Spatiotemporal Analysis of Cropping Pattern Changes Using Sentinel-2 Data and Support Vector Machine Classification in Varamin Plain, Iran | ||
| فصلنامه علوم محیطی | ||
| مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 28 اردیبهشت 1405 | ||
| نوع مقاله: مقاله پژوهشی | ||
| شناسه دیجیتال (DOI): 10.48308/envs.2025.240747.1523 | ||
| نویسندگان | ||
| جعفر کامبوزیا* ؛ آزاده قنبرزاده قصبه؛ فاطمه آقامیر | ||
| چکیده | ||
Introduction: The increasing strain on water and land resources, combined with recurring droughts and unsustainable farming practices, has posed serious challenges to agricultural sustainability in Iran’s semi-arid regions. Remote sensing techniques offer a powerful, cost-effective, and repeatable method for monitoring changes in cropping patterns. The Sentinel-2 satellite, with its high spectral resolution and five-day revisit time, provides an ideal data source for such studies. The Varamin Plain, located in southeastern Tehran, represents a critical agricultural zone affected by water scarcity and policy restrictions on crop cultivation. Understanding how cropping patterns have changed over time in this area can help inform adaptive agricultural planning. In recent years, intensive agricultural activities combined with climate fluctuations have significantly transformed cropping systems, especially in semi-arid regions such as Varamin Plain. Accurate monitoring of these changes using advanced satellite imagery is critical to ensure agricultural sustainability and effective water management at both local and regional scales. Materials and Methods: This study utilized Sentinel-2 imagery from the years 2016, 2018, 2020, and 2022. Images were sourced from USGS and Copernicus databases and were preprocessed in ENVI 5.3 using QUAC atmospheric correction. Monthly NDVI maps were generated and stacked to form annual composites. Four crop classes were defined: Garden, rainfed, irrigated, and rice. Classification was performed using the Support Vector Machine (SVM) algorithm. Training data were derived from field surveys, Google Earth imagery, and local agricultural records. Accuracy assessments were carried out using confusion matrices, calculating user accuracy, producer accuracy, overall accuracy, and Kappa coefficient. GIS techniques were employed to calculate class areas and visualize transitions using Sankey diagrams. Results and Discussion: Overall classification accuracy was consistently above 80% in all four years, with Kappa values ranging from 0.74 to 0.77. The 2018 classification had the highest accuracy. Sankey diagrams revealed that the most significant transitions occurred between irrigated and rain fed lands, reflecting farmers’ adaptive responses to climatic variability and water availability. Notably, rice cultivation dropped significantly in 2020 due to groundwater depletion and local restrictions on high-consumption crops. These observations align with findings from other semi-arid areas of Iran, where reduced rainfall and policy interventions have led to shifts in cropping strategies. These outcomes correspond closely with similar studies in semi-arid regions, highlighting farmers’ common adaptive responses to drought conditions. However, limitations such as insufficient phenological ground data were also noted. Methodologically, Sentinel-2 imagery—particularly its Red-Edge bands—proved highly effective in distinguishing crop types, while SVM maintained robustness across datasets. However, some limitations were noted, including the absence of phenological ground truth data and the biennial analysis approach, which may overlook within-year crop changes or mixed cropping systems. Conclusion: The findings underscore the potential of combining Sentinel-2 imagery with SVM classification for effective agricultural monitoring in semi-arid regions. This approach can be effectively applied to similar semi-arid regions, potentially enhancing water management and agricultural policy decisions. This methodology can support evidence-based decision-making for water resource management, crop planning, and climate adaptation strategies. To further improve classification accuracy and temporal coverage, future studies should incorporate ground-based phenological data and higher-frequency satellite imagery. Keywords Adaptation, Agricultural transition, Classification accuracy, Crop dynamics, Sustainability. | ||
| کلیدواژهها | ||
| Adaptation؛ Agricultural transition؛ Classification accuracy؛ Crop dynamics؛ Sustainability | ||
