The invention discloses a ground surface temperature downscaling algorithm based on a temperature change rate. A ground surface temperature image with high temporal and spatial resolution is obtained through static downscaling and dynamic downscaling. According to the method, a change law of ground surface temperature in space is explored in term of the temperature itself; at the same time, a temperature annual-cycle model is combined, the ground surface temperature of a long time sequence is combined into a downscaling model, and the ground surface temperature at any time is dynamically acquired; a used downscaling factor is a ground surface temperature gradient, change of the downscaling factor in a year is stable, and compared with downscaling factors of a NDVI, an albedo and the like, the downscaling factor is easier to be acquired; a factor with similar weather conditions can be used for replacement when temporal-consistent scale factors are deficient in the case of data absence, and impacts on a result are lesser; and an effect of the downscaling method is stable.