Summary form only given. Surface flashover formation at dielectric/air interfaces during pulsed high power microwave (HPM) excitation can severely limit the power densities which can be transmitted into atmospheric medium. Previous studies on HPM surface flashover in the S-band at 5 MW power levels have reported on the contributing factors to flashover development including the effects of gas type, pressure and relative humidity. Furthermore, analysis on optical emission spectra collected from the developing discharge has determined that the vibrational and rotational temperatures of the plasma are approximately 2700 degK and 300 degK, respectively. In addition to experimental efforts, a Monte Carlo-type electron motion simulation code, MC, has been developed to calculate the increasing electron density during flashover formation. Results from this code have exhibited a quantitative agreement with experimental data over a wide range of atmospheric conditions. A critical parameter to flashover development is the stochastic process involving the appearance of initiatory or "seed" electrons, as seen by the reduction in flashover delay time by approximately 10-20% in the presence of external UV illumination. While the current version of the MC code seeds the flashover location with electron densities on the order of background ion densities produced by cosmic radiation, it fails to incorporate the field assisted collisional detachment processes which are the primary origin of these electrons on the time scales of interest. Investigation of these processes and development of more accurate seeding in the MC code is a key step towards predicting HPM flashover over a wide range of parameters, particularly in the presence of highly electronegative gasses such as SF6 or O2, in which there is an absence of free electrons. Theoretical results of HPM surface flashover with the improved seeding model will be benchmarked against previously measured data obtained with HPM pulse excitation. Further, the slow rise-time data (~500-600 ns risetime) that revealed a distinct reduced field vs. pressure delay time product dependence will be supplemented by short rise-time pulse data.