A model for the onset of the reduction in SRF cavity quality factor, the so-called Q-drop, at high accelerating electric fields is presented. Breakdown of the surface barrier against magnetic flux penetration at the cavity equator is considered to be the critical event that determines the onset of Q-drop. The worst case of triangular grooves with low field of first flux penetration Hp, as analyzed previously by Buzdin and Daumens, [1998 Physica C 294: 257], was adapted. This approach incorporates both the geometry of the groove and local contamination via the Ginzburg-Landau parameter kappa, so the proposed model allows new comparisons of one effect in relation to the other. The model predicts equivalent reduction of Hp when either roughness or contamination were varied alone, so smooth but dirty surfaces limit cavity performance about as much as rough but clean surfaces do. When in combination, contamination exacerbates the negative effects of roughness and vice-versa. To test the model with actual data, coupons were prepared by buffered chemical polishing and electropolishing, and stylus profilometry was used to obtain distributions of angles. From these data, curves for surface resistance generated by simple flux flow as a function of magnetic field were generated by integrating over the distribution of angles for reasonable values of kappa. This showed that combined effects of roughness and contamination indeed reduce the Q-drop onset field by ~30%, and that that contamination contributes to Q-drop as much as roughness. The latter point may be overlooked by SRF cavity research, since access to the cavity interior by spectroscopy tools is very difficult, whereas optical images have become commonplace. The model was extended to fit cavity test data, which indicated that reduction of the superconducting gap by contaminants may also play a role in Q-drop.