Monitoring the condition of water surfaces using the integrated environment of Google Earth Engine and Google Colab

Abstract

Monitoring the state of aquatic ecosystems is one of the key tasks of modern ecology and natural resource management. One of the most informative indicators of water quality is the concentration of chlorophyll-a, which reflects the level of phytoplankton and is directly related to the phenomenon of eutrophication. Traditional sampling and laboratory analysis methods are accurate but have limitations in terms of scale and speed. In this context, remote sensing and cloud information systems open up new opportunities for regular monitoring of the biological productivity of water bodies. The work involves preliminary image processing, including cloud masking (QA60) and water surface extraction using the NDWI index. To assess phytoplankton levels, the chlorophyll-a spectral index was used, which takes into account the ratio of reflectance in the red and red edge bands. To reduce the impact of noise, median composites were constructed, which allowed the identification of general spatial patterns, and difference maps provided an assessment of the dynamics of changes between selected intervals. Additionally, threshold segmentation was used to identify areas with elevated chlorophyll-a levels and to quantitatively assess blooming areas. The results confirmed the effectiveness of the methodology used to identify both local sources of nutrients and large-scale centres of summer blooms. The resulting cartographic materials clearly reflected seasonal fluctuations in phytoplankton levels and the spatial structure of blooms. At the same time, limitations of the index approach were identified, related to the influence of atmospheric conditions and different types of phytoplankton. A promising direction for further research is the application of machine learning and deep neural network methods, which are capable of integrating multispectral and hyperspectral data, increasing the accuracy of detecting and predicting water blooms. This paves the way for the creation of intelligent real-time environmental monitoring systems