Modeling the Relationship between Land Surface Temperature, Topography and Vegetation Cover Using Landsat 8 Satellite Imagery

Introduction
Land surface temperature plays a vital role in a wide range of scientific researches including climatology, meteorology, hydrology, ecology, geology, medical sciences, design and optimization of transportation services, fire location and especially in calculating the real evaporation and transpiration. There are some determining factors affecting the land surface temperature, such as, the kind of surface elements, topography conditions, environmental conditions, climate condition and the amount of emitted energy from the sun. Recognition and analysis of the relation between the land surface temperature and various factors are so critical. The remote sensing method has a widespread application in preparing the land surface temperature images due to the extensive covering and continuous data. The purpose of this study was to investigate the effects of vegetation cover indices and topographic factors on land surface temperature and modeling the relationship between land surface temperature, topographic conditions and vegetation cover using Landsat 8 satellite imagery.
Materials and Methods
In this study, sensor reflective bands OLI, Jimenez and Sobrino method were utilized to calculate the emissivity of the available phenomena in the area. By using TIRS Landsat 8 sensor thermal bands 10 and 11 and utilizing Split-window algorithm, the land surface temperature was calculated. Topography parameters, such as elevation, slope, aspect and vegetation were extracted using digital elevation model and NDVI index, respectively. Then, the relation between the land surface temperature and topography factors in diverse conditions was investigated by statistical analysis, and then, the validity of relations was analyzed with a confidence level of 95%. For this purpose, we employed ENVI 5.3, ArcGIS 10.4, ERDAS IMAGING 2014 software as well as SPSS statistical software.
Results and Discussion
The obtained results indicates that the study area has a uniform vegetation cover in most areas and a high percentage of the areas have the NDVI of 0.45-0.6. Nonetheless, due to the diversity in topographic and climatic conditions the area surface temperature is inhomogeneous and non-uniform. Consequently, there is no relation with a high correlation coefficient between the land surface temperature and vegetation in the area. However, there is a reverse linear relationship between the land surface temperature and vegetation in the area. This relation gains a higher correlation coefficient in the form of linear relation compared with second order polynomial, Pearson, and logarithmic equations. The areas with southern and southeast slope have higher land surface average temperature compared to other aspects during imaging due to their position exposed to direct sun radiation. The temperature average is different in various slopes. Investigating the relations of temperature and elevation independent of slope parameters and slope aspect, gives rise to an increase in the correlation coefficient between the two parameters. The relation of the land surface temperature and elevation, regardless of slope and aspect for the studied area, is a reverse linear relation with the correlation coefficient of 0.54, whereas for the relation between the land surface temperature and elevation in the western slope and slope of 40-50 degree, there is a reverse linear relation with the correlation coefficient of 0.76. Moreover, in investigating the relation between the land surface temperatures with topographic conditions, simultaneous consideration of both elevation and slope variables as independent variables for modeling the dependent variable of surface temperature reveals a strong relation. The addition of vegetation index parameter to relation independent parameters brings about a rise in relation’s correlation coefficient. For instance, relation’s correlation coefficient of the land surface temperature with elevation independent variables, slope, and vegetation in the western, northwest and southeast direction, are 0.84, 0.81 and 0.8, respectively. All the obtained relations are investigated in the confidence level of 95%, and validity of relations was confirmed by “t” statistic and search for relations’ coefficients.
Conclusion
Results of this research have indicated that taking elevation, slope, aspect and NDVI parameters independently for modeling the land surface temperature, can give adverse results and by simultaneous application of both topographic parameters and vegetation and also their combination, as dependent parameters, the land surface temperature can be precisely calculated. In addition, for accurate modeling of the land surface temperature all topographic, climatic and environmental conditions for the area should be taken into account. The thermal and reflective remote sensing technology are economical, fast and effective due to several positive aspects to provide uniform topographic, vegetation data and environmental parameters. So, further researches and investigation are necessary.