Color Figures from Geographical Information Systems, Second Edition

Plate 24 In-vehicle use of GPS. ( 25 of 67)	Plate 25 Use of a hand-held GPS receiver. ( 26 of 67)	Plate 26 Interpolation of a DEM from scattered point data using methods available in GIS: (a) given data and Voronoi polygons; (b) TIN-based linear interpolation; (c) inverse distance weighting; (d) Kriging (spherical variogram); (e) spline with tension and stream enforcement; and (f) regularised spline with tension and smoothing. ( 27 of 67)	Plate 27 Influences of interpolation on the results of an erosion/deposition model: (a) observed depth of colluvial deposits in the sampling points and as interpolated raster shown in colour; (b) artificial erosion/deposition pattern predicted from a DEM interpolated from contours by spline with tension set too high; and (c) realistic erosion/deposition pattern predicted using a DEM interpolated from contours by regularised spline with tension and smoothing. ( 28 of 67)
Plate 28 Interpolation of a 50-m resolution DEM (7OO × 3OO) from a large dataset with complex topography using a regularised spline with tension. Inset shows the principle of segmented processing based on quadtrees and the detail of given contours. ( 29 of 67)	Plate 29 lnterpolation of annual precipitation in tropical South America using: (a) bivariate regularised spline with tension and smoothing; (b) trivariate interpolation with incorporation of the influence of topography. ( 30 of 67)	Plate 30 Snapshot from a result of spatio-temporal interpolation (3-D + time) of nitrogen concentrations in Chesapeake Bay (January 1991). For the complete animated result see http://www.cecer.army.mil/grass/viz/ches.html. ( 31 of 67)	Plate 31 Influence of tension on the surface (a–c) and volume (d–f) models of nitrogen concentrations in the middle section of Chesapeake Bay. Low tension (a, d) leads to overshoots; high tension (c, f) creates extrema in data points. Appropriate tension (b, e) was found by minimising the cross-validation error. ( 32 of 67)
Plate 32 Multi-criteria evaluation using varying degrees of trade-off. ( 33 of 67)	Plate 33 Decision risk: (bottom) the probability (risk) that an area would be flooded by a rise in sea level if one were to assume that it would not, based on errors in measured elevation; (top) the flood expected at a 5% decision risk level, draped upon a colour-composite LANDSAT-TM (Bands 3, 4, 5) image of a region in north-central Vietnam. ( 34 of 67)	Plate 34 Watershed drainage network and drainage area partition produced by a graph optimisation algorithm in the GRASS algorithm r.watershed. Note the bias towards straight-line drainage connections to minimise the cumulative cost function. ( 35 of 67)	Plate 35 Partitions of the South Platte River basin produced by two methods of pruning the full drainage direction tree: (a) simple thresholding of the drainage area accumulation image; (b) adaptively pruning the drainage tree using the surface spherical variance. Both images have approximately the same number of partition units. ( 36 of 67)