Detecting the changes in Miqan lagoon zone by using NDWI, MNDWI, AWEI and supervised SVM models in the period of 1373 to 1401

Climate changes and mismanagement in recent years have caused the destruction of wetlands and lakes in large parts of the world, especially arid and semi-arid areas. In the central parts of Iran, this phenomenon is more evident and studies have shown that wetlands and lakes are drying up. Miqan lagoon is one of the cases that has faced a sharp decrease in water level. In this research, spatial and temporal changes of Miqan lagoon from 1373 to 1401 were evaluated using TM and OLI images of Landsat 5 and 8 satellites.

For this purpose, image data from SVM and NDWI, MNDWI and AWEI were used. The spectral and spatial performance of each classification was compared using Pearson's correlation.

In general, SVM model along with NDWI, MNDWI and AWEI indices achieved better results in terms of spectral and spatial quality. Based on the applied methods, the lake level shows a sharp decrease between 1373 and 1401. The results show the effectiveness of AWEI and MNDWI models in detecting water level changes in certain time intervals.

According to the AWEI model, the area of the lagoon has decreased from 112.6 square kilometers in 1373 to 78.5 square kilometers. The highest Pearson correlation between AWEI and MNDWI was observed in 1373, 1381, 1393 and 1401 with values of 0.93, 0.96, 0.97 and 0.97 respectively.

Innovation: 

So far, spectral indices and learning algorithm models have not been used in the investigation of Miqan wetland. In this research, new indices such as AWEI have been used along with the support vector machine model.