Abstract
In the present study, the health status of the Choghakhor wetland, as one of the twenty-five Iranian wetlands of international importance, is assessed using the ESCOM's wetland classification and risk assessment index (WCRAI). Furthermore, the ability of the FCM-ANFIS model is investigated to predict the dissolved oxygen (DO) with electrical conductivity, water level, air temperature and water temperature as input variables. The results highlight that the Choghakhor wetland is categorized as "D" based on the A-F ecological category, which means that the wetland ecosystem has changed and a large number of local species have become extinct. Also, the output of the FCM-ANFIS model indicates the acceptable performance of the model in predicting the amount of DO due to the high values of the efficiency criteria (for the best model, DC = 0.92 and RMSE = 0.07).
Keywords
Wetland health assessment, ESCOM's WCRAI, Online monitoring station, Choghakhor wetland, FCM-ANFIS

1. Introduction
Wetlands are among the most valuable and productive ecosystems on the earth. They provide various supporting, provisioning, regulating, and cultural services, all of which make a significant contribution to the well-being of human populations. These delicate ecosystems have been long exposed to unsustainable resource extraction and various anthropogenic activities such as the spread of urbanization, industrialization, and land-use change. Iran has 25 sites designated as wetlands of international importance (Ramsar sites) out of 2,290 worldwide. From 25 sites, about one-third are under pressure or in critical condition. Although Wetland management in Iran has received more and more attention in recent years, in many instances, these efforts are being held back by the lack of comprehensive assessment tools, insufficient coordination among stakeholders, and weak policy and legislative frameworks. The risk assessment of these endangered ecosystems is highly required so that the appropriate management actions to prevent or limit the risks can be carried out in time.
In recent years, wetland risk assessment has gained increased attention among scholars, becoming an integral part of sustainable wetland management worldwide. In this regard, various studies, using different indicators and methodologies, have been conducted to evaluate wetland health status and determine the factors threatening these delicate ecosystems. One of the first and most widely used methods for assessing wetland health status was proposed by Cowardin et al. (1979). This method tries to classify wetlands with a comprehensive approach, but it is relatively complex and has some drawbacks. Kotze et al. (2005) based on the Cowardin method proposed a new method that had the ability to classify wetlands in any climatic conditions. This method categorizes wetlands by evaluating the three characteristics of hydrology, geology and wetland plant species.
Until the past few years, there have been relatively few investigations dealing with real-time and water quality data to assess wetlands within arid and semi-arid regions. The development of time-series data mining techniques has provided an emerging method for wetland health assessment. Soft computing methods such as adaptive neuro-fuzzy inference system (ANFIS) can play an important role in this regard. In the present study, ESCOM's wetland classification and risk assessment index (WCRAI) is used for the first time to evaluate Choghakhor wetland health status. Furthermore, the ability of the FCM-ANFIS model is investigated to predict the dissolved oxygen (DO) in order to provide a model with high reliability to approximate the wetland parameters in the future.

2. Methodology
2.1. Case study
The Choghakhor wetland in the south of Shahr-e Kord, Chaharmahal and Bakhtiari province (31° 55' 23" latitude North, 50° 53' 59" longitude East) is located at the northern slopes of Kallar mountain (Fig. 1). The topography of the region is such that the plain formed by the mountains "Baraftab" in the north, "Shapurnaz" in the west, and "Kallar" from the west to the south has created a drainage basin of 114 square kilometers as a suitable location to feed a wetland with a surface area of about 1500 hectares.
2.2. wetland risk assessment
Increases in the need for effective wetland assessment and monitoring programs have resulted in numerous rapid appraisal methods. Among them, the ESCOM's WCRAI is a reliable approach to determine the overall wetland health status, which can be implemented by non-wetland experts. This index involves field sampling and in situ measurement of some water physicochemical parameters, including DO, EC, and pH, as well as various physical characteristics of the wetland.
In order to facilitate the measurements, and considering the fact that different parts of the selected wetland are exposed to different anthropogenic activities, including agricultural activities, wastewater disposal, tourism, and livestock grazing, it is divided into four parts, and the results obtained from these parts are averaged to calculate the overall values (Fig. 2). Water samples are collected periodically every month during 2017 and 2018 in the Choghakhor wetland. Four sampling points are located in the study area and the periodic values obtained from these sampling points are averaged to calculate the overall values of the indicators.
2.3. Adaptive Neuro-Fuzzy Inference System (ANFIS)
Jang (1993) first proposed the adaptive neural fuzzy inference system by combining the capabilities of fuzzy theory with neural networks. It provides accelerated learning capacity and adaptive interpretation capabilities to model complex patterns and apprehends nonlinear relationships. In the present study, the fuzzy c-means clustering method (FCM) is used to derive the rules in Takaki-Sugeno fuzzy model.

3. Results and Discussion
3.1. Modeling of Water Quality Parameters Using FCM-ANFIS
In this study, nonlinear regression is performed on the qualitative data obtained from the wetland, in order to model the relationship between its parameters. For this purpose, an adaptive neural fuzzy inference system is utilized to model changes in DO at a depth of 1.4 m in the Choghakhor wetland. Water temperature, electrical conductivity at three depths of 0.3, 0.85 and 1.4 m, air temperature, water level and DO at two depths of 0.3 and 0.85 m are selected as input parameters. Daily and one-hour normalized time series data are used for the modeling. The results of modeling the changes in the DO parameter at a depth of 1.4 m are given in Table 8. The values of the efficiency criteria are measured to evaluate the performance of the developed models to the related inputs, using the determination coefficient (DC) and the root mean square error (RMSE). The results of the model in scenario M14 with electrical conductivity (0.3 m), water level, air temperature and water temperature (1.4 m) as input variables led to higher performance with the best efficiency criteria, including DC = 0.92 and RMSE = 0.07.
3.2. Evaluation of the wetland health level
Risk assessment is an integral part of sustainable wetland management, where the ecosystem's health and wise exploitation are ensured. In this regard, the health status of the Choghakhor wetland, as one of the most important Iranian Ramsar sites, is assessed with ESCOM's WCRAI. Tables 10 and 11 represent the results of field measurements for four parts of the study area in 2017 and 2018, respectively. The results obtained confirmed a decline in the wetland health level in 2018 compared to 2017. The ESCOM's WCRAI generated overall scores of 48% and 41% for the Choghakhor wetland health level for 2017 and 2018, respectively.

4. Conclusion
In this study, the health status of the Choghakhor wetland is assessed using the ESCOM's WCRAI. In this method, by performing field sampling, important chemical and physical characteristics of the wetland are measured and a score is assigned to the wetland based on these measurements, which indicates the health status of the wetland. The ESCOM's WCRAI generated overall scores of 48% and 41% for the Choghakhor wetland health level for 2017 and 2018, respectively. According to the results, the Choghakhor wetland is categorized as "D" based on the ecological category. This means that the wetland ecosystem has undergone a transformation and a large number of local species have become extinct.
Moreover, the FCM-ANFIS model is utilized to model changes in DO at a depth of 1.4 m in the Choghakhor wetland. The results highlight that the FCM-ANFIS model has a high ability to predict DO based on physical and water quality parameters. The results of the M14 model with electrical conductivity (0.3 m), water level, air temperature and water temperature (1.4 m) as input variables led to higher performance with the best efficiency criteria, including DC = 0.92 and RMSE = 0.07.