Iranian Jornal of Watershed Management Science&Engineering

fa سنجش تخریب اکوسیستم‌های ماندابی در حوزه‌های آبخیز ساحلی با استفاده از داده‌های ماهواره‌ای Assessing the Degradation of Wetland Ecosystems in Coastal Watersheds Using Satellite Data تخصصي Special كاربردي Applicable جنگل‌های مانگرو، مجموعه‌ای از درختان همیشه‌سبزی هستند که در سواحل حوزه‌های آبخیز، به‌ویژه در مصب‌ها و خورهای مجاور دریا رشد می‌کنند. این بوم‌سازگان ارزشمند به دلیل اهمیت بالای زیستی و حساسیت زیاد نسبت به آسیب‌های محیطی، نیازمند حفاظت و مدیریت مؤثر می‌باشند. در این پژوهش، جنگل‌های مانگرو تیاب و کلاهی، واقع در شهرستان میناب از توابع استان هرمزگان با مساحت تقریبی 31/126 کیلومترمربع، به منظور بررسی وضعیت تخریب این بوم‌سازگان‌های ارزشمند انتخاب شدند. تصاویر ماهواره‌ای سنتینل-2 مربوط به سال‌های 2019 و 2024 از طریق سامانه Google Earth Engine استخراج و با استفاده از شاخص‌های پوشش گیاهی NDVI، EVI و MVI مورد تحلیل قرار گرفتند. نتایج حاصل از تحلیل شاخص‌های پوشش گیاهی نشان داد که در سال 2024، مساحت جنگل‌های مانگروی با طراوت براساس شاخص‌های NDVI، EVI و MVI به ترتیب 07/4، 31/2 و 59/4 درصد برآورد شده است. این مقادیر در مقایسه با سال 2019 به ترتیب افزایش 98/0، 74/0 و 39/1 درصدی را نشان می‌دهند. همچنین در همین بازه زمانی، مساحت مانگروهای بدون ‌طراوت نیز بر اساس شاخص‌های مذکور به ترتیب 6/0 درصد (NDVI)، 03/0 درصد (EVI) و 85/0 درصد (MVI) افزایش یافته است. به منظور بررسی عدم قطعیت طبقه‌بندی از شبیه‌سازی مونت-کارلو و روش IQR استفاده شد. نتایج مونت-کارلو افزایش دقت کلی مدل از 67/0 درصد در سال 2019 به 73/0 درصد در سال 2024 را نشان داد. نتایج IQR عملکرد خوب NDVI و ثبات MVI را نشان داد. با وجود این، شاخص EVI به دلیل پایداری و حساسیت بالا نسبت به پوشش متراکم، عملکرد بهتری در شناسایی جنگل‌های مانگرو دارد. این مطالعه بر اهمیت انتخاب شاخص مناسب و استفاده از عدم قطعیت در تحلیل صحت و دقت کلاس های مختلف حاصل از طبقه‌بندی آن‌ها در پایش بوم‌سازگان‌های ساحلی تاکید دارد. Extended Abstract Introduction Mangrove forests, consisting of evergreen trees that grow along the coasts of adjacent sea watersheds, estuaries, are one of the most productive ecosystems in the world. They provide food for humans and wildlife and also play a major role in protecting and stabilizing coastlines, preventing soil erosion and sequestering carbon. Mangroves are suitable habitats for the reproduction of a variety of fish, crabs, amphibians, mammals, birds and arthropods. Due to the increasing pressure and stress caused by human activities, the destruction of mangrove forests has been accelerated. Therefore, monitoring the process and estimating the extent of destruction of these ecosystems provides a comprehensive view for their restoration and protection. While field monitoring of mangrove forests is difficult and costly, recent advances in access to remote sensing data, image processing, information technology and computing, as well as advancements in human technology, have provided an opportunity for continuous and systematic monitoring of mangrove forests. Platforms such as Google Earth Engine (GEE) provide access to satellite imagery and the ability to analyze spatiotemporal changes. These platforms can be used to calculate and analyze vegetation indices. In this study, vegetation indices were classified using ground-based reference data and matched with satellite data, and the accuracy of each index in the classification was estimated. Materials and Methods In the present study, by comparing ground reference data and satellite data in 2019 and 2024, an attempt has been made to provide a reliable classification to show the ecological status and health status of mangrove trees in the Tyab and Kolahi regions with an area of 126.31 square kilometers. In this regard, Sentinel-2 satellite images with a spatial resolution of 10 meters were used through the Google Earth Engine platform and three widely used vegetation indices in previous research in the field of vegetation analysis and assessment were used, including NDVI (Normalized Difference Vegetation Index), EVI (Enhanced Vegetation Index), and MVI (Mangrove Vegetation Index). Ground reference data were obtained via field surveys to the area. Image preprocessing steps, including atmospheric correction and removal of cloud and shadow effects, were performed using the Sen2Cor algorithm in the GEE environment. The extracted data were matched with the field data, thus providing a valid and reliable set of training samples for image