Field Measurements and Machine Learning Algorithms to Monitor Water Quality in Lakes Located in Landscape Parks – A Case Study

Abstract

One of the greatest threats to many lakes is their accelerated eutrophication resulting from anthropogenic pressure, agricultural intensification, and climate change. A very important element of surface water protection in environmentally conserved areas is the proper monitoring of water quality and detection of potential threats by examining the physicochemical properties of water and performing statistical analyses that enable possible exposure of unfavourable trends. The article presents the analyses of the results of measurements made in three lakes located in the Sierakowski Landscape Park. As part of the measurements, water quality indicators i.e., phosphorus, nitrogen, BOD 5 and COD, were determined monthly for a year at the inflows and outflows of the studied lakes. The test results of selected water quality indicators were analysed using machine learning algorithms i.e., PCA and k-means. The conducted tests enabled statistical estimation of changes in water quality indicators in the reservoirs and evaluation of their correlation.