Background and Objective: Land use changes in mountainous and summer pasture regions of Iran, due to reverse migration and the expansion of second homes, have become one of the fundamental challenges of regional planning. Masal and Shanderman County in Gilan Province, due to its Hyrcanian forests, summer pastures, and temperate climate, has undergone rapid transformations in land use patterns over the past decade. The aim of this study is spatio-temporal analysis of these changes during the period 2016 to 2025.
Methodology:This study used Sentinel-2 satellite imagery (Level L2A) for three time periods: 2016, 2021, and 2025. Supervised classification was performed using Support Vector Machine (SVM) algorithm with Radial Basis Function (RBF) kernel in six land use classes. Accuracy assessment was conducted using confusion matrix, Kappa coefficient, and overall accuracy. Change analysis was extracted using Compute Change Raster tool and change matrix. Hot spot analysis was performed using Getis-Ord Gi* algorithm with a neighborhood distance of 100 meters. For spatial autocorrelation analysis, global and local Moran's indices were used in Python environment with PySAL library.
Findings and Conclusion:The Kappa coefficient of classification for the periods 2016, 2021, and 2025 was 0.927, 0.957, and 0.929, respectively, and overall accuracy for all three periods was above 94%. The results showed that forest area decreased from 7,377 hectares in 2016 to 5,224 hectares in 2025 (29.2% decrease). The change matrix for the 2016–2021 period indicates that the largest conversion flow in the forest class was to residential (1,871.73 hectares) and in the pasture class was to agriculture (1,437.07 hectares). The apparent decrease in the area of the residential class in the classified data is mainly due to the transfer of 5,210.69 hectares of residential land in 2016 to the NoData extent in 2021 and the dispersion of villa constructions in host classes. Hot spot histogram analysis indicated a transition from a dispersed and uniform pattern in the first period (2016–2021) to a bimodal and concentrated pattern in the second period (2021–2025), followed by neutralization of the polarization effect in the long-term period (2016–2025). The global Moran's index for all three periods showed a significant clustered pattern at the 99% confidence level (Moran's I: 0.525, 0.527, and 0.519, respectively). Masal and Shanderman region is transitioning from an agricultural-pastoral region to a residential-villa region with elements of ecosystem degradation. This transition has been accompanied by deforestation and pasture degradation, increased barren land, and dispersed construction development. The findings emphasize the necessity of revising land use management policies and adopting approaches based on ecosystem sustainability.