| عنوان مقاله [English] | ||
| Spatiotemporal Analysis of Cropping Pattern Changes Using Sentinel-2 Data and Support Vector Machine Classification in Varamin Plain, Iran | ||
| نویسندگان [English] | ||
| Jafar Kambouzia؛ Azade Ghanbarzade Ghasabe؛ Fateme Aghamir | ||
| چکیده [English] | ||
| Extended Abstract Introduction The increasing strain on water and land resources, combined with recurring droughts and unsustainable farming practices, has posed serious challenges to agricultural sustainability in Iran’s semi-arid regions. Remote sensing techniques offer a powerful, cost-effective, and repeatable method for monitoring changes in cropping patterns. The Sentinel-2 satellite, with its high spectral resolution and five-day revisit time, provides an ideal data source for such studies. The Varamin Plain, located in southeastern Tehran, represents a critical agricultural zone affected by water scarcity and policy restrictions on crop cultivation. Understanding how cropping patterns have changed over time in this area can help inform adaptive agricultural planning. In recent years, intensive agricultural activities combined with climate fluctuations have significantly transformed cropping systems, especially in semi-arid regions such as Varamin Plain. Accurate monitoring of these changes using advanced satellite imagery is critical to ensure agricultural sustainability and effective water management at both local and regional scales. Materials and Methods This study utilized Sentinel-2 imagery from the years 2016, 2018, 2020, and 2022. Images were sourced from USGS and Copernicus databases and were preprocessed in ENVI 5.3 using QUAC atmospheric correction. Monthly NDVI maps were generated and stacked to form annual composites. Four crop classes were defined: Garden, rainfed, irrigated, and rice. Classification was performed using the Support Vector Machine (SVM) algorithm. Training data were derived from field surveys, Google Earth imagery, and local agricultural records. Accuracy assessments were carried out using confusion matrices, calculating user accuracy, producer accuracy, overall accuracy, and Kappa coefficient. GIS techniques were employed to calculate class areas and visualize transitions using Sankey diagrams. Results and Discussion Overall classification accuracy was consistently above 80% in all four years, with Kappa values ranging from 0.74 to 0.77. The 2018 classification had the highest accuracy. Sankey diagrams revealed that the most significant transitions occurred between irrigated and rain fed lands, reflecting farmers’ adaptive responses to climatic variability and water availability. Notably, rice cultivation dropped significantly in 2020 due to groundwater depletion and local restrictions on high-consumption crops. These observations align with findings from other semi-arid areas of Iran, where reduced rainfall and policy interventions have led to shifts in cropping strategies. These outcomes correspond closely with similar studies in semi-arid regions, highlighting farmers’ common adaptive responses to drought conditions. However, limitations such as insufficient phenological ground data were also noted. Methodologically, Sentinel-2 imagery—particularly its Red-Edge bands—proved highly effective in distinguishing crop types, while SVM maintained robustness across datasets. However, some limitations were noted, including the absence of phenological ground truth data and the biennial analysis approach, which may overlook within-year crop changes or mixed cropping systems. Conclusion The findings underscore the potential of combining Sentinel-2 imagery with SVM classification for effective agricultural monitoring in semi-arid regions. This approach can be effectively applied to similar semi-arid regions, potentially enhancing water management and agricultural policy decisions. This methodology can support evidence-based decision-making for water resource management, crop planning, and climate adaptation strategies. To further improve classification accuracy and temporal coverage, future studies should incorporate ground-based phenological data and higher-frequency satellite imagery. Keywords: Adaptation, Agricultural transition, Classification accuracy, Crop dynamics, Sustainability. | ||
| کلیدواژهها [English] | ||
| Adaptation, Agricultural transition, Classification accuracy, Crop dynamics, Sustainability | ||
| مراجع | ||
|
| ||
|
آمار تعداد مشاهده مقاله: 313 |
||