Estimation of daily average near-surface air temperature using MODIS and AIRS data

Abstract

The near-surface air temperature is an essential climatic variable wildly used in studies of meteorology, climate, and environmental health. Numerous studies have developed approaches to estimate near-surface air temperature from remote sensing data for clear sky conditions, but efforts to estimate air temperature for cloudy sky conditions and daily average air temperature using remote sensing data still few. The current study introduces an approach to estimate daily average near-surface air temperature using the estimated daily maximum and minimum air temperatures with the help of time series of the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Atmospheric Infrared Sounder (AIRS) observations. The daily maximum temperatures of clear sky pixels are estimated from three MODIS products data using a linear regression model as expressed in [1], and the AIRS standard surface air temperature products data are used to fill the cloudy sky pixels for the images after a downscaling process. The retrieved land surface temperatures (LST) from Aqua MODIS night-overpass observations are used as the daily minimum air temperatures for the clear sky pixels, and the cloudy sky pixels are also filled by AIRS standard surface air temperature products data. Thus, the daily average near-surface air temperature can be estimated according to the diurnal variation of near-surface air temperature. This method was validated using field observed air temperature data of 176 ground meteorological stations in August 2013. The mean absolute error (MAE) and the root mean square error (RMSE) are 1.2 °C and 1.6 °C. The strength of the proposed methodology is that it can obtain reasonable near-surface air temperature data from remote sensing data, so it is useful in regions with sparse ground stations.