بررسی خطر وقوع بیابان‌زایی با استفاده از شاخص‌های طیفی در محدوده‌ی پیرامونی دریاچه‌ی ارومیه

پژوهش حاضر با هدف بررسی وقوع بیابان زایی در محدوده ی پیرامون دریاچه ی ارومیه انجام شده است. بدین منظور در ابتدا، تصاویر ماهواره ی سنتینل2 با استفاده از نرم افزار qgis مورد پیش پردازش قرار گرفته و پس از انجام تصحیحات اتمسفری، اقدام به استخراج شاخص های طیفی نشانگر بیابان زایی (پوشش گیاهی تفاضلی نرمال شده (ndvi)، آلبدوی سطحی، میزان نمناکی (wetness)، ضریب روشنایی (brightness)، میزان سبزینگی (greenness) شد. پس از استخراج شاخص های طیفی مذکور و در جهت شناسایی مناسب ترین زوج شاخص های طیفی، میزان همبستگی و رابطه ی رگرسیونی موجود بین شاخص های مورد مطالعه با استفاده از تحلیل های آماری صورت پذیرفته در نرم افزارspss(22)  بررسی شد. بر طبق نتایج حاصل، میزان همبستگی برای زوج شاخص (میزان سبزینگی ضریب روشنایی (برابر با 9/4 و برای زوج شاخص (میزان نمناکی ضریب روشنایی) برابر با 33/0 می باشد. در مرحله ی بعد نقشه ی خطر بیابان زایی بر اساس دو زوج شاخص مذکور تهیه و با استفاده از الگوریتم  jenks natural break  در محیط نرم افزار arcgis 10.6 در پنج کلاس خطر شدید، نسبتا شدید، متوسط، ضعیف و بدون خطر بیابان زایی، طبقه بندی شد. نتایج نشان داد که 89/9 درصد از کل مساحت محدوده ی مورد مطالعه در کلاس خطر شدید، 60/30 درصد در کلاس خطر نسبتا شدید، 48/37 درصد در کلاس خطر متوسط، 42/12 درصد در کلاس خطر ضعیف و 61/9 درصد در کلاس خطر بدون بیابان زایی قرار دارد. نتایج به دست آمده با استفاده از مشاهدات میدانی و ماتریس خطا ( confusion matrix using ground truth roi ) ارزیابی و با کسب ضریب کاپا 95/0 و درجه ی صحت 51/90 درصد مورد صحت سنجی قرار گرفت.

Evaluating the risk of desertification using the spectral indices in the surrounding area of Lake Urmia

Extended Abstract IntroductionDesertification should be considered as the destruction of the fragile balance generating human, plant, and animal life in arid, semiarid, and subhumid arid areas.During the recent decade, Urmia catchment and the surrounding area of Lake Urmia have encountered imbalance by experiencing severe environmental fluctuations. Thus, due to the ecological and ecosystemic significance of this area, studying the occurrence of desertification seems essential. Thus, the present study including:1. the evaluation of the occurrence of desertification in the surrounding area of Lake Urmia using the spectral indices of vegetation, Albedo, tasseled cap, 2) Identifying a base pixel relationship between different biophysical indices (normalized vegetation difference, greenery rate, humidity rate, surface Albedo, brightness degree) and detection are among the best study pairs for evaluating the status of desertification due to the highest negative correlation among them.The studied areaThe studied is part of Urmia catchment located on northwestern Iran with geographical coordinates of 44 degrees and 0 minute to 47 degrees and 0minute east longitude and also 37 degrees and 0 minute to 38 degrees and 20 minutes’ north latitude and has an area of 14395 square kilometers.MethodologyIn the first step, Satellite images of Sentinel2 (Level1C) were downloaded from Copernicus Open Access Hub (https://scihub.copernicus.eu). The images were acquired on July 2018. The cloud free images were selected on July because it is the period when natural and annual vegetation is minimal and crops are harvested. Desertification during this period is best assessed to avoid confusion with seasonal vegetation. In the second step the images were preprocessed and processed using QGIS software. Then, ArcGIS 10.3, SAGAGIS, and SPSS (22) software were used for the statistical regression analysis between vegetation and Albedo (the first pair of spectral index), greenery and brightness coefficient (the second pair of spectral index), as well as humidity rate and brightness coefficient (the third pair of spectral index) were used to identify the pair of spectral indices with higher negative correlation. Results and discussionBased on the results, the high rate of NDVI index and greenery rate are related to the points like the Alluvial fans located at the foot of Misudagh hillside. However, the low rate of the abovementioned indices is related to the areas like aquatic zones (Lake Urmia). The Albedo rate in the studied area is between 0.01 to 0.9. The low values of Albedo index are related to the areas full of vegetation and aquatic bodies. The high values of Albedo and brightness rate are related to bright soils.The high rates of humidity are related to aquatic bodies, the areas with vegetation, and the humid lands around Lake Urmia while its low values are related to the soils with bright texture, the land without vegetation, and the lands and salt marsh without humidity resulted from the retrogression of Lake Urmia and poor soils without organic materials.In order to evaluate the regression relationship between the spectral indices, the SAGAGIS was used. The analysis showed a strong negative correlation between the spectral index pair of humidity rate and brightness coefficient (r=0.37).After the spectral index pair of brightness coefficienthumidity rate, the spectral index pair of normalized difference VegetationAlbedo is equal to r=0=1.14.The negative correlation is between the spectral index rate of brightness coefficient greenery rate (r=4.9) allocating higher rate to itself than the previous spectral index pair.Thus, first the spectral index pair of brightness coefficient and humidity rate as well as normalized difference vegetation and Albedo was normalized. ConclusionAfter normalizing the two abovementioned index pairs, the correlation coefficient for normalized vegetation Albedo is equal to 0.34 and for humidity ratebrightness coefficient is equal to 0.31. Due to the close correlation obtained for the abovementioned indices, the map for desertification risk was prepared based on the two abovementioned pairs and then verified using the Confusion Matrix using Ground truth ROI. Kappa coefficient obtained for the map is equal to 0.89 and the obtained accuracy rate is equal to 90.1. The achieved desertification map was classified in five categories of without risk, poor, average, relatively severe, and severe. From the total 14350.81 square kilometers of the studied area, 557.391 square kilometers (3.9%) is at the severe risk of desertification, 2015.330 square kilometers (14.04$) is at the relatively severe risk, 4007.073 square kilometers (27.92%) is at the average risk, 3660.534 square kilometers (25.50%) is at the poor risk and 4110.473 square kilometers (28.64%) is at the class of no risk for desertification .Based on the abovementioned issues, any effort for exploiting the lands located on the studied area should be made with full cation and based on sufficient knowledge on the conditions of these lands.