Monitoring ecological status of wetlands using linked fuzzy inference system- remote sensing analysis

Abstract

The present study develops an applicable model to simulate the ecological status of saltwater lakes in which depth and total dissolved solids are selected as the effective factors on aquatic habitats. First, spectral images of the operational land imager of Landsat 8 were used to simulate distribution of depth and total dissolved solids by applying two feed forward neural networks. Next, a Mamdani fuzzy inference system was used to develop habitat suitability rules of Artemia and Flamingo as the selected target species in which expert opinions were considered. Finally, habitat suitability maps of target species were generated by linking distribution maps of selected effective parameters and fuzzy inference system. Based on the results in the Urmia lake as a case study, the Nash–Sutcliffe efficiency coefficients of depth and total dissolved solids are 0.88 and 0.5 which indicates the proposed method for simulating distribution of these parameters is reliable. Average depth in the simulated date is 227 cm, while average simulated total dissolved solids is 264 g per litre. Simulation of habitat suitability maps demonstrated that average habitat suitability of Artemia is less than 30% in the most areas of the lake. Moreover, average habitat suitability of the Flamingo is less than 10% which implies the ecological status of the lake is critical and ecological restoration is necessary. The main advantage of the proposed method is to develop a framework for combining the expert opinions with remote sensing data processing to generate habitat suitability maps in lakes.