Estimating maximum daily temperature using NOAA satellite images- Case study in Oroomieh lake basin

Air temperature, as measured at standard meteorological stations, is an important climatologically variable and since these stations are collected as point samples, their data are not applicable for regional purposes. Therefore these data are necessary to be converted or modified to a regional data. Land surface temperature and vegetation index have the most correlation with air temperature and can be estimated by satellite images. Thus, for retrieving air temperature at regional scale regression models have been used where the input variables are land surface temperature and vegetation index. In this study three regression models; linear ordinary, linear multiple and non-linear multiple in estimating maximum air temperature are used and compared. In multiple regression models, in addition to variables mentioned above, also geographic variables used as model inputs. Land surface temperature and normalized deference vegetation index (NDVI) has been estimated using NOAA satellite images. The results show that non-linear multiple regression model has the highest precision among the models where the root mean square error (RMSE) and mean absolute error (MAE) estimated are equal to 1.57°C and 1.3°C, respectively.