GISI Foundation cheat sheet

Vector data represents discrete features as points, lines, and polygons; raster data represents continuous fields as a grid of cells, each holding one attribute value.

Topology records spatial relationships (adjacency, connectivity, containment) independently of coordinates, enabling network tracing and polygon neighbor queries without recomputation.

Raster resolution is the ground size of one cell; halving cell size quadruples cell count and storage, so resolution choice trades precision against volume.

Vector storage scales with vertex count, raster storage with cell count; a coarse raster of a large extent can exceed a detailed vector layer in size.

Discrete objects suit vector models (parcels, pipes); continuous phenomena (elevation, temperature, reflectance) suit raster models.

A TIN stores terrain as irregular triangles with vertices at measured points, preserving breaklines better than a fixed grid of the same point budget.

Scale is the ratio of map distance to ground distance; large-scale maps (1:1,000) show small areas in high detail, small-scale maps (1:1,000,000) the reverse.

Feature geometry types are fixed per layer in most formats; mixing points and polygons in one layer complicates symbology, queries, and overlay operations.

A geographic coordinate system (GCS) locates positions on a datum-defined ellipsoid in degrees; a projected coordinate system (PCS) applies a map projection to express positions in linear units on a plane.

A datum pairs an ellipsoid with an origin and orientation; WGS84 and NAD83 differ by roughly 1-2 m in North America, enough to matter at large scales.

Transformations between datums require published parameters and introduce error; reprojection between projections on the same datum is exact mathematics.

Conformal projections preserve local angles and shapes (Mercator, Transverse Mercator, Lambert Conformal Conic); equal-area projections preserve area (Albers, Lambert Azimuthal Equal-Area).

No projection preserves shape, area, distance, and direction simultaneously; choose by the measurement the analysis depends on.

UTM divides the globe into 60 zones of 6 degrees longitude, each a Transverse Mercator with a central meridian scale factor of 0.9996.

Web Mercator (EPSG:3857) uses spherical formulas on WGS84 coordinates; it is convenient for tiling but unsuitable for area or distance measurement.

Distance in a GCS is angular (degrees); computing Euclidean length on unprojected coordinates yields meaningless units, so project before measuring.

A spatial database stores geometry as a native type with a spatial index (R-tree or grid) so bounding-box and topological predicates run without full table scans.

The OGC Simple Features model defines geometry types (Point, LineString, Polygon, MultiPolygon) and predicates such as ST_Intersects, ST_Contains, and ST_Within.

ST_Within(A, B) and ST_Contains(B, A) are equivalent; ST_Intersects is symmetric; ST_Touches excludes overlap of interiors.

Normalization removes redundancy in attribute tables; geometry columns stay denormalized as single objects, so spatial and attribute design are tuned separately.

A spatial index accelerates filter steps but not exact geometry tests; queries run index filter first, then precise predicate on candidates.

File geodatabases and PostGIS both enforce geometry type and spatial reference per layer; shapefiles store one geometry type, lack true curves, and cap field names at 10 characters.

GeoJSON stores coordinates as longitude-latitude arrays in WGS84 and supports only 2D geometry; topology is not encoded, so shared boundaries are duplicated.

A join table keyed on a stable identifier links attributes to geometry; keying on geometry itself breaks when coordinates are edited or reprojected.

Overlay operations: clip cuts input to a boundary; intersect keeps overlapping areas of both layers; union keeps all areas; erase removes the second layer's area from the first.

Buffer distance is measured in the layer's coordinate units; buffering in a GCS produces distorted rings, so project first or use a geodesic buffer.

Spatial join transfers attributes by a spatial relationship (intersects, contains, nearest) and can produce one-to-many output rows unless a match option limits it.

Dissolve merges features sharing attribute values and can compute aggregate statistics; it is the standard step before area or length summarization.

Slope and aspect derive from the gradient of a DEM; slope is the magnitude of the gradient, aspect its compass direction, and flat cells have undefined aspect.

Viewshed analysis marks cells visible from an observer point given elevation and earth curvature; output is binary visibility, not distance.

Kriging, inverse distance weighting, and spline interpolate surfaces from sample points; IDW is exact at samples and produces bull's-eye artifacts, kriging models spatial autocorrelation.

Raster map algebra applies cell-by-cell expressions across aligned grids; misaligned extents or cell sizes require resampling, which alters values.

Visual hierarchy orders elements by contrast and size so the title and main data read first, supporting layers second, marginalia last.

Color schemes follow data type: sequential for ordered data, diverging for data around a meaningful midpoint, qualitative for unordered categories.

Choropleth maps encode rates or ratios, never raw counts, because area differences otherwise dominate the pattern.

Classification methods (equal interval, quantile, natural breaks, standard deviation) change the same data's appearance; state the method in the legend.

Map scale, north arrow, legend, and data source are required map elements; a representative fraction states scale independent of unit.

Label placement rules: avoid overlap with features and other labels, follow line curvature, keep text upright, and prioritize by feature importance.

Figure-ground contrast separates the data layer from the basemap; muting basemap saturation and value pushes it behind thematic content.

Layout resolution for print is typically 300 dpi; screen products use 72-96 dpi and RGB, print uses CMYK, and conversion shifts saturated colors.

Spatial data quality elements include positional accuracy, attribute accuracy, logical consistency, completeness, and temporal accuracy.

Root mean square error summarizes positional error against checkpoints; it does not bound maximum error, so report sample size and method too.

Metadata records lineage, accuracy, projection, contact, and use constraints; ISO 19115 is the international geospatial metadata standard.

FGDC Content Standard for Digital Geospatial Metadata is the US predecessor to ISO 19115; the ISO standard is now generally preferred for new work.

A data dictionary defines field names, types, domains, and null rules; it prevents semantic drift when layers pass between teams.

Governance assigns data ownership, stewardship, access rules, and review cadence; without an owner, updates and corrections stall.

Lineage documents source data, processing steps, and parameters so results can be reproduced and errors traced to their origin.

Version control for spatial data should record geometry and attribute changes with timestamps; overwriting files in place destroys audit trails.

A GIS project lifecycle runs needs assessment, data acquisition, database design, analysis, product delivery, and maintenance; skipping maintenance shortens data lifespan.

Requirements define the decision the map supports; deliverable format, accuracy, and scale follow from that decision, not from available data.

Positional accuracy requirements should match the smallest feature or decision threshold; over-specifying accuracy raises cost without benefit.

Ethical practice requires disclosing data limitations, protecting locational privacy of individuals, and not presenting derived estimates as measurements.

Licensing governs redistribution: some datasets permit derived products, others restrict sharing; check terms before publishing maps built on them.

Reproducibility requires recording software versions, tool parameters, and input dataset versions alongside outputs.

Documented quality assurance includes independent checkpoint validation, topology checks, and review of symbology against the data classification.

Stakeholder review at each milestone reduces rework; late discovery of a wrong extent or classification forces full redo of maps and analysis.