Geographic Information Systems and Digital Data
This domain tests data models, topology, coordinate reference handling, accuracy reporting, and metadata for spatial data. Expect questions on vector and raster structure, datum transformation pitfalls, surface models, and how positional accuracy is quantified.
Data models and topology
- Vector data uses points, lines, and polygons with topology: connectivity, contiguity, adjacency, and coincidence. Polygons are built from arcs; dangling nodes and pseudonodes signal editing problems.
- Raster data uses cells whose size sets the resolution; categorical rasters represent classes, continuous rasters represent elevations or imagery. A raster cannot store detail finer than its cell size regardless of display scale.
- Accuracy has several dimensions: positional, attribute, logical consistency, completeness, and lineage. Metadata standards such as the FGDC Content Standard and ISO 19115 record all of these.
Coordinate reference and datum handling
- Project-on-the-fly reprojects display only; a true datum transformation (for example NAD27 to NAD83 through NADCON or HTDP methods) is required before coordinates are compared or area is computed.
- Two coordinate systems with the same name can reference different datums or epochs; never assume a match from the name alone.
- Surface models: a DEM represents bare earth with breaklines; a DSM includes buildings and vegetation; a DTM is a broader digital terrain model. LiDAR point clouds carry class codes, intensity, and multiple returns.
Worked example
Three checkpoints give coordinate errors of +1.2 m, -0.8 m, and +0.5 m. Square them: 1.44, 0.64, 0.25. Sum = 2.33; divide by 3 = 0.7767; square root gives RMSE = 0.881 m. Apply the 95 percent horizontal factor: 0.881 times 1.7308 = 1.53 m. The defensible statement is 'horizontal accuracy 1.5 m at 95 percent confidence', not 'accuracy 0.88 m', because the RMSE is a one-sigma-style statistic, not a confidence bound.
Traps that cost marks
Reporting RMSE as the accuracy; overlaying layers in different datums without transformation; assuming topology cleaning corrects attribute errors; treating a DSM as bare-earth elevation; setting a raster cell size finer than the source data can support; omitting lineage from metadata; assuming a shapefile stores topology or true curves.