Spatial modeling of chlorophyll-a parameter by Landsat-8 satellite data and deep learning techniques: The case of Lake Mogan

Water is essential for the sustainability of life and the healthy functioning of ecosystems. Increasing pollution poses a serious threat to the world's waters, making the monitoring and protection of water quality a strategic imperative. Chlorophyll-a is one of the most important indicators of water quality and ecosystem health, as it is a measure of photosynthetic activity and phytoplankton density, the lifeblood of aquatic ecosystems. Remote sensed data provide a unique opportunity to analyse chlorophyll-a changes in lake ecosystems. In this study, chlorophyll-a concentration was modelled by machine and deep learning techniques using chlorophyll-a measurements, Landsat-8 surface reflectance values and spectral indices of Lake Mogan between 2018 and 2024. The RF, ANN, and CNN models achieved R² values of 0.84, 0.85, and 0.92, respectively. With its ability to learn spectral relationships, identify patterns in complex datasets, and its superior ability to process remote sensing imagery, thematic maps were generated using the CNN model, which performed best in the study. The results of the study demonstrate the potential of remote sensing-based deep learning approaches for monitoring chlorophyll-a. With its ability to produce highly accurate results, this study provides the literature with an effective tool for future strategic monitoring studies.