HESS Opinions: Never train a Long Short-Term Memory (LSTM) network on a single basin

Machine learning (ML) has played an increasing role in the hydrological sciences. In particular, Long Short-Term Memory (LSTM) networks are popular for rainfall–runoff modeling. A large majority of studies that use this type of model do not follow best practices, and there is one mistake in particul...

Full description

Saved in:
Bibliographic Details
Published in:Hydrology and earth system sciences Vol. 28; no. 17; pp. 4187 - 4201
Main Authors: Kratzert, Frederik, Gauch, Martin, Klotz, Daniel, Nearing, Grey
Format: Journal Article
Language:English
Published: Katlenburg-Lindau Copernicus GmbH 12.09.2024
Copernicus Publications
Subjects:
Basins
Best practice
Best practices
Camelids
Datasets
Deep learning
Hydrologic models
Hydrology
Long short-term memory
Machine learning
Precipitation
Rainfall-runoff modeling
Rainfall-runoff relationships
Runoff models
Stream flow
Watersheds
ISSN:1607-7938, 1027-5606, 1607-7938
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Machine learning (ML) has played an increasing role in the hydrological sciences. In particular, Long Short-Term Memory (LSTM) networks are popular for rainfall–runoff modeling. A large majority of studies that use this type of model do not follow best practices, and there is one mistake in particular that is common: training deep learning models on small, homogeneous data sets, typically data from only a single hydrological basin. In this position paper, we show that LSTM rainfall–runoff models are best when trained with data from a large number of basins.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1607-7938
1027-5606
1607-7938
DOI:10.5194/hess-28-4187-2024