A source of accurate and reliable heading is vital for the navigation of autonomous systems such as micro-air vehicles (MAVs). It is desirous that a passive computationally efficient measure of heading is available even when magnetic heading is not. To confront this scenario, a biologically inspired methodology to determine heading based on atmospheric scattering patterns is proposed. A simplified model of the atmosphere is presented, and a hardware analog to the insect Dorsal Rim Area (DRA) photodetection is introduced. Several algorithms are developed to map the patterns of polarized and unpolarized celestial light to heading relative to the sun. Temporal information is used to determine current solar position, and then merged with solar relative heading resulting in absolute heading. Simulation and outdoor experimentation are used to validate the proposed heading determination methodology. Celestial heading measurements are shown to provide closed loop heading control of a ground based robot.