Artificial Intelligence and Remote Sensing for Freshwater Quality Monitoring: A Systematic Review of Methods, Applications, Model Performance, and Research Gaps
- 27 October 2026
- journal article
- review article
- Published by Wiley in Environmental Quality Management
- Vol. 36 (4)
Abstract
Artificial intelligence (AI) integrated with remote sensing offers a scalable approach to freshwater‐quality monitoring, potentially extending the spatial and temporal coverage of conventional field measurements. However, evidence on model performance, validation, and transferability remains heterogeneous. This systematic review synthesized evidence on AI and remote sensing for freshwater‐quality monitoring, focusing on remote‐sensing platforms, target parameters, AI/machine‐learning (ML) methods, predictive performance, validation, and generalizability. Web of Science Core Collection, Scopus, PubMed/MEDLINE, ScienceDirect, and IEEE Xplore were searched for studies published from March 2000 to April 30, 2026. Study selection followed PRISMA 2020 guidance. Eligible studies used remote‐sensing data with AI/ML or deep‐learning methods to assess freshwater quality. Data on sensors, target variables, algorithms, performance metrics, validation, uncertainty, explainability, and reproducibility were extracted and synthesized descriptively. Thirty‐one primary studies (S01–S31) were included and consistently cross‐referenced throughout the manuscript and supplementary tables. Sentinel‐2/MSI (16/31; 51.6%) and Landsat (10/31; 32.3%) were the most frequently used platforms. Chlorophyll‐a (15/31; 48.4%), turbidity (11/31; 35.5%), and total suspended solids (8/31; 25.8%) were the most studied parameters. Random Forest (17/31; 54.8%), support‐vector methods (11/31; 35.5%), and artificial neural networks (10/31; 32.3%) were commonly applied. Internal validation predominated, while independent temporal, spatial, cross‐site, cross‐region, cross‐sensor, and external validation was uncommon. Uncertainty, explainability, and data/code availability were inconsistently reported. Overall, evidence supports strong within‐study predictive performance more than environmental generalizability. Future research should prioritize independent multi‐site and multi‐season validation, uncertainty quantification, explainable AI, standardized reporting, and reproducible workflows.Keywords
This publication has 15 references indexed in Scilit:
- Deep learning-based remote sensing retrieval of inland water quality: A reviewJournal of Hydrology: Regional Studies, 2025
- TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methodsBMJ, 2024
- Retrieving water chlorophyll-a concentration in inland waters from Sentinel-2 imagery: Review of operability, performance and ways forwardInternational Journal of Applied Earth Observation and Geoinformation, 2023
- Transfer learning in environmental remote sensingRemote Sensing of Environment, 2023
- Monitoring inland water via Sentinel satellite constellation: A review and perspectiveISPRS Journal of Photogrammetry and Remote Sensing, 2023
- Machine Learning in Environmental Research: Common Pitfalls and Best PracticesEnvironmental Science & Technology, 2023
- Remote Sensing Big Data for Water Environment Monitoring: Current Status, Challenges, and Future ProspectsEarth's Future, 2022
- The PRISMA 2020 statement: an updated guideline for reporting systematic reviewsBMJ, 2021
- Monitoring inland water quality using remote sensing: potential and limitations of spectral indices, bio-optical simulations, machine learning, and cloud computingEarth-Science Reviews, 2020
- block CV : An r package for generating spatially or environmentally separated folds for k ‐fold cross‐validation of species distribution modelsMethods in Ecology and Evolution, 2018