A surveillance area is to be monitored using a grid network of heterogeneous sensor nodes. There are two types of nodes; type 0 nodes which perform sensing and relaying of data within a cluster, and type 1 nodes which act as cluster heads or fusion points. A surveillance aircraft visits the area periodically, and gathers information about the activity in the area. During each data gathering cycle, the sensor nodes use multihopping to communicate with their respective cluster heads, while the cluster heads perform data fusion, and transmit the aggregated data directly to the aircraft. We formulate and solve a cost based optimization problem to determine the optimum number of sensor nodes (n0), cluster head nodes (n1) and the battery energy in each type of nodes (E0 and E1 respectively) to ensure at least T data gathering cycles. We observe that the number of cluster heads required, n1, scales approximately as n0 1- k 4 where k is the propagation loss exponent.