Remote sensing technology has been applied more widely and deeply with its development and, meanwhile, it has been asked for obtaining higher and higher spatial and temporal resolution. However, it is very difficult to overcome the contradiction between spatial resolution and temporal resolution of remote sensing images. Considering the influence of cloud, frog, snow and shadow, obtaining clear image with high spatial and high temporal resolution is impractical. To solve this problem, the authors proposed a method based on harmonic model for generating synthetic GF-1 images, which can take advantage of all available clear history images of GF-1 satellites to simulate the surface reflectance data at any specified date, so that obtaining time serial satellite images at any temporal frequency theoretically and overcoming the limits of methods based on fusion models for synthesizing satellite images become possible. Synthetic GF-1 images generated based on the harmonic model which will firstly establish a model parameterized by Julia date for every pixel of every band using all clear GF-1 time serial images since GF-1 satellite was launched and then the synthetic image at the specified day with the models would be generated. Finally, to illustrate the availability of harmonic model based method, the authors applied visual assessment and quantitative assessment. The synthetic images generated by this method were very visually similar to the real images and provide good result in quantitative assessment. Most difference of pixel values between synthetic image and real image ranged -0.03~0.03, and the root mean square error (RMSE) between synthetic image and real image ranged 0.02~0.05. The method based on harmonic model showed relatively high accuracy and stability, and effectively improved the temporal resolution of GF-1 images and could be applied in real production environment.