Case Study on the Application of Machine Learning in Hydrology
(2026) In TVVR 5000 VVRM05 20252Division of Water Resources Engineering
- Abstract (Swedish)
- This thesis presents a case study on the application of artificial intelligence (AI) in hydrology using two approaches. This topic was chosen since AI is an important tool, and water is a vital resource. The first case is on leakage detection with artificial data, while the second is on precipitation prediction with real data. For leakage detection, a random forest (RF) model and an artificial neural network (ANN) model were developed in series. The RF detects if there has been a leak or not; the detected leaks are forwarded to the ANN. The ANN determines where in the water distribution network the leak is located. This design achieved an accuracy of 88.6% and a precision of 94.1%. For precipitation, six models were developed; one spatial... (More)
- This thesis presents a case study on the application of artificial intelligence (AI) in hydrology using two approaches. This topic was chosen since AI is an important tool, and water is a vital resource. The first case is on leakage detection with artificial data, while the second is on precipitation prediction with real data. For leakage detection, a random forest (RF) model and an artificial neural network (ANN) model were developed in series. The RF detects if there has been a leak or not; the detected leaks are forwarded to the ANN. The ANN determines where in the water distribution network the leak is located. This design achieved an accuracy of 88.6% and a precision of 94.1%. For precipitation, six models were developed; one spatial model and five area-specific prediction models. The models developed for precipitation outperformed the current satellite products with R2 values above 0.5 for five out of six models. The results from both parts of the case study show promise, and hydrology is definitely a field into which AI should be further integrated. (Less)
- Popular Abstract (Swedish)
- Vatten är en livsviktig resurs, men trots detta går globalt cirka 30 procent av det vatten som pumpas in i ledningsnäten förlorat. Genom att implementera artificiell intelligens (AI) har denna avhandling utvecklat metoder för att lokalisera läckor med hög precision, samt modeller som förbättrar noggrannheten på satellitbaserad nederbördsdata.
Avhandlingen har undersökt två tillämpningar av AI på verkliga problem: detektering av läckage och förbättring av nederbördsdata. Läckagedetekteringen löstes med en metod bestående av två steg. Först används en modell som avgör om systemet har en läcka eller inte. Om en läcka upptäckts förutsäger den andra modellen var den är belägen. Denna metod uppnådde hög noggrannhet, vilket visar på stor... (More) - Vatten är en livsviktig resurs, men trots detta går globalt cirka 30 procent av det vatten som pumpas in i ledningsnäten förlorat. Genom att implementera artificiell intelligens (AI) har denna avhandling utvecklat metoder för att lokalisera läckor med hög precision, samt modeller som förbättrar noggrannheten på satellitbaserad nederbördsdata.
Avhandlingen har undersökt två tillämpningar av AI på verkliga problem: detektering av läckage och förbättring av nederbördsdata. Läckagedetekteringen löstes med en metod bestående av två steg. Först används en modell som avgör om systemet har en läcka eller inte. Om en läcka upptäckts förutsäger den andra modellen var den är belägen. Denna metod uppnådde hög noggrannhet, vilket visar på stor potential för tidig läckagedetektering. Detta kan hjälpa VA-bolag att minska vattenförluster och spara både pengar och vatten.
Den andra delen av studien fokuserar på förbättring av nederbördsuppskattningar i Skåne. Satellitbaserad nederbördsdata har stor täckningsgrad men uppvisar inte alltid hög noggrannhet. Genom att kombinera satellitdata med historisk nederbördsdata kunde de utvecklade modellerna överträffa samtliga utforskade satellitprodukter. Fem av sex modeller uppnådde hög prestanda, vilket demonstrerar nyttan av AI och maskininlärning när det gäller att förutspå nederbörd. (Less) - Popular Abstract
- Water is the most vital resource, yet globally, 30% is lost through leaks. By implementing artificial intelligence (AI), this thesis has developed methods to locate leaks with high precision, as well as models to improve accuracy on satellite-based precipitation data.
This thesis investigates two applications of AI on real-world problems; detecting water leaks and improving precipitation data. The leak detection case was solved with a pipeline of two AI models. First, a model that predicts whether the system contains a leak or not. If a leak is detected, a second model predicts where it is located. This resulted in an accuracy of almost 89%, showing strong potential for early leak detection. This could help utilities reduce water losses... (More) - Water is the most vital resource, yet globally, 30% is lost through leaks. By implementing artificial intelligence (AI), this thesis has developed methods to locate leaks with high precision, as well as models to improve accuracy on satellite-based precipitation data.
This thesis investigates two applications of AI on real-world problems; detecting water leaks and improving precipitation data. The leak detection case was solved with a pipeline of two AI models. First, a model that predicts whether the system contains a leak or not. If a leak is detected, a second model predicts where it is located. This resulted in an accuracy of almost 89%, showing strong potential for early leak detection. This could help utilities reduce water losses and save both money and water.
In the second part of the study, the focus is on improving rainfall estimates in the region of Skåne in southern Sweden. Satellite-based precipitation data provides wide coverage but often lacks accuracy. By combining satellite data with ML and historical precipitation data, we developed models that outperformed existing satellite products in providing accurate precipitation data. Five out of six models achieved strong predictive performance, which demonstrates the usefulness of AI and ML within the field of estimating precipitation.
Together, these results show how AI can be used to achieve more sustainable water management by enabling earlier leak detection and more accurate precipitation data. This work highlights the potential for the integration of AI and ML into the field of hydrology. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/student-papers/record/9224746
- author
- Davidsson, Ola LU and Sweger, Axel
- supervisor
- organization
- alternative title
- Fallstudie om Tillämpningen av Maskininlärning inom Hydrologi
- course
- VVRM05 20252
- year
- 2026
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Machine Learning, ANN, Leakage Detection, Precipitation
- publication/series
- TVVR 5000
- report number
- TVVR25/5021
- ISSN
- 1101-9824
- language
- English
- additional info
- Examiner: Kenneth M. Persson
- id
- 9224746
- date added to LUP
- 2026-03-26 08:52:09
- date last changed
- 2026-03-26 08:52:09
@misc{9224746,
abstract = {{This thesis presents a case study on the application of artificial intelligence (AI) in hydrology using two approaches. This topic was chosen since AI is an important tool, and water is a vital resource. The first case is on leakage detection with artificial data, while the second is on precipitation prediction with real data. For leakage detection, a random forest (RF) model and an artificial neural network (ANN) model were developed in series. The RF detects if there has been a leak or not; the detected leaks are forwarded to the ANN. The ANN determines where in the water distribution network the leak is located. This design achieved an accuracy of 88.6% and a precision of 94.1%. For precipitation, six models were developed; one spatial model and five area-specific prediction models. The models developed for precipitation outperformed the current satellite products with R2 values above 0.5 for five out of six models. The results from both parts of the case study show promise, and hydrology is definitely a field into which AI should be further integrated.}},
author = {{Davidsson, Ola and Sweger, Axel}},
issn = {{1101-9824}},
language = {{eng}},
note = {{Student Paper}},
series = {{TVVR 5000}},
title = {{Case Study on the Application of Machine Learning in Hydrology}},
year = {{2026}},
}