Distributed compressive video sensing (DCVS) for wireless visual sensor networks faces challenges due to computational limitations and bandwidth constraints. This paper presents a rate-adaptive DCVS scheme that dynamically allocates measurements based on temporal correlation and sparsity estimation. By skipping highly correlated blocks and adaptively sampling others, the proposed method achieves improved rate-distortion performance with reduced sampling complexity and transmission burden. Experimental results demonstrate substantial gains over state-of-the-art methods, especially for videos with varying motion speeds. Codes and data are available at https://github.com/SongHere/USE_DCVS.