Highlighting the Importance of the Vegetation Variable on Distributed Land surface temperature on different land use/land cover in Javanrud city range

In environmental models, Vegetations are considered as an important part in controlling environmental changes. To determine the importance of vegetation on land surface temperature (LST), preliminary preprocessing was performed on Landsat 8 image and a split window procedure was used to determine surface temperature. Temperature difference with the surrounding synoptic stations was estimated to be about 2.5 °C, which confirmed the accuracy of the computational temperature. In addition to Normalized difference vegetation index (NDVI), other variables such as Normalized difference water index (NDWI), elevation, Aspect, Landuse/Landcover (LU/LC) and heat loading index were used to show the effect of vegetation on temperature. LU/LC map of city area was obtained by using a 7 categories training samples by Neural Network Classification (MLP). Total Kappa index was calculated 0.82. Then, the points were randomly extracted by stratified sampling method separated for 7 LU/LC categories and divided into two groups of testing and training. Values of the mentioned variables were extracted for these points and random forest regression was used to investigate their effect on surface temperature. Performance of the model was evaluated and validated using mean absolute error (MAE), mean square error (MSE) and root mean squared error (RMSE). Based on sensitivity analysis by Mean decrease accuracy, NDVI, Aspect and NDWI had the most effect on surface temperature distribution. Highest surface temperature with 29 °C was for uncoated rangelands and its lowest with 25 °C was for human-made areas, including Javanrud and peripheral villages. The results of this study showed that the NDVI plays a very important role in land surface temperature changes.