A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater Models

Abstract Hydrogeologic models generally require gridded subsurface properties, however these inputs are often difficult to obtain and highly uncertain. Parametrizing computationally expensive models where extensive calibration is computationally infeasible is a long standing challenge in hydrogeolog...

Full description

Saved in:
Bibliographic Details
Main Authors: Amanda Triplett, Andrew Bennett, Laura E. Condon, Peter Melchior, Reed M. Maxwell
Format: Article
Language:English
Published: Wiley 2025-04-01
Series:Geophysical Research Letters
Subjects:
Online Access:https://doi.org/10.1029/2024GL114285
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849397339824848896
author Amanda Triplett
Andrew Bennett
Laura E. Condon
Peter Melchior
Reed M. Maxwell
author_facet Amanda Triplett
Andrew Bennett
Laura E. Condon
Peter Melchior
Reed M. Maxwell
author_sort Amanda Triplett
collection DOAJ
description Abstract Hydrogeologic models generally require gridded subsurface properties, however these inputs are often difficult to obtain and highly uncertain. Parametrizing computationally expensive models where extensive calibration is computationally infeasible is a long standing challenge in hydrogeology. Here we present a machine learning framework to address this challenge. We train an inversion model to learn the relationship between water table depth and hydraulic conductivity using a small number of physical simulations. For a 31M grid cell model of the US we demonstrate that the inversion model can produce a reliable K field using only 30 simulations for training. Furthermore, we show that the inversion model captures physically realistic relationships between variables, even for relationships that were not directly trained on. While there are still limitations for out of sample parameters, the general framework presented here provides a promising approach for parametrizing expensive models.
format Article
id doaj-art-148ff8d590f54e4e88bce34cd44fe4a4
institution Kabale University
issn 0094-8276
1944-8007
language English
publishDate 2025-04-01
publisher Wiley
record_format Article
series Geophysical Research Letters
spelling doaj-art-148ff8d590f54e4e88bce34cd44fe4a42025-08-20T03:39:03ZengWileyGeophysical Research Letters0094-82761944-80072025-04-01528n/an/a10.1029/2024GL114285A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater ModelsAmanda Triplett0Andrew Bennett1Laura E. Condon2Peter Melchior3Reed M. Maxwell4Department of Hydrology and Atmospheric Sciences University of Arizona Tucson AZ USADepartment of Hydrology and Atmospheric Sciences University of Arizona Tucson AZ USADepartment of Hydrology and Atmospheric Sciences University of Arizona Tucson AZ USADepartment of Astrophysical Sciences Princeton University Princeton NJ USAIntegrated Groundwater Modeling Center Princeton University Princeton NJ USAAbstract Hydrogeologic models generally require gridded subsurface properties, however these inputs are often difficult to obtain and highly uncertain. Parametrizing computationally expensive models where extensive calibration is computationally infeasible is a long standing challenge in hydrogeology. Here we present a machine learning framework to address this challenge. We train an inversion model to learn the relationship between water table depth and hydraulic conductivity using a small number of physical simulations. For a 31M grid cell model of the US we demonstrate that the inversion model can produce a reliable K field using only 30 simulations for training. Furthermore, we show that the inversion model captures physically realistic relationships between variables, even for relationships that were not directly trained on. While there are still limitations for out of sample parameters, the general framework presented here provides a promising approach for parametrizing expensive models.https://doi.org/10.1029/2024GL114285groundwaterinversionmodelingcalibrationmachine learninghydraulic conductivity
spellingShingle Amanda Triplett
Andrew Bennett
Laura E. Condon
Peter Melchior
Reed M. Maxwell
A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater Models
Geophysical Research Letters
groundwater
inversion
modeling
calibration
machine learning
hydraulic conductivity
title A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater Models
title_full A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater Models
title_fullStr A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater Models
title_full_unstemmed A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater Models
title_short A Deep‐Learning Based Parameter Inversion Framework for Large‐Scale Groundwater Models
title_sort deep learning based parameter inversion framework for large scale groundwater models
topic groundwater
inversion
modeling
calibration
machine learning
hydraulic conductivity
url https://doi.org/10.1029/2024GL114285
work_keys_str_mv AT amandatriplett adeeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT andrewbennett adeeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT lauraecondon adeeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT petermelchior adeeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT reedmmaxwell adeeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT amandatriplett deeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT andrewbennett deeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT lauraecondon deeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT petermelchior deeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels
AT reedmmaxwell deeplearningbasedparameterinversionframeworkforlargescalegroundwatermodels