The paper got the data of spectral reflectance of ground features from field surveying by using field spectroradiometer. The spatial distribution information of the ground features was obtained from the land-use map. Based on above mentioned, the generating algorithm of spectral reflectance image of ground features was developed by Modeler module of ERDAS Imaging software. The four bands were selected as example image bands, including the blue band (0.45-0.52im), the green band (0.52-0.60im), the red band (0.63-0.69im) and the infrared band (0.76-0.90im). The four band images with real geographical coordinates were generated from the spectral reflectance of ground features. In order to present the following images, the true color and the standard false color images were merged with four individual band images. By using the field spectroradiometer, relatively simple compared with hyperspectral imaging radiometer, the similar spectral reflectance images of ground features could be obtained with the secondary developed generating algorithm on the ERDAS Imaging software platform. Through the analysis of the spectral reflectance images of ground features, we can prove that the generated images are close to the real land scenes. Therefore, this paper provides a new idea and a new method for the first step of simulating remote sensing images with real geographic coordinates. Finally, the authors prefer to explain that further studies should be developed in two aspects. One issue is how to describe the spatial distributing information of ground features more accurately, and the other is how to differentiate the same class ground features with different spectral reflectance. Based on above, further more studies should include the effect of topographic factors on the spectral reflectance of ground features.