Assessment and forecasting of surface water changes in the Hamoun Lakes using Landsat imagery and CA-Markov model

Introduction

In recent decades, the Hamoun Lakes have almost dried up due to drought and unsustainable management of water resources (construction of dams, expansion of agricultural lands and systematic diversion of water). This issue, considering the importance of these lakes in the Sistan region, has caused many environmental and socio-economic challenges, including dust storms in this region. Therefore, it seems necessary to study the changes of water areas and their effects on the surrounding environment. Accordingly, the forecasting/projection of the future perspective of land use/cover (LULC) changes is considered as one of the key and fundamental tools for planning and sustainable management of territorial resources and decision-making to address the resulting challenges. In this regard, the literature shows that the use of multispectral remote sensing satellite imagery has found widespread use due to its wide spatial-temporal coverage and lack of dependence on costly field work. Various approaches have been used to model LULC changes and project future trends, among which the results of the CA-Markov model are more reliable. Despite the many studies that have been conducted to monitor LULC changes in the Hamoun Lakes, their focus has mostly been on agriculture and vegetation classes, and the issue of lake drying and its different dimensions have not been well understood. On the other hand, considering the rapid changes of the Hamoun lakes in the last two decades, it seems that we should focus on projecting the state of the lakes in the near future, which has not been given much attention in previous studies.

Methodology

The main objective of this study is to forecast the changes in the water areas of Hamoun Lakes located in southeastern Iran. For this purpose, the Landsat 5-TM and Landsat 8-OLI images were used for the years 1991 and 2022, respectively. Initially, the support vector machine (SVM) classifier was used to produce LULC maps for these years. The main advantage of SVM is its ability to solve complex classification problems with a large number of features and few training samples, making it a suitable option for classifying remote sensing images. Then, the changes in the water areas of the Hamoun Lakes for the year 2030 were forecasted taking advantage of the CA-Markov model. The CA-Markov analysis is a suitable tool for modeling LULC changes.

Results and discussion

The SVM classifier gave the best performance for the Radial Basis Function (RBF) kernel. This kernel has the highest (maximum 98.3%) and lowest (minimum 86.3%) intra-class accuracy in detecting water body and built-up classes, respectively, in the years studied. Achieving an overall classification accuracy of more than 93% for the RBF kernel indicates the acceptable performance of the SVM classifier and the reliability of its results. The findings show that the area of lakes and wetlands in the region has decreased by 88% and 94%, respectively, in 2022 compared to 1991. During the study period, the area of salty lands has almost doubled, and the greatest loss of area corresponds to water and vegetation classes that have been transformed into this class. Forecasting of the LULC changes using the CA-Markov model indicates that in less than the upcoming decade, water areas including Hamoun Puzak and Saberi lakes and wetlands will become completely dried up and transformed into salty lands.

Conclusion

The very high rate of LULC change in the study area indicates the very critical situation of the region in terms of the loss of water and plant resources and the expansion of desert lands. Therefore, it is necessary to focus on the near future and target setting to achieve sustainable territorial management in this region based on short-term time horizons with high priority. In general, the results of the current research emphasize the critical situation of the surface water resources in the study area and the need for special attention to manage the resulting problems.