Terrain Analysis

Digital Terrain Data

Terrain data most commonly take the raster format (i.e., DEMs), which records elevation on a cell-by-cell basis for each cell, but irregular sampling points, contour lines, and triangulated irregular networks (TIN) are also common elevation data formats. DEMs may be produced from one or multiple data sources, such as conventional topographic maps, sample points, and remotely sensed imagery, and the production process often requires substantial preprocessing and interpolation of source data. The user should pay special attention to understanding the effect of data lineage when dealing with various elevation data sets. The raster DEM format has the advantage of simplicity, and it matches the format of remotely sensed imagery, making DEMs easier to update and improve using imagery compared with other data formats. DEM resolutions vary from very fine (e.g., < 5 m) to very coarse (e.g., 1 km or coarser) based on data source and/or application purpose. However, each specific data set has a uniform cell size for the entire represented area, making it impossible to present more details for steep places than flat ones.

The spatial resolution of DEMs as indicated by the cell size is critical in terrain analysis because most terrain analysis conclusions depend on and may vary with the DEM resolution. For example, slope gradient calculated at a very fine spatial resolution (i.e., with small cells such as 1 m) may help identify a small hollow (or depression) of 3 m to 10 m in the middle of a hillslope; the same calculation at 30 m would not be able to identify the hollow but may better describe the overall steepness of the hillslope. Scientists have also found that surface and subsurface water flow across the entire terrain surface could be traced—thereby connected—on a cell-by-cell basis, but the outputs of this modeling activity are notably less accurate when DEM cells are larger than 10 m, and especially 30 m. The cell size also determines the size (and possibly cost) of DEM data sets and may influence the spatial extent of the analysis that is conceived or conducted.

The spatial extent as another component of spatial scale is important in terrain analysis for several reasons. First, the terrain analysis results for one area may not be applicable to another area. Second, the topography-based modeling of biophysical processes involving a particular point, such as being shadowed by remote hills or receiving runoff from a remote ridgeline, requires the complete consideration of all relevant areas that may impact this point. Third, the study area should be sufficiently large in comparison with the cell size, so that the analytical results for the edge cells, whose characterization is not supported by [Page 466]a complete neighboring area, would not severely bias the conclusion for the entire area.

...