A new physically based method for Air temperature downscaling

An international widespread concern about scaling is how to choose appropriate scale or resolution, and how to evaluate the impact of them[1]. Air temperature is an important input variable to estimate terrestrial evapotranspiration based on satellite remote sensing. The air temperature obtained by the observations from surface meteorological stations is limited in their spatial and temporal representation, while the validated GDAS (Global Data Assimilation System) has many advantages, it can provide the simulated temperature data every 3 hours, and it has great value in downscaling analysis. There are three major driving factors of the near surface air temperature: the surface long-wave radiative balance, land-air turbulent heat exchange, and advection. The fluctuation of the Air temperature (2m height level above ground) mainly depends on underlying surface feedback. Northern China was chosen as the study area. Using air temperature data from the GDAS forcing dataset as a data source, we proposed a new method for downscaling air temperature based on land surface temperature. In order to evaluate the performance of our methods, bilinear interpolation, spline interpolation were used in the comparison. To assess the performance of the downscaling approaches, the ground measurements were used to compare with the downscaling results. Experiments show that the effect of static feedback interpolation is the best based on the surface temperature. What we have got are as follows. First, In most plain areas, the air temperature(2m height level above ground) mainly depends on the temperature of the surface temperature. Second, during the process of downscaling, pure mathematic methods appear to be not sufficient. It is necessary that the effects of physical basis be taken into consideration.