The invention discloses an all-day sampling optimization layout method for the surface temperature of a heterogeneous underlying surface. The method comprises the following steps: obtaining and preprocessing related products; fitting a daily cycle change curve of the surface temperature of the stationary meteorological satellite pixel by pixel; reconstructing a high spatial resolution polar orbital satellite hourly surface temperature product; constructing a surface temperature sampling cost function; setting the number of sampling samples; determining an optimal sample layout position based on a simulated annealing algorithm; carrying out iteration stop condition judgment; and determining the number and positions of optimal sampling samples in the research area. According to the method, the optimal number of sampling points can be determined, the positions of the sampling points can also be determined, the space-time representativeness of the sampling points is effectively guaranteedall day long, then the surface temperature truth value of the pixel scale can be accurately represented, and the application requirements of surface temperature products for carrying out truth inspection under various underlying surface conditions are met.