Machine learning-based detection and mapping of riverine litter utilizing Sentinel-2 imagery

Aug 5, 2026

Upload by Patrick

Rivers strongly influence the global marine litter problem, yet riverine litter receives limited attention. This study detects and maps riverine litter using middle-scale multispectral satellite imagery and machine learning, using the Tisza River in Hungary as the case study. Very High Resolution images from Google Earth support labeling for five supervised algorithms (ANN, SVC, RF, NB, DT) trained and validated on Sentinel-2. Models show strong validation metrics but only medium to poor test performance under different hydrological conditions and litter sizes, with accuracy constrained by Sentinel-2 pixel resolution. Spatio-temporal analysis identifies hydraulic structures, such as Kisköre Dam, as major accumulation hotspots.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Abstract

Keywords

Introduction

Environmental Science and Pollution Research [https://doi.org/10.1007/s1](https://doi.org/10.1007/s1) 1356-023-27068-0 Machine learning‑based detection and mapping of riverine litter utilizing Sentinel‑2 imagery Ahmed Mohsen1,2 · Tímea Kiss1 · Ferenc Kovács1 Received: 20 September 2022 / Accepted: 12 April 2023 © The Author(s) 2023 Abstract Despite the substantial impact of rivers on the global marine litter problem, riverine litter has been accorded inadequate consideration. Therefore, our objective was to detect riverine litter by utilizing middle-scale multispectral satellite images and machine learning (ML), with the Tisza River (Hungary) as a study area. The Very High Resolution (VHR) images obtained from the Google Earth database were employed to recognize some riverine litter spots (a blend of anthropogenic and natural substances) . These litter spots served as the basis for training and validating five supervised machine-learning algorithms based on Sentinel-2 images [Artificial Neural Network (ANN), Support Vector Classifier (SVC), Random Forest (RF), Naïve Bays (NB) and Decision Tree (DT)]. To evaluate the generalization capability of the developed models, they were tested on larger unseen data under varying hydrological conditions and with different litter sizes. Besides the best-performing model was used to investigate the spatio-temporal variations of riverine litter in the Middel Tisza. According to the results, almost all the developed models showed favorable metrics based on the validation dataset (e.g., F1-score; SVC: 0.94, ANN: 0.93, RF: 0.91, DT: 0.90, and NB: 0.83); however, during the testing process, they showed medium (e.g., F1-score; RF:0.69, SVC: 0.62; ANN: 0.62) to poor performance (e.g., F1-score; NB: 0.48; DT: 0.45) . The capability of all models to detect litter was bounded to the pixel size of the Sentinel-2 images. Based on the spatio-temporal investigation, hydraulic structures (e.g., Kisköre Dam) are the greatest litter accumulation spots. Although the highest transport rate of litter occurs during floods, the largest litter spot area upstream of the Kisköre Dam was observed at low stages in summer. This study represents a preliminary step in the automatic detection of riverine litter; therefore, additional research incorporating a larger dataset with more representative small litter spots, as well as finer spatial resolution images is necessary. Keywords Tisza River · Plastic indices · Litter transport · Support vector classifier · Artificial neural network · Macroplastic Introduction The rivers are the main source of marine litter, as approximately 80% of marine litter originates from inland sources (González et al. 2021 , González et al. 2016) . The term“riverine litter” refers to the natural and artificial materials transported by a river, which are often trapped by the Responsible Editor: Philippe Garrigues * Ferenc Kovács [kovacsf@geo.u-szeged.hu](kovacsf@geo.u-szeged.hu) 1 Department of Geoinformatics, Physical and Environmental Geography, University of Szeged, Egyetem u. 2–6, Szeged 6722, Hungary 2 Department of Irrigation and Hydraulics Engineering, Tanta University, Tanta 31512, Egypt banks and hydraulic structures. The natural and artificial litter is mixed and drifting together in a form of litter spots. The litter can enter the fluvial system by run-off and direct input, and during its transportation, it degrades to smaller particles (González et al. 2016). The artificial floating materials, especially plastics, originate from public waste, landfill sites, agricultural fields, and industrial disposals (Van Emmerik et al. 2018) . Rech et al. (2014) classified artificial litter according to its ability to float, referring to persistent buoyant (plastics and wood), short-time buoyant (cigarette stubs, paper, and textiles), and non-buoyant litter (concrete, metal, and glass) . The monitoring of riverine litter aims to identify its sources, quantity, and flux

FAQs

What is the main objective of the study?

To detect and map riverine litter using middle-scale multispectral satellite images and machine learning, with the Tisza River as the study area.

Which supervised machine-learning algorithms are used?

The study trains and validates five supervised algorithms: Artificial Neural Network (ANN), Support Vector Classifier (SVC), Random Forest (RF), Naïve Bayes (NB), and Decision Tree (DT).

How do the models perform when tested on new unseen data?

Validation performance is favorable, but testing under varying hydrological conditions and different litter sizes degrades results to medium or poor levels. The detected litter capability is also limited by the Sentinel-2 pixel size.

Aug 5, 2026

Upload by Patrick