We propose a Spatial-based Multi-resolution Data Dissemination(SMRDD) scheme for data retrieval at multiple levels of detail in wireless sensor networks. In addition to saving energy by avoiding query flooding and extra data transmission for data storage elsewhere, SMRDD achieves further energy efficiency by reducing dissemination of redundant data. In SMRDD, instead of being addressed to nodes, queries are sent to pixel points, a set of carefully selected locations across the sensor field, on the support of a geographic routing protocol. The query finally reaches the home node of the pixel point, the node geographically closest to the pixel point, which reports data to sinks in response. The density of the pixel points corresponds to data resolution level. The higher the density, the higher the data resolution level. The highest density of pixel points is determined by the sensing range of sensors for the application, which allows sensing ranges of home nodes of all pixel points to cover the whole sensor field. Analysis and simulation results show that the performance of our method is much better than early flooding based methods like Directed Diffusion.