identity":"a6bc9c81-e014-463c-8141-c527a7b46d15","added_by":"auto","created_at":"2026-01-16 13:25:15","extension":"pdf","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":50771,"visible":true,"origin":"","legend":"","description":"","filename":"regionloglossimp.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/0d0dbc4a69e25504a7f5e370.pdf"},{"id":100418296,"identity":"467eac60-1aa3-42e7-8850-3c8f556cd98e","added_by":"auto","created_at":"2026-01-16 13:25:39","extension":"json","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20451,"visible":true,"origin":"","legend":"","description":"","filename":"NCOMMS26001302.json","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1/1d269eebcd64694af65cbb47.json"},{"id":102297335,"identity":"23d87dae-838f-4d80-b14b-3d3bb529f340","added_by":"auto","created_at":"2026-02-10 10:27:01","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":9840378,"visible":true,"origin":"","legend":"Article File","description":"","filename":"SEWAAenslogreg.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8534913/v1_covered_b32ee8f3-1e84-4cbf-b12a-cfb158d6f1a4.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Predicting East Africa’s rainfall extremes with calibrated, hybrid physical and AI systems","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Rainfall, ML/AI, extremes, forecast value, Early Warning Systems","lastPublishedDoi":"10.21203/rs.3.rs-8534913/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8534913/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Accurate and reliable rainfall forecasts are crucial for Early Warning Systems (EWS). Hybrid Artificial Intelligence (AI) approaches, which integrate global physical forecasts with AI, enable low-cost ensemble generation, extending skill to medium-range lead times and offering scalable systems to inform EWS particularly under resource-constrained settings. However, assessing their practical value in representing user-relevant rainfall extremes and guiding decisions for EWS remains critical. We present a low-cost, fine-tuning method that calibrates ensemble forecasts to probabilities of exceedances for rainfall extremes. Applied to physical and hybrid AI rainfall forecasts over East Africa, case-study results show encouraging skill improvement from hybrid AI systems at higher rainfall thresholds. Sub-regional analysis shows the hybrid AI systems better reflecting the observed chances of extremes at longer lead times. Forecast skill decomposed across rainfall and probability thresholds further highlights the ranges of rainfall and risk tolerances where most actionable improvement arises, whether using satellite or rain-gauge data as reference. This tailored evaluation helps guide forecast adoption for EWS.","manuscriptTitle":"Predicting East Africa’s rainfall extremes with calibrated, hybrid physical and AI systems","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-16 13:07:55","doi":"10.21203/rs.3.rs-8534913/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"24dc1c80-459d-4eec-8a7a-96a5e602777c","owner":[],"postedDate":"January 16th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":61080386,"name":"Earth and environmental sciences/Climate sciences/Climate change/Projection and prediction"},{"id":61080387,"name":"Physical sciences/Mathematics and computing/Statistics"},{"id":61080388,"name":"Scientific community and society/Developing world"}],"tags":[],"updatedAt":"2026-02-09T15:01:11+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-16 13:07:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8534913","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8534913","identity":"rs-8534913","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}