YaBeSH Engineering and Technology Library

    • Journals
    • PaperQuest
    • YSE Standards
    • YaBeSH
    • Login
    View Item 
    •   YE&T Library
    • AMS
    • Artificial Intelligence for the Earth Systems
    • View Item
    •   YE&T Library
    • AMS
    • Artificial Intelligence for the Earth Systems
    • View Item
    • All Fields
    • Source Title
    • Year
    • Publisher
    • Title
    • Subject
    • Author
    • DOI
    • ISBN
    Advanced Search
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Archive

    Can We Integrate Spatial Verification Methods into Neural Network Loss Functions for Atmospheric Science?

    Source: Artificial Intelligence for the Earth Systems:;2022:;volume( 001 ):;issue: 004
    Author:
    Ryan Lagerquist
    ,
    Imme Ebert-Uphoff
    DOI: 10.1175/AIES-D-22-0021.1
    Publisher: American Meteorological Society
    Abstract: In the last decade, much work in atmospheric science has focused on spatial verification (SV) methods for gridded prediction, which overcome serious disadvantages of pixelwise verification. However, neural networks (NN) in atmospheric science are almost always trained to optimize pixelwise loss functions, even when ultimately assessed with SV methods. This establishes a disconnect between model verification during versus after training. To address this issue, we develop spatially enhanced loss functions (SELF) and demonstrate their use for a real-world problem: predicting the occurrence of thunderstorms (henceforth, “convection”) with NNs. In each SELF we use either a neighborhood filter, which highlights convection at scales larger than a threshold, or a spectral filter (employing Fourier or wavelet decomposition), which is more flexible and highlights convection at scales between two thresholds. We use these filters to spatially enhance common verification scores, such as the Brier score. We train each NN with a different SELF and compare their performance at many scales of convection, from discrete storm cells to tropical cyclones. Among our many findings are that (i) for a low or high risk threshold, the ideal SELF focuses on small or large scales, respectively; (ii) models trained with a pixelwise loss function perform surprisingly well; and (iii) nevertheless, models trained with a spectral filter produce much better-calibrated probabilities than a pixelwise model. We provide a general guide to using SELFs, including technical challenges and the final Python code, as well as demonstrating their use for the convection problem. To our knowledge this is the most in-depth guide to SELFs in the geosciences.
    • Download: (6.133Mb)
    • Show Full MetaData Hide Full MetaData
    • Item Order
    • Go To Publisher
    • Price: 5000 Rial
    • Statistics

      Can We Integrate Spatial Verification Methods into Neural Network Loss Functions for Atmospheric Science?

    URI
    http://yetl.yabesh.ir/yetl1/handle/yetl/4290401
    Collections
    • Artificial Intelligence for the Earth Systems

    Show full item record

    contributor authorRyan Lagerquist
    contributor authorImme Ebert-Uphoff
    date accessioned2023-04-12T18:52:38Z
    date available2023-04-12T18:52:38Z
    date copyright2022/11/28
    date issued2022
    identifier otherAIES-D-22-0021.1.pdf
    identifier urihttp://yetl.yabesh.ir/yetl1/handle/yetl/4290401
    description abstractIn the last decade, much work in atmospheric science has focused on spatial verification (SV) methods for gridded prediction, which overcome serious disadvantages of pixelwise verification. However, neural networks (NN) in atmospheric science are almost always trained to optimize pixelwise loss functions, even when ultimately assessed with SV methods. This establishes a disconnect between model verification during versus after training. To address this issue, we develop spatially enhanced loss functions (SELF) and demonstrate their use for a real-world problem: predicting the occurrence of thunderstorms (henceforth, “convection”) with NNs. In each SELF we use either a neighborhood filter, which highlights convection at scales larger than a threshold, or a spectral filter (employing Fourier or wavelet decomposition), which is more flexible and highlights convection at scales between two thresholds. We use these filters to spatially enhance common verification scores, such as the Brier score. We train each NN with a different SELF and compare their performance at many scales of convection, from discrete storm cells to tropical cyclones. Among our many findings are that (i) for a low or high risk threshold, the ideal SELF focuses on small or large scales, respectively; (ii) models trained with a pixelwise loss function perform surprisingly well; and (iii) nevertheless, models trained with a spectral filter produce much better-calibrated probabilities than a pixelwise model. We provide a general guide to using SELFs, including technical challenges and the final Python code, as well as demonstrating their use for the convection problem. To our knowledge this is the most in-depth guide to SELFs in the geosciences.
    publisherAmerican Meteorological Society
    titleCan We Integrate Spatial Verification Methods into Neural Network Loss Functions for Atmospheric Science?
    typeJournal Paper
    journal volume1
    journal issue4
    journal titleArtificial Intelligence for the Earth Systems
    identifier doi10.1175/AIES-D-22-0021.1
    treeArtificial Intelligence for the Earth Systems:;2022:;volume( 001 ):;issue: 004
    contenttypeFulltext
    DSpace software copyright © 2002-2015  DuraSpace
    نرم افزار کتابخانه دیجیتال "دی اسپیس" فارسی شده توسط یابش برای کتابخانه های ایرانی | تماس با یابش
    yabeshDSpacePersian
     
    DSpace software copyright © 2002-2015  DuraSpace
    نرم افزار کتابخانه دیجیتال "دی اسپیس" فارسی شده توسط یابش برای کتابخانه های ایرانی | تماس با یابش
    yabeshDSpacePersian