نوع مقاله : مقاله پژوهشی
نویسندگان
1 گروه فناوریهای محیطزیست، پژوهشکده علوم محیطی، دانشگاه شهید بهشتی، تهران، ایران
2 گروه اگرواکولوژی، پژوهشکده علوم محیطی، دانشگاه شهید بهشتی، تهران، ایران
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
Introduction: Mangrove forests are among the richest coastal ecosystems in tropical and subtropical regions, possessing high ecological and economic value. Despite their immense importance, these forests are being degraded and destroyed at an alarming rate worldwide. In recent decades, extensive restoration projects, including the planting of mangrove seedlings by industries in the Bandar Mahshahr Special Economic Zone, have been initiated to develop these green areas. However, comprehensive and up-to-date studies on the dynamics of these ecosystems in the Mahshahr tidal creeks (Khooriats) are limited. The present study was conducted with the aim of investigating changes in mangrove forests within the Bandar Mahshahr Special Economic Zone from 2003 to 2023 and producing an accurate classification map of these forests for the year 2023. The results of this study can provide a basis for future conservation planning and evaluating the effectiveness of restoration projects in the region.
Material and methods: To monitor the spatiotemporal changes of mangrove forests during the 2003–2023 period, Landsat (series 7 and 8) satellite images under low tide and minimal cloud cover conditions were used. Four spectral indices (NDVI, NDWI2, MVI, and CMRI) were calculated for three time points (2003, 2013, and 2023). By calibrating detection thresholds based on a baseline map for 2023, distribution maps for each year were generated. The final change map was extracted by integrating these four indices and applying a majority rule. For producing the precise classification map for 2023, Sentinel-2 images and a dataset comprising spectral bands and the calculated indices were utilized. Training samples for the mangrove class were collected from a baseline map provided by the petrochemical company, and samples for other classes were gathered through visual interpretation of high-resolution imagery. A Random Forest classification model was trained and optimized. The accuracy of the final map was evaluated using an error (confusion) matrix, and the area of each class was calculated. All stages of image processing, index extraction, classification, and accuracy assessment were performed in the R software environment.
Results and discussion: Long-term change monitoring using Landsat data showed that the planted mangrove forests in the area have developed from approximately 2 hectares in 2003 to 109.5 hectares in 2013 and further to 221.6 hectares in 2023. This net growth of 219.6 hectares over two decades confirms the success of the restoration projects. The final map produced using Sentinel-2 data and the Random Forest algorithm, with an overall accuracy of 94.78%, revealed a more precise area of 630.8 hectares for established mangrove forests in 2023. A key methodological finding was the discrepancy of approximately 55% between the Landsat-based estimate (285 hectares) and the Sentinel-2-based estimate for the mangrove forest area in 2023. This gap reveals the limitation of moderate-resolution data in identifying small, and scattered patches and the systematic underestimation error it causes. Comparison with previous studies that used simpler methods or lower-resolution data indicates that the multi-index, multi-class integrated approach of this research, by increasing spectral discrimination power, has significantly improved identification accuracy and enabled better separation of mangroves from mudflats and similar coastal vegetation.
Conclusion: By employing a novel integrated framework, this study provided an accurate and reliable estimate of the extent of planted mangrove forests in the Mahshahr Special Economic Zone and demonstrated that reliance on moderate-resolution data can lead to underestimation. The high-accuracy map produced will serve as a reliable baseline for future monitoring, assessment of ecosystem services, and sustainable management of these valuable reserves in industrial–coastal environments. To complete this framework, continuous monitoring using time series and integration of auxiliary data such as LiDAR for forest structure analysis is proposed as a future research path.
کلیدواژهها [English]
)