A Data-Driven Framework for Identifying Tropical Wetland Model

Abstract

A wetland is a land area that is saturated with water. Most of the wetlands exhibit seasonal variations because of soil characteristics, climate variables and orography of a site. This study applies the orthogonal least square (OLS) algorithm under the system identification methodology for the identification of a nonlinear dynamic model structure of the tropical wetlands, using a remotely sensed dataset. Despite the availability of data from the multiple tropical sites, a single dynamic-model structure is able to explain the underlying processes, governing the wetland extents of the tropics. The model is validated against a fresh data set, derived using the similar remote sensing technique. Overall, this study is a novel application of the systems identification for obtaining a single model structure of a category of wetlands, enabling some understanding about their dynamics. The model can also be employed for the assessment of future wetlands in the advent of climate change.