In this paper, we characterize the joint network capacity region (JNCR) for a licensed broadcast (primary) and ad hoc cognitive (secondary) network in a heterogeneous environment, including indoor and outdoor transmissions, under various spectrum (white space) detection techniques. Each technique delivers a different degree of RF-environment awareness - the more a device knows about its environment the larger the network capacity region. To quantify the gains, we develop a simple stochastic model capturing the interdependency amongst primary and secondary nodes and compare their joint capacity. Cognitive devices using the classical signal energy detection method are shown to perform poorly due to limitations on detecting primary transmitters in environments with indoor shadowing. This can be circumvented through direct use (e.g., database access) of location information on primary transmitters, or better yet, on that of primary receivers. The specific capacity trade-off between primary and secondary networks depends on white space detection techniques, resulting in JNCRs which range from complement convex to linear to (almost) convex. Our results show that, for example, the gain of positioning-assisted method over signal energy detection is 76% and the gain of receiver location-aware approach is 177% when the density of primary transmitters is 2 × 10-10 m-2, the indoor shadowing level is -10dB, and the fraction of indoor nodes is 0.5. Furthermore we show that if cognitive devices have positioning information then the secondary network's capacity increases monotonically with increased indoor shadowing in the environment. These are the first analytical results quantifying, albeit for simple heterogeneous environmental model, the capacity gains one can expect when cognitive devices leverage additional information.