كليدواژه :
دماي سطح زمين , پنجره مجزا , ادغام گسيل مندي ها , ماديس
چكيده فارسي :
آب و هوا، عامل اصلي تغييرات دماي سطح زمين است و دماي سطح زمين يك متغير مهم در مطالعات اقليمي و زيست محيطي محسوب مي شود و محاسبه دماي سطح زمين با استفاده از داده هاي سنجش از دور نيازمند محاسبه دقيق ضريب گسيل مندي مي باشد. در بحث مطالعات حرارتي نياز به تصاوير ماهواره اي با قدرت تفكيك بالا مي باشد چون در مناطق شهري به دليل تراكم انواع كاربري ها استخراج اطلاعات در پيكسل هاي مختلط كار دشواري خواهد بود. در اين تحقيق از تصاوير ماهواره ترا، محصولات سنجنده ماديس (MOD021KM,MOD11A1,MOD05) مربوط به سال 2012 - 2013 و تصاوير لندست TM5- مربوط به سال 2010 و داده هاي ساعتي هواشناسي استفاده گرديد. در اين پژوهش به منظور محاسبه دماي سطح زمين از روش پنجره مجزا و قانون پلانك استفاده گرديد و همچنين روش هاي مختلف استخراج گسيلمندي شامل روش طبقه بندي، روش شاخص تفاضل گياهي نرمال شده، روش ضريب گسيل مندي نرمال به منظور استخراج گسيلمندي بكارگرفته شد تا در كنار محصولات گسيل مندي ماديس (MOD11A1) به منظور تهيه نقشه دماي سطح زمين مورد مقايسه قرارگيرند. در اين راستا امكان ادغام داده هاي گسيل مندي مستخرج از روش هاي مختلف نيز مورد بررسي قرارگرفت. نتايج بدست آمده نشان مي دهد روش شاخص تفاضل گياهي نرمال شده و روش تركيب با استفاده از ميانه گيري داده هاي گسيلمندي مناسبي را براي محاسبه دماي سطح زمين فراهم مي آورند. همچنين در محاسبه دماي سطح زمين، روش پلانك در باند 31 و 32 دقت بالاتري را نسبت به روش پنجره مجزا ايجاد مي نمايد.
چكيده لاتين :
Land surface temperature (LST) is one of the most important variables required in environmental and climatological studies. In order to calculate LST, accurate emissivity is needed. Recently, several methods have been developed to calculate LAST and emissivity. Some of these methods estimate LST based on a pre-known emissivity, while the others calculate LST and emissivity, simultaneously. LST mapping in urban areas can be difficult due to the high variation of the land cover and the formation of mixed pixels. Accordingly, the LST calculation based on the emissivity derived from a single method can be erroneous, especially using a low spatial resolution image in the urban areas. Integration of the emissivity values derived from different methods seems to be an effective solution in this situations. In this study, LST was calculated using Split Window and Planck Law methods for Tehran city. Three different methods including classification, normalized difference vegetation index (NDVI)-based method, and normalization emissivity method were applied to derive emissivity from MODIS images. NDVI-based method is a common method used NDVI thresholding to determine the emissivity of different pixels. In classification method, each pixel is classified into one of 14 classes for which the emissivity is known. Normalization emissivity method assumes a constant value as emissivity for a pixel in different bands to calculate temperature for these bands and then the maximum temperature derived through, is used for calculation of emissivity coefficients which are used for actual LST calculation using Planck function. In addition, MODIS emissivity product (MOD11A1) was used to compare with the emissivity derived from the other methods. In order to implement this study, the remotely sensed data including Landsat-TM data acquired in 2010, and MODIS products (MOD021KM, MOD05, MOD11A1) acquired in 2012 to 2013 were downloaded. Temperature data measured by three meteorological stations around Tehran were provided to validate the results. In order to integrate the emissivity values, averaging and median methods were used to fuse the emissivity values derived from three methods and MODIS emissivity product. The results showed that NDVI-based method produces more accurate emissivity as the LST calculated based on this emissivity was more accurate than that derived from other emissivity values. Fusing the emissivity values through mean and median methods, the fused emissivity values were used for calculating LST using Planck’s equation and Split Window methods. It was shown that the fused emissivity derived from averaging method can improve the accuracy of the LST maps derived from each emissivity method. Moreover, Planck Law performed better for calculating LST using MODIS bands 31 and 32 with error of 1.6 and 1.63 Kelvin degrees, respectively, compared to that derived from Split Window method.
