Assessing the Robustness of Deep Learning Streamflow Models Under Climate Change

Simple item page
dc.contributorNearing, Grey S.
dc.contributorZhang, Bo
dc.contributorZhang, Yong
dc.contributor.advisorTick, Geoffrey R.
dc.contributor.authorQualls, Logan Michelle
dc.contributor.otherUniversity of Alabama Tuscaloosa
dc.date.accessioned2022-09-28T14:55:44Z
dc.date.available2022-09-28T14:55:44Z
dc.date.issued2022
dc.descriptionElectronic Thesis or Dissertationen_US
dc.description.abstractLong Short-Term Memory networks provide the most accurate rainfall-runoff predictions to-date, but their reliability under climate change is not well understood. We explore the robustness of these models under climate nonstationarity by creating train and test data splits thatare designed to simulate climate bias. By training on forcing data from hydrological years of high (low) aridity and testing on data from hydrological years of low (high) aridity, we can begin to quantify the performance and relative robustness of that performance under climate nonstationarity. We benchmark against a calibrated conceptual model (the Sacramento Soil Moisture Accounting model) and a calibrated process-based model (the NOAA National WaterModel), and found that LSTMs were generally more accurate than both, even when trained on climatologically biased data splits. The process-based model did not show as large of a performance gap as the conceptual and deep learning models, however (i) this model was not calibrated on a climate-biased data split and (ii) LSTMs always out-performed the process-based benchmark, even when the LSTM training data had climatological bias. We find that although all hydrologic models reported here degrade under nonstationarity, DL models demonstrate greater robustness. We also tested the hypothesis that dynamic climate attributes as inputs into the LSTM would improve performance under climate nonstationarity. We found no predictive value with the addition of dynamic, as opposed to static, climate attribute inputs.en_US
dc.format.mediumelectronic
dc.format.mimetypeapplication/pdf
dc.identifier.otherhttp://purl.lib.ua.edu/186575
dc.identifier.otheru0015_0000001_0004534
dc.identifier.otherQualls_alatus_0004M_14923
dc.identifier.urihttps://ir.ua.edu/handle/123456789/9561
dc.languageEnglish
dc.language.isoen_US
dc.publisherUniversity of Alabama Libraries
dc.relation.hasversionborn digital
dc.relation.ispartofThe University of Alabama Electronic Theses and Dissertations
dc.relation.ispartofThe University of Alabama Libraries Digital Collections
dc.rightsAll rights reserved by the author unless otherwise indicated.en_US
dc.subjectClimate change
dc.subjectDeep learning
dc.subjectDL
dc.subjectHydrology
dc.subjectLong Short-Term Memory
dc.subjectLSTM
dc.titleAssessing the Robustness of Deep Learning Streamflow Models Under Climate Changeen_US
dc.typethesis
dc.typetext
etdms.degree.departmentUniversity of Alabama. Department of Geological Sciences
etdms.degree.disciplineGeology
etdms.degree.grantorThe University of Alabama
etdms.degree.levelmaster's
etdms.degree.nameM.S.

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
u0015_0000001_0004534.pdf
Size:
2.42 MB
Format:
Adobe Portable Document Format
Download

Collections

Theses and Dissertations