Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks

Robust hydrological simulation is key for sustainable development, water management strategies, and climate change adaptation. In recent years, deep learning methods have been demonstrated to outperform mechanistic models at the task of hydrological discharge simulation. Adoption of these methods ha...

Full description

Bibliographic Details
Main Authors: Tlhomole, James, Borgomeo, E., Matheswaran, Karthikeyan, Garcia Andarcia, Mariangel
Format: Preprint
Language:Inglés
Published: 2026
Subjects:
Online Access:https://hdl.handle.net/10568/180273
_version_ 1855533346903293952
author Tlhomole, James
Borgomeo, E.
Matheswaran, Karthikeyan
Garcia Andarcia, Mariangel
author_browse Borgomeo, E.
Garcia Andarcia, Mariangel
Matheswaran, Karthikeyan
Tlhomole, James
author_facet Tlhomole, James
Borgomeo, E.
Matheswaran, Karthikeyan
Garcia Andarcia, Mariangel
author_sort Tlhomole, James
collection Repository of Agricultural Research Outputs (CGSpace)
description Robust hydrological simulation is key for sustainable development, water management strategies, and climate change adaptation. In recent years, deep learning methods have been demonstrated to outperform mechanistic models at the task of hydrological discharge simulation. Adoption of these methods has been catalysed by the proliferation of large sample hydrology datasets, consisting of the observed discharge and meteorological drivers, along with geological and topographical catchment descriptors. Deep learning methods infer rainfall-runoff characteristics that have been shown to generalise across catchments, benefitting from the data diversity in large datasets. Despite this, application to catchments in Africa has been limited. The lack of adoption of deep learning methodologies is primarily due to sparsity or lack of the spatiotemporal observational data required to enable downstream model training. We therefore investigate the application of deep learning models, including LSTMs, for hydrological discharge simulation in the transboundary Limpopo River basin, emphasising application to data scarce regions. We conduct a number of computational experiments primarily focused on assessing the impact of varying the LSTM model input data on performance. Results confirm that data constraints remain the largest obstacle to deep learning applications across African river basins. We further outline the impact of human influence on data-driven modelling which is a commonly overlooked aspect of data-driven large-sample hydrology approaches and investigate solutions for model adaptation under smaller datasets. Additionally, we include recommendations for future efforts towards seasonal hydrological discharge prediction and direct comparison or inclusion of SWAT model outputs, as well as architectural improvements.
format Preprint
id CGSpace180273
institution CGIAR Consortium
language Inglés
publishDate 2026
publishDateRange 2026
publishDateSort 2026
record_format dspace
spelling CGSpace1802732026-01-21T08:01:25Z Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks Tlhomole, James Borgomeo, E. Matheswaran, Karthikeyan Garcia Andarcia, Mariangel river basins stream flow forecasting water management models Robust hydrological simulation is key for sustainable development, water management strategies, and climate change adaptation. In recent years, deep learning methods have been demonstrated to outperform mechanistic models at the task of hydrological discharge simulation. Adoption of these methods has been catalysed by the proliferation of large sample hydrology datasets, consisting of the observed discharge and meteorological drivers, along with geological and topographical catchment descriptors. Deep learning methods infer rainfall-runoff characteristics that have been shown to generalise across catchments, benefitting from the data diversity in large datasets. Despite this, application to catchments in Africa has been limited. The lack of adoption of deep learning methodologies is primarily due to sparsity or lack of the spatiotemporal observational data required to enable downstream model training. We therefore investigate the application of deep learning models, including LSTMs, for hydrological discharge simulation in the transboundary Limpopo River basin, emphasising application to data scarce regions. We conduct a number of computational experiments primarily focused on assessing the impact of varying the LSTM model input data on performance. Results confirm that data constraints remain the largest obstacle to deep learning applications across African river basins. We further outline the impact of human influence on data-driven modelling which is a commonly overlooked aspect of data-driven large-sample hydrology approaches and investigate solutions for model adaptation under smaller datasets. Additionally, we include recommendations for future efforts towards seasonal hydrological discharge prediction and direct comparison or inclusion of SWAT model outputs, as well as architectural improvements. 2026-01-11 2026-01-21T07:51:54Z 2026-01-21T07:51:54Z Preprint https://hdl.handle.net/10568/180273 en Open Access Tlhomole, J.; Borgomeo, E.; Matheswaran, K.; Garcia Andarcia, M. 2026. Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks. arXiv, 14p. doi: https://doi.org/10.48550/arXiv.2601.06941
spellingShingle river basins
stream flow
forecasting
water management
models
Tlhomole, James
Borgomeo, E.
Matheswaran, Karthikeyan
Garcia Andarcia, Mariangel
Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks
title Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks
title_full Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks
title_fullStr Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks
title_full_unstemmed Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks
title_short Towards operational streamflow forecasting in the Limpopo River Basin using long short-term memory networks
title_sort towards operational streamflow forecasting in the limpopo river basin using long short term memory networks
topic river basins
stream flow
forecasting
water management
models
url https://hdl.handle.net/10568/180273
work_keys_str_mv AT tlhomolejames towardsoperationalstreamflowforecastinginthelimpoporiverbasinusinglongshorttermmemorynetworks
AT borgomeoe towardsoperationalstreamflowforecastinginthelimpoporiverbasinusinglongshorttermmemorynetworks
AT matheswarankarthikeyan towardsoperationalstreamflowforecastinginthelimpoporiverbasinusinglongshorttermmemorynetworks
AT garciaandarciamariangel towardsoperationalstreamflowforecastinginthelimpoporiverbasinusinglongshorttermmemorynetworks