In order to improve the accuracy of similarity analysis of precipitation images, a rainfall similarity search method based on daily precipitation images of watershed is proposed. The algorithm first extracts the daily precipitation, precipitation distribution, precipitation center characteristics of the precipitation images, and calculates the similarity distance of each characteristic respectively. Then, an ensemble weighting method of normalized discounted cumulative gain-improved particle swarm optimization is proposed to weight and fuse the three extracted features as the similarity measure of precipitation image. The similarity search experiments of daily precipitation images on the Jialing River Basin illustrate that the method proposed in this paper can better characterize the spatiotemporal characteristics of the precipitation image and quickly discover similar rainfall processes from precipitation images.