Analyzing and predicting the trend of land cover degradation and determining the changes in landscape metrics using remote sensing

Introduction
Human-induced land use changes have widespread effects on the landscape. Examining and quantifying these changes can be beneficial in planning and sustainable land management. Landscape metrics are tools for measuring and describing the underlying spatiotemporal patterns and structures statistically in any landscape. These metrics can be used as a basis for comparing different scenarios of landscape or recognizing changes and developments in landscape over time. The use of landscape metrics, while saving time, makes it possible to assess the environmental impact of human activities in the shortest time. Also, detecting and predicting land use changes provide beneficial information about the trend of land development and degradation. Halilroud watershed is one of the important parts of the Jazmurian watershed located in the southeast of Iran, which is one of the most important agricultural hubs in Iran. Therefore, the purpose of this research is to investigate land use changes in the past and to predict land use in the future using remote sensing and land change modeler (LCM) and quantitative analysis of these changes based on landscape metrics in Halilrud Watershed.
 
Materials and Methods
The data of the Landsat 5 thematic mapper (TM) sensor (1991), enhanced thematic mapper plus (ETM+) sensor (2002), and Landsat 8 operational land imager (OLI) sensor (2019) were used to evaluate the trend of the land use and land cover (LULC) changes in Halilrud Watershed. The land use maps were classified into seven land uses including dam lake, residential lands, agricultural lands, rock, orchards, rangelands, and bare lands. The classification was done based on the maximum likelihood method. Then, the land use map of 2040 was simulated using a land change modeler and artificial neural network. Finally, landscape metrics were calculated at both landscape and class levels using Fragstats 4.2 to quantify structural changes. The metrics used at the class level include class area (CA), largest patch index (LPI), percentage of lands (PLAND), and number of patches (NP). cohesion index (COHESION), Shannon’s diversity index (SHDI), Simpson diversity index (SIDI), interspersion juxtaposition index (IJI), and patch density (PD) are used for quantifying changes at the landscape level.
 
Results and Discussion
The results showed that from 1991 to 2019, and the future period (2040), the landscape of the studied area changed in terms of structure and composition. A significant increase in the area of agricultural, barren, and residential lands and a decrease in the rangelands are evident in this Watershed. Shannon's and Simpson’s diversity indices show an increase in 2019 and 2040 compared to 1991, showing that the study area has become more fragmented and heterogeneous under exploitation and human activities. In addition, the value of the patch density has increased during the study period, which indicates the division of the landscape into smaller parts. The IJI also increased from 1991 to 2040, which indicates landscape diversity in the study area. The value of the contagion has decreased in the study period, which indicates that the patches have been spatially separated from each other. The results of the metrics at the class level are in line with the results of the metrics at the level of the landscape. In general, the analysis of landscape metrics has shown the extensive replacement of average rangelands by agricultural, residential lands, orchards, and bare lands. According to PLAND and CA metrics, the patches of agriculture, residential land, orchard, and bare land have increased and the patches of rangeland and dam lake have decreased during the study periods. The results obtained from the application of the metrics used in the research show the effectiveness of the metrics of the percentage of lands (PLAND), class area (CA), largest patch index (LPI), and patch number (NP) in examining and analyzing the changes. According to the results of the analysis of this metric at the level of the agricultural land class, the LPI metric was initially at the lowest level, and then with the increase of agricultural land, its values increased from 0.86% to 2.26% during 1991 to 2019 and will increase in the next period (2040) and will reach to 3.8%. Also, the rangeland class has faced an increase in the patch number during 1991-2040, and this indicates that the existing patches have fragmented and lost their integrity over time.
 
Conclusion
One of the limitations and challenges of the research is the lack of access to socio-economic and soil science data as one of the factors affecting land use changes. Therefore, it is suggested to investigate the role of other variables affecting land use changes such as soil types and socio-economic information to improve the performance of the model and prepare a more accurate prediction map. Regarding the uncontrolled growth of residential and agricultural lands in recent years, to prevent more degradation and also to preserve rangelands, it is suggested to accomplish land use planning based on the concepts of landscape. The change in the landscape structure has occurred in land use types with different degrees, and quantifying these changes using landscape metrics is one of the issues that can help to analyze the pattern of spatial changes. According to the high ability of landscape metrics to quantify the landscape pattern, the results of these studies can be used in the planning and integrated landscape planning.