تهيۀ نقشۀ نوع محصول كشاورزي از تركيب تصاوير ماهواره‌هاي سنتينل-1 و سنتينل-2 با استفاده از مفهوم يادگيري انتقالي (مناطق مطالعاتي: شهر اراك و دشت مرودشت)

Crop Mapping From the Combination of Sentinel-1 an‎d Sentinel-2 Images Using Transfer Learning Concept (Study Area: Arak & Marvdasht)

مهندسي نقشه برداري

Monitoring an‎d identifying agricultural crop types is one of the fundamental requirements in resource management, crop planning, water allocation, an‎d macro-level agricultural policymaking. In recent years, remote sensing data, particularly satellite imagery with high spatial resolution an‎d suitable temporal frequency, have provided the possibility of producing accurate an‎d up-to-date maps of cultivated areas. This study aims to eva‎luate the capability of satellite data for crop type classification an‎d to analyze the performance of transfer learning in both spatial an‎d temporal dimensions, with the goal of generating crop-type maps for the study regions. The first study area, Arak, located in Markazi Province, is characterized by a cold semi-arid climate with relatively limited crop diversity. The second area, Marvdasht in Fars Province, has a temperate mountainous climate an‎d higher crop diversity. These two regions were selec‎ted due to their climatic an‎d agricultural differences, allowing a robust eva‎luation of transfer learning generalization. In the spatial transfer experiment, a combination of optical an‎d radar data, including Sentinel-2 (ban‎ds 3, 4, an‎d 5) an‎d Sentinel-1 (VV an‎d VH polarizations), from the 2016–2017 cropping year was used for the two regions. In the temporal transfer experiment, Lan‎dsat 8 imagery from the 2014–2015 cropping years was employed to assess the transferability of models trained in one agricultural year to the next within the same region. Spectral indices including NDVI, MNDWI, an‎d BUI were used as input features for the models. All stages of preprocessing, feature extraction, an‎d model implementation were performed in the Google Earth Engine (GEE) cloud platform, which enabled large-scale data processing an‎d model comparison. The Ran‎dom Forest (RF), Decision Tree (DT), an‎d Support Vector Machine (SVM) algorithms were implemented under four scenarios: supervised an‎d unsupervised learning, with an‎d without Principal Component Analysis (PCA). The results indicated that in spatial transfer learning, transferring the model from Arak to Marvdasht achieved higher accuracy than the reverse direction, with the RF algorithm combined with PCA yielding the best performance. Conversely, in temporal transfer learning, applying PCA—especially in the unsupervised mode—led to a reduction in accuracy due to the loss of certain phenological features of crops. Overall, the RF algorithm demonstrated superior stability an‎d generalization capability compared with DT an‎d SVM in all scenarios. These findings highlight that integrating multi-source satellite data, applying transfer learning, an‎d utilizing the cloud-based processing capabilities of Google Earth Engine can serve as an effective approach for near real-time crop monitoring an‎d reducing dependence on extensive field data across different agricultural regions.

6

153006