It is often necessary to track objects or events that gather as a large group, such as a school of fishes, or that do not have a rigid shape, such as forest fires and toxic gas. Their aggregate features, such as shape, area, moving direction, and speed are often queried. These objects or events often exhibit interesting behavior, such as splitting into multiple groups or merging into a single group. In this paper, we discuss the use of wireless sensor networks to track such events. A framework, CLOUD, is proposed, which constructs a collective structure over every region covered by the events to collect and aggregate related information. CLOUD is evaluated by simulations in NS2. The results show that CLOUD can effectively reduce bandwidth consumption in tracking such events while conserve energy.