مقایسه دقت داده های ماهواره لندست9 و سنتینل3 در برآورد دمای سطح زمین (مطالعه موردی: شهرستان اردبیل)

The Land surface temperature (LST) is an important Quantity that affects the cycle of energy and moisture between the earth's surface and the atmosphere. (xin et al, 2023). In fact, land surface temperature is an important index related to climatic, meteorological, hydrological, and environmental phenomena and processes and is widely used in climate change investigation, hydrological process modeling, drought monitoring, and fire risk assessment (Yang et al, 2020). One of the methods of measuring temperature is meteorological stations, which are less used due to problems such as lack of proper spatial distribution and difficult access to data, especially on a large scale. (Heidari and Akhundzadeh, 2019). Thermal infrared remote sensing is an efficient approach in this field and has attracted the attention of researchers. (Yi et al, 2023). Considering that Ardabil city, located in the northwest of the country and on the foothills of Sablan mountain, is considered one of the big cities of Iran and is affected by climate changes, including the increase in temperature (Malkian et al, 2018), it is necessary to investigate climatic factors, including temperature, in this region. The purpose of this study is to investigate the temperature in different land uses by LANDSAT9 satellites by applying the split window (SW) algorithm and Sentinel3 using the SLSTR sensor and its LST tool. Checking the temperature using two different images, by applying different algorithms and checking these temperatures in different uses in the study area can be considered as an innovation of the present study.

Materials and methods

Ardabil city with an area of 2165 square kilometers is one of the big cities of Iran and the capital of Ardabil province. In this study, Landsat 9 and Sentinel 3 images from 20/7/2022 have been used to extract the land surface temperature (LST). For this purpose, the split window algorithm was implemented on the Landsat 9 image, and the LST algorithm of the SLSTR sensor was implemented on Sentinel 3. To check the results related to the temperature, the data from the meteorological station has been used. Also, the land use map of the region has been extracted using the Landsat 9 image on 5 residential and industrial floors, vegetation, agricultural land, water areas, and soil cover using the object-oriented technique to check the temperature in different uses.

Results and discussion

The land use map of the studied area was extracted with overall accuracy and a kappa coefficient of 98 and 97%, respectively. Investigations showed that in Landsat image 9 agricultural land use with an average temperature of 35 degrees Celsius is the minimum temperature and soil cover use with an average temperature of 42 degrees Celsius is the maximum temperature. In the image of Sentinel 3, in the same way, agricultural land use and soil cover have minimum and maximum temperatures of 37 and 42 degrees Celsius, respectively. The minimum and maximum temperatures obtained by both sensors show close values with a difference of one degree in the maximum and 3 degrees in the minimum temperature. The temperature of the station related to Landsat 9 and Sentinel 3 is estimated to be 35 and 36 degrees, respectively, which are close to each other and show a difference of one degree. However, the temperatures obtained from the Landsat 9 sensor have a smaller difference from the temperature of the meteorological station, and this Temperature compliance is especially noticeable in the minimum temperature with a difference of 10 degrees Celsius. It should be mentioned that the spatial resolution of 30 meters of Landsat 9 and 1 km of Sentinel 3, also the location of the weather station should be considered.

Conclusion

The minimum and maximum temperature has been estimated using the split window algorithm on the Landsat 9 image, 22 and 51 degrees Celsius, and using the SLSTR sensor on the Sentinel 3 image, 25 and 52 degrees Celsius. The split window algorithm has high accuracy and capability and is considered one of the most effective and widely used algorithms in extracting land surface temperature in many studies (Grace et al, 2020), (Eon et al, 2023), (Zhang et al, 2019). SLSTR is also a high-precision infrared radiometer and enables more accurate LST measurements using common algorithms as well as the development of new algorithms (Kuppo et al, 2016), (Sobrino et al, 2015). The obtained temperatures show that reasonable and close values were obtained by both sensors and are by different uses. The results of this research show that the temperature calculated using the satellite images of both sensors is consistent with the data obtained from the meteorological station, however, the temperatures obtained from Landsat 9 show a greater agreement with these obtained data, especially in the minimum temperature. Regarding the maximum temperature and the station temperature, the difference between the Landsat 9 temperature and the station temperature is one degree less than the Sentinel 3 temperature. LST for each pixel in remote sensing is equivalent to the average temperature of different earth surface covers (Alavi Panah, 2017), Therefore, it seems that the high spatial and radiometric resolution of the Landsat 9 sensor has not been ineffective in the high accuracy of the obtained values.