In cognitive radio, wireless ad hoc networks, the rapid changes in environment in addition to heterogeneous spectrum availability among the CR users resulted in the connectivity and stability problem of the network. A cluster structure is seen as an effective technique to overcome this problem. Presently, most existing works in cognitive radio focus on forming clusters under the homogenous channel model. Withal, the channels that are situated in vastly isolated frequency bands show a remarkable heterogeneity in transmission ranges. In this paper, we proposed a greedy heuristic algorithm for clustering to identify the effect of channel heterogeneity on cluster formation performance. A new metric for cluster head selection named selection factor is proposed. This metric is computed based on the number of available channels, the number of common channels that a CR user share with each of its neighbors, channel reward (coverage) for the available channels and the number of neighbors. To the best of our knowledge this is the first effort to add the channel heterogeneity into the formulation of cluster head election in cognitive wireless networks. We investigate the clustering formation, performance for the CRAHN under both heterogeneous and homogeneous channel models. Particularly, we evaluate the average number of clusters, cluster size, number of common available channels within each cluster and load balance among clusters. The algorithm has been implemented and tested using Matlab. Simulation results revealed that the performance of clustering is significantly affected by the channel heterogeneity.