Development of a Novel Fuzzy Logic-Based Wetland Health Assessment Approach for the Management of Freshwater Wetland Ecosystems

In the present study, a new wetland health assessment approach based on the fuzzy inference system (FIS) was developed and proposed for the first time to improve a traditional wetland classification and assessment index. One primary purpose of the study is to modify the indicators of the traditional assessment approach to suit the regional environmental conditions of the selected areas. As two of the twenty-five Iranian wetlands with international importance, Kanibarazan and Choghakhor wetlands were selected as the study areas due to their significant roles in protecting the biodiversity of their regions. The wetlands are supported by the international Ramsar Convention on Wetlands to mandate and encourage the local authorities towards their conservation and sustainable exploitation. In this regard, the Iranian Department of Environment, in cooperation with the Global Environment Facility (GEF) and the United Nations Development Programme (UNDP), selected these wetlands to demonstrate new approaches of managing the wetland areas protected by the Conservation of Iranian Wetlands Project (CIWP). A real-time wetland monitoring station with hydrological instruments, including water level, air temperature, air humidity, and water quality multi-parameter sensors recording water temperature, pH, electrical conductivity, and dissolved oxygen (DO), was implanted at the deepest part of both wetlands. The manual sampling of water quality parameters was also carried out periodically during specific intervals. The relative importance of the wetland health indicators involved in the FIS was determined via the analytic hierarchy process (AHP), utilizing the knowledge of local experts, including academic staff, environmental specialists, and natives, to localize the traditional assessment approach. In turn, the health level categories of both wetlands were assessed using the traditional and proposed wetland health assessment approaches. The efficiency of the proposed method was evaluated with the selected case studies, and it proved to be a more flexible and appropriate approach for wetland health assessment. Furthermore, the observed differences between the health level of the first case study pointed to the efficiency of the AHP-FIS method in improving the traditional index. Besides, feedforward neural network (FFNN) and support vector regression (SVR) artificial intelligence (AI) methods were used to model DO as one of the most critical water quality parameters in main