classification. Using field data and connecting them to the spectral ranges of the indicators significantly improved the classification accuracy and increased the ability to distinguish different land covers. However, various factors such as measurement error, atmospheric conditions, solar zenith angle variations, and computational errors during index calculation the indices can affect the accuracy of satellite data. Therefore, uncertainty was investigated using Monte Carlo simulation and IQR methods. Results and Discussion The classification was designed based on the field observations and using ground control points, in the ArcMap environment in such a way that it can identify and differentiate between water, soil, and stressed vs. healthy mangroves. The results of the analysis of the indicators indicate significant changes in the amount of vegetation cover in the study area during the years 2019 and 2024. The results obtained from the classification of mangrove tree cover in ArcMap environment showed that in 2024, compared to 2019, there was an increase of 0.98% (NDVI), 0.74% (EVI) and 1.39% (MVI) of fresh trees. In addition, the cover of non-fresh mangrove trees increased by 0.6% (NDVI), 0.03% (EVI) and 0.85% (MVI) during the same period. The MVI index detected the highest coverage of fresh and non-fresh cover, which can be attributed to its high sensitivity to specific mangrove covers. Both Monte Carlo and IQR simulation methods were used to assess classification uncertainty. Monte Carlo results showed a decrease in overall model accuracy from 71.4% in 2019 to 62.5% in 2024, likely attributable to environmental variability or spectral class overlap. IQR analysis showed strong performance of NDVI and high uncertainty associated with MVI. However, EVI showed superior performance in identifying mangrove forests due to its stability and high sensitivity to dense vegetation. Conclusion In this study, using satellite data and ground reference data, an attempt was made to quantify ecological changes in mangrove forests. Three indices, MVI, EVI, and NDVI, were selected, and by comparing the results obtained from these three indices with ground reality, the best index for examining ecological changes was introduced. Monte Carlo and IQR uncertainty methods were used to examine the accuracy of the classifications, and the findings showed that EVI is more suitable for continuous monitoring, while NDVI is preferable for baseline assessments. MVI can serve as a complementary indicator in certain situations. All three indicators show a decrease in vegetation cover from 2019 to 2024, indicating heightened anthropogenic pressure. To improve the obtained results, more advanced algorithms such as Random Forest, SVM or Deep Learning networks can be used to increase the accuracy of the model. Supplementary data from higher resolution sensors could improve the separation of classes with high overlap. Sensitivity analysis and assessment of sources of uncertainty should be expanded to select appropriate indicators in specific regional conditions. By providing a detailed analysis of the process of mangrove destruction, this study emphasizes the need for continuous monitoring and the use of selected indicators appropriate to regional characteristics, and establishes a framework for conservation policies in estuarine areas in coastal watersheds.   جنگل‌های مانگرو, شاخص‌های پوشش گیاهی, روش مونت-کارلو, IQR,Sentinel-2 ,Google Earth Engine Mangrove forests, vegetation indices, Monte Carlo method, IQR, Sentinel-2, Google Earth Engine. 93 111 http://jwmsei.ir/browse.php?a_code=A-10-1723-1&slc_lang=fa&sid=1 Laleh sharifipour لاله شریفی پور l.sharifipour.phd@hormozgan.ac.ir 100319475328460018970 0009-0003-6982-274X No University of Hormozgan دانشگاه هرمزگان Marzieh Rezai مرضیه رضایی m.rezai@hormozgan.ac.ir 100319475328460018971 100319475328460018971 Yes University of Hormozgan دانشگاه هرمزگان Mohammad Kazemi محمد کاظمی mohamad.kazemi86@gmail.com 100319475328460018972 100319475328460018972 No University of Hormozgan دانشگاه هرمزگان Ali Reza Nafarzadegan علیرضا نفرزادگان a.r.nafarzadegan@hormozgan.ac.ir 100319475328460018973 100319475328460018973 No University of Hormozgan دانشگاه هرمزگان Rasool Mahdavi رسول مهدوی نجف‌آبادی ra_mahdavi2000@hormozgan.ac.ir 100319475328460018974 100319475328460018974 No University of Hormozgan دانشگاه هرمزگان