Georeferencing is the process of tying a dataset that lacks spatial positioning — typically a scanned paper map, an aerial photograph, or a raw raster — to a known coordinate reference system so it overlays correctly with other geographic data.
How it works
The analyst places ground control points (GCPs): identifiable features in the image (road junctions, building corners, grid ticks) matched to their known real-world coordinates. A transformation is then fitted to warp the image. Common transformations include polynomial (1st-order/affine for simple shift-scale-rotate, higher orders for distortion) and thin-plate spline for localized warping. The fit quality is reported as RMS error in map units; a few control points well distributed across the image give a more reliable result than many clustered in one corner.
Why it matters
Most legacy geological and topographic knowledge lives on paper or in un-located scans. Georeferencing is the first step in digitizing those maps into usable GIS layers — without it, you cannot digitize contacts, faults, or contours into the correct location. It is also essential for drone imagery and for any raster delivered without spatial metadata.
Common pitfall
Confusing georeferencing with geocoding (turning an address or place name into coordinates) — they are unrelated. Another pitfall is over-fitting: using a high-order polynomial with few or poorly distributed GCPs can produce a low reported RMS while badly distorting areas between control points. Choose the simplest transformation that fits, and spread GCPs across the whole image including the edges.