Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia

Timely and accurate agricultural impact assessments for droughts are critical for designing appropriate interventions and policy. These assessments are often ad hoc, late, or spatially imprecise, with reporting at the zonal or regional level. This is problematic as we find substantial variability in...

Descripción completa

Detalles Bibliográficos
Autores principales: Mann, Michael L., Malik, Arun S., Warner, James
Formato: Artículo preliminar
Lenguaje:Inglés
Publicado: International Food Policy Research Institute 2018
Materias:
Acceso en línea:https://hdl.handle.net/10568/145591
_version_ 1855523544282169344
author Mann, Michael L.
Malik, Arun S.
Warner, James
author_browse Malik, Arun S.
Mann, Michael L.
Warner, James
author_facet Mann, Michael L.
Malik, Arun S.
Warner, James
author_sort Mann, Michael L.
collection Repository of Agricultural Research Outputs (CGSpace)
description Timely and accurate agricultural impact assessments for droughts are critical for designing appropriate interventions and policy. These assessments are often ad hoc, late, or spatially imprecise, with reporting at the zonal or regional level. This is problematic as we find substantial variability in losses at the village-level that are missing when reporting even at the zonal level. In this paper we propose a new data fusion method combining remotely-sensed data with agricultural survey data that might address these limitations. We apply the method to Ethiopia, which is regularly hit by droughts and is a substantial recipient of ad hoc imported food aid. We then utilize remotely-sensed data obtained near mid-season to predict substantial crop losses of greater than or equal to 25 percent due to drought at the village level for five primary cereal crops. We train machine learning models to predict the likelihood of losses and explore the most influential variables. On independent samples, the models identify substantial drought loss cases with up to 70 percent accuracy by mid- to late-September. We believe the proposed models could be used to help monitor and predict yields for disaster response teams and policy makers, particularly with further development of the models and integration of newly available high resolution remotely-sensed data, such as the Harmonized Landsat Sentinel (HLS) data set.
format Artículo preliminar
id CGSpace145591
institution CGIAR Consortium
language Inglés
publishDate 2018
publishDateRange 2018
publishDateSort 2018
publisher International Food Policy Research Institute
publisherStr International Food Policy Research Institute
record_format dspace
spelling CGSpace1455912025-11-06T06:03:20Z Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia Mann, Michael L. Malik, Arun S. Warner, James weather hazards shock surveys machine learning drought cereal crops crop losses resilience impact assessment Timely and accurate agricultural impact assessments for droughts are critical for designing appropriate interventions and policy. These assessments are often ad hoc, late, or spatially imprecise, with reporting at the zonal or regional level. This is problematic as we find substantial variability in losses at the village-level that are missing when reporting even at the zonal level. In this paper we propose a new data fusion method combining remotely-sensed data with agricultural survey data that might address these limitations. We apply the method to Ethiopia, which is regularly hit by droughts and is a substantial recipient of ad hoc imported food aid. We then utilize remotely-sensed data obtained near mid-season to predict substantial crop losses of greater than or equal to 25 percent due to drought at the village level for five primary cereal crops. We train machine learning models to predict the likelihood of losses and explore the most influential variables. On independent samples, the models identify substantial drought loss cases with up to 70 percent accuracy by mid- to late-September. We believe the proposed models could be used to help monitor and predict yields for disaster response teams and policy makers, particularly with further development of the models and integration of newly available high resolution remotely-sensed data, such as the Harmonized Landsat Sentinel (HLS) data set. 2018-07-25 2024-06-21T09:04:43Z 2024-06-21T09:04:43Z Working Paper https://hdl.handle.net/10568/145591 en Open Access application/pdf International Food Policy Research Institute Ethiopian Development Research Institute Mann, Michael; Warner, James M.; and Malik, Arun S. 2018. Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia. ESSP Working Paper 120. Washington, DC and Addis Ababa, Ethiopia: International Food Policy Research Institute (IFPRI) and Ethiopian Development Research Institute (EDRI). https://hdl.handle.net/10568/145591
spellingShingle weather hazards
shock
surveys
machine learning
drought
cereal crops
crop losses
resilience
impact assessment
Mann, Michael L.
Malik, Arun S.
Warner, James
Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia
title Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia
title_full Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia
title_fullStr Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia
title_full_unstemmed Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia
title_short Predicting high-magnitude, low-frequency crop losses using machine learning: An application to cereal crops in Ethiopia
title_sort predicting high magnitude low frequency crop losses using machine learning an application to cereal crops in ethiopia
topic weather hazards
shock
surveys
machine learning
drought
cereal crops
crop losses
resilience
impact assessment
url https://hdl.handle.net/10568/145591
work_keys_str_mv AT mannmichaell predictinghighmagnitudelowfrequencycroplossesusingmachinelearninganapplicationtocerealcropsinethiopia
AT malikaruns predictinghighmagnitudelowfrequencycroplossesusingmachinelearninganapplicationtocerealcropsinethiopia
AT warnerjames predictinghighmagnitudelowfrequencycroplossesusingmachinelearninganapplicationtocerealcropsinethiopia