Using Ensemble Learning for Remote Sensing Inversion of Water Quality Parameters in Poyang Lake

Changchun Peng, Zhijun Xie () and Xing Jin
Additional contact information
Changchun Peng: Faculty of Electrical Engineering and Computer Science, Ningbo University, Ningbo 315211, China
Zhijun Xie: Faculty of Electrical Engineering and Computer Science, Ningbo University, Ningbo 315211, China
Xing Jin: Faculty of Electrical Engineering and Computer Science, Ningbo University, Ningbo 315211, China

Sustainability, 2024, vol. 16, issue 8, 1-19

Abstract: Inland bodies of water, such as lakes, play a crucial role in sustaining life and supporting ecosystems. However, with the rapid development of socio-economics, water resources are facing serious pollution problems, such as the eutrophication of water bodies and degradation of wetlands. Therefore, the monitoring, management, and protection of inland water resources are particularly important. In past research, empirical models and machine learning models have been widely used for the water quality assessment of inland lakes. Due to the complexity of the optical properties of inland lake water bodies, the performance of these models is often limited. To overcome the limitations of these models, this study uses in situ water quality data from 2017 to 2018 and multispectral (MS) remote sensing data from Sentinel-2 to construct experimental samples of Poyang Lake. Based on these experimental samples, we constructed a spatio-temporal ensemble model (STE) to evaluate four common water quality parameters: chlorophyll-a (Chl-a), total phosphorus (TP), total nitrogen (TN), and chemical oxygen demand (COD). The model adopts an ensemble learning strategy, improving the model’s performance by merging multiple advanced machine learning algorithms. We introduced several indices related to water quality parameters as auxiliary variables, such as NDCI and Enhanced Three, and used band data and these auxiliary variables as predictive variables, thereby greatly enhancing the predictive potential of the model.The results show that the inversion accuracy of these four inversion models is high ( R 2 of 0.94, 0.88, 0.92, and 0.93; RMSE of 1.15, 0.01, 0.02, and 0.02; MAE of 0.81, 0.01, 0.09, and 0.10), indicating that the STE model has good evaluation accuracy. Meanwhile, we used the STE model to reveal the spatio-temporal distribution of Chl-a, TP, TN, and COD from 2017 to 2018, and analyzed their seasonal and spatial variation rules. The results of this study not only provide an effective and practical method for monitoring and managing water quality parameters in inland lakes, but also provide water security for socio-economic and ecological environmental safety.

Keywords: remote sensing inversion; water quality monitoring; inland water; machine learning; ensemble learning; Poyang Lake (search for similar items in EconPapers)
JEL-codes: O13 Q Q0 Q2 Q3 Q5 Q56 (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2071-1050/16/8/3355/pdf (application/pdf)
https://www.mdpi.com/2071-1050/16/8/3355/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jsusta:v:16:y:2024:i:8:p:3355-:d:1377197

Access Statistics for this article

Sustainability is currently edited by Ms. Alexandra Wu

More articles in Sustainability from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().