Meta-Classifier Driven Exploration of Large-Scale Wetland Dynamics with Time Series Satellite Imagery

Wetland vegetation dynamic monitoring and analysis are crucial for identifying changes, ecosystem environment restoration, and reconstruction. Wetlands all around the world are threatened by anthropogenic and climatic forces that interfere with vital ecosystem functions. This research study proposes a meta- classifier-driven approach to analyze the combined attributes using the labelled training data and classify features or images. Classification error was calculated using validation data to assess the accuracy of the meta-classifier. The Google Earth Engine an open-source planetary processing engine is used for this research. The harmonic series of time is then utilized to complete the missing NDVI facts, and the map that results is classified with a high overall accuracy of 98 and a suitable Kappa coefficient of 0.9. The outcomes of this paper provided an extensive overview of how the research region's pixel range changed when water was present, as well as how locations changed in terms of seasonality and persistence.