Relative state estimation is an important component in multi-vehicle applications in GPS-denied environments. In this paper, we consider relative heading estimation between two Unmanned Aerial Systems (UAS). We examine observability properties of relative states between the two UAS in the presence of constant disturbances. We assume that UAS 2 measures UAS 1's position in its own frame. We first present observability conditions of the relative states when UAS 1 maintains a constant heading rate and a constant airspeed under the disturbance. We next establish state observability properties when UAS 1 is tracking a stationary ground target using a feedback control. In this case, the heading rate is non-constant. We design an extended Kalman filter (EKF) and conduct simulations to demonstrate its effectiveness in estimating the relative states.