Giles wrote many research proposals. He did this to gain funds for graduate students who would participate in expanding the slope dimension of the statewide data system (then in declining operation ,1980). A draft proposal contained:

1. Are we using the best available algorithm for slope analysis (then Topo-II by Grender of the Va Tech Geology Dept.) to give as accurate slope information as is available from the sampled elevations?
2. Can the computational costs be reduced?
3. Are there alternatives (data structure and storage) (Wong 1980- a MS thesis of the Computer Sci. Dept.)?
4. What is the effect of changing cell size on the algorithms? On accuracy?(Mathur in a MS thesis in 1990 of the Geography Dept at Va Tech explored some of these relations)
5. Can resulting data and its form be useful in various state information systems or does the system and use constrain the algorithm?
6. Are alternative approaches feasible and useful in gaining cost effectiveness such as:
   • a person computes interactions such as using an additional file of observed topographic peaks (perhaps as a constant set)
   • using a Fibonnaci search to bracket the computational interval (e.g., in coastal areas where the slope changes very gradually)
   • using planar equations and a quad-tree searching algorithm for storage and retrieving efficiency
   • using steepest ascent optimization routines to select starting points, computational regions, or as a means to cut down on an "all possible combinations" approach to the problem?
   • given that the slopes of a diverse three-dimensional terrain surface can be represented as a polynomial, what are the storage and retrieval efficiencies in the current and contemplated uses and computer environment?
   • How do the answers compare to mapped reality? To ground truth? (now possible with GPS)
   • What is the fewest known points (e.g., in peaks and valleys) that can be used to adequately represent a topography for reasonable estimation of slopes?
   • What are the "reasonable classes "for different users? (The questions of accuracy, precision, confidence, and power)?
   • Can a rapid procedure be created for plotting a cumulative percent of land in an ownership or county (vertical axis) with slope intervals e.g., in degrees (horizontal). A vertical line (or series of labeled lines ) might remark classes too steep for certain operations such as machine logging.

     RHG - October, 2000