this range. The analyst can then determine if questionable points are errone- 

 ous or represent genuine but unexpected topography. The X and Y points 

 should typically represent Cartesian coordinates, which is the case if the origi- 

 nal maps were based on State Plane coordinates. X and Y points that are 

 latitude and longitude must be converted by the program. 



Gridding operations 



Gridding is a mathematical process in which a continuous surface is 

 computed from a set of randomly distributed X, Y, and Z data 1 . The result 

 of the gridding operation is a data structure (usually a surface) called a grid. 

 Note that the grid is an artificial structure. It is based on the original data 

 (and hopefully is an accurate representation of the topography which was 

 surveyed in the field), but the grid points are not identical with the original 

 survey points (Figures 44 and 45). Because the grid represents the surface 

 that is being modeled, the accuracy of the grid directly affects the quality of 

 any output based on it or on comparisons with other grids generated from 

 other data sets. Computing a grid is necessary before operations such as 

 contouring, volume calculation, profile generation, or volume comparison can 

 be performed. The advantage of a grid is that it allows the program to 

 manipulate the surface at any scale or orientation. For example, profiles can 

 be generated across a channel even if the original survey lines were not run in 

 these locations. In addition, profiles from subsequent surveys can be directly 

 compared, even if the survey track lines were very different. 



Several steps must be considered as part of the grid generation. These 

 include: 



• Selecting a gridding algorithm. 



• Identifying the input data. 



• Specifying the limits of the grid coverage. 



• Specifying gridding parameters. 



• Specifying gridding constraints. 



• Computing the grid. 



The choice of a gridding algorithm can have a major effect on the ultimate 

 appearance of the grid. Software companies have proprietary algorithms 

 which they claim are universally superior. Often, however, the type or distri- 

 bution of data determines which procedure to use, and some trial and error is 

 necessary at the beginning of a project. Because a computed grid is an artifi- 

 cial structure, often it is a subjective evaluation whether one grid is "better" 

 than another. For subaerial topography, an oblique aerial photograph can be 

 compared with a computer-generated three-dimensional drawing oriented at 

 the same azimuth and angle. But for a subaqueous seafloor, other than 



Material in this section has been condensed from course notes provided by Radian Corpora- 

 tion during a CPS-3 training seminar presented at CERC in November 1989. 



Chapter 5 Analysis and Interpretation of Coastal Data 



111 



