Natural disasters, such as earthquake, typhoon and flood, have caused great loss of lives and property each year, which makes emergency monitoring and rescue an imperative problem to be addressed. In this paper, we designed a novel monitoring and rescue system based on wireless sensor network (WSN) for disaster scenarios, which combines environment monitoring, information transmission, and emergency localization. In our system, fast localization through WSN is a crucial technique for searching and rescuing. We proposed an improved weighted centroid algorithm with adaptive error correction to increase localization accuracy of sensor networks with sparse anchors. This method based on received signal strength indication (RSSI) is low cost, expandable, easy to implement, and suitable for disaster rescues.