3) At larger incidence angles (i.e. greater than 35 degrees) it appears that surface 
roughness, perhaps at a macro-scale of structure edges, vegetation morphology, etc., plays 
an increasingly dominant role in generating specular return from settlements and 
minimizes topographic effects. 
A) As might be expected, settlements were most readily detected in terrain 
conditions where they generated maximum contrast from the surroundings in terms of 
signal response. Specifically, settlements located in open meadow areas, forest clearings, 
and level terrain were identified with the highest detection rates and least error. 
5) Overall, the detectability, accuracy of detection, and rate of commission errors 
improved as the angle of incidence increased up to a point and then decreased with an 
increase in incidence angle. But, there were exceptions for specific environmental 
conditions. 
6) In examining look direction differences there appeared to be a marked change in 
settlement detectability. Two data takes were flown almost perpendicular to each other 
with slightly different incidence angles. Over diverse terrain, the pass with a northwest 
look direction and AO.9 degree incidence angle generated a higher accuracy (76.A per cent) 
than the pass with a northeast look direction and 52.3 degree incidence angle (62.7 per 
cent). This result raised the question as to whether the difference was due to look 
direction, incidence angle, or both. 
7) When only forested/mountainous areas were examined using the same passes the 
imagery with the largest incidence angle generated the highest accuracy--reversing the 
results found in examining the total areas covered by each pass. 
Thus, it is suggested that the optimum incidence angle varies between 
forest/mountain and open level terrain environs, even in such a relatively small area of the 
earth. If this is true, these data support and expand the observations made by MacDonald 
and Waite (1971) with regard to geological studies. Unfortunately, the terrain imaged by 
both passes in this study was not entirely identical, precluding an exact comparison. 
The preliminary effort reported here points to some interesting questions regarding 
settlement detectability in specific and the theoretical relationships among surface 
roughness, incidence angle, and terrain target in general.