Skip to content

Georeferencing in QGIS

Georeferencing is the process of assigning real-world coordinates to a raster dataset (such as a scanned paper map, a historical aerial photo, or an unreferenced topographic sheet). By defining geographic coordinates for specific pixels, GIS analysts can integrate historical data and paper maps directly into modern spatial analysis pipelines.

1. Core Georeferencing Concepts

To georeference a raster, you must align it with a known Coordinate Reference System (CRS) using reference control points.

  • Ground Control Points (GCPs): Coordinates of identifiable features on both the unreferenced raster and a reference map (e.g., road intersections, river confluences, mountain peaks, or coordinate grid intersections).

  • Transformation Algorithms: Mathematical models used to stretch, rotate, and scale the raster grid to match the target coordinates:

  • Linear: Performs simple scaling, rotation, and translation. Best for maps that are already orthorectified and only lack coordinate headers.
  • Polynomial (1, 2, 3): Applies complex polynomial equations to warp the image. Polynomial 1 (Affine) preserves parallel lines. Polynomial 2 and 3 stretch the image non-linearly to correct for paper folds, camera distortion, or terrain displacement.

  • Thin Plate Spline (TPS): Localized transformation that warps the image locally around GCPs. Highly effective for severely distorted historical maps.

  • Resampling Methods: Algorithms used to calculate the new cell values of the output raster grid:

  • Nearest Neighbor: Assigns the value of the closest input pixel. Fastest method; preserves original discrete values (ideal for categorized maps).
  • Bilinear Interpolation: Computes a weighted average of the nearest 4 pixels. Smooths the output (ideal for continuous elevation datasets or satellite images).
  • Cubic Convolution: Computes a weighted average of the nearest 16 pixels. Sharpens edges but is computationally intensive.

2. Step-by-Step Georeferencing Workflow in QGIS

This workflow demonstrates how to georeference a scanned topographic basin map using a reference base map.

Step 1: Open the Georeferencer Tool

  • In QGIS Desktop, navigate to the main menu and select Layer > Georeferencer.
  • In the Georeferencer window, click the Open Raster button (blue checkerboard icon) and select your scanned map file (e.g., scanned_catchment_map.jpg).

Step 2: Set the Target CRS

  • Click the Transformation Settings button (yellow gear icon).
  • Target CRS: Select the projected coordinate system for your project area, such as WGS 84 / UTM Zone 44N (EPSG:32644) or WGS 84 / UTM Zone 45N (EPSG:32645).

Step 3: Add Ground Control Points (GCPs)

  • Identify a clear, distinguishable landmark on the scanned map (e.g., a coordinate grid intersection or river junction).
  • Select the Add Point tool (green dot with crosshair) and click precisely on that landmark.
  • A dialog box will appear asking for coordinates. You can:
  • Manually enter the X (Easting) and Y (Northing) coordinates printed on the scanned map's graticules grid.
  • Click From Map Canvas and click on the same feature on your active QGIS base map layer to pull coordinates dynamically.
  • Repeat this step to add at least 4 to 6 GCPs distributed evenly across the raster corners and center. Avoid clustering GCPs in a single region, as this causes distortion at the edges.

Step 4: Configure Transformation Parameters

  • Re-open Transformation Settings and configure:
  • Transformation type: Select Polynomial 1 (or Thin Plate Spline if the paper map is folded).
  • Resampling method: Select Bilinear.
  • Output raster: Name your file georeferenced_catchment_map.tif (always save as GeoTIFF).
  • Use 0 for transparency when needed: Check this to hide black borders around the warped output raster.
  • Load in QGIS when done: Check this to view the output map directly on the canvas.
  • Click OK.

Step 5: Start the Transformation

  • Click the Start Georeferencing button (green play icon) on the toolbar.
  • The Georeferencer will warp the raster image and load georeferenced_catchment_map.tif onto your QGIS map canvas.

3. Residual Error Analysis

The Georeferencer table displays a Residual error (measured in pixels) for each GCP:

\[\text{Residual} = \sqrt{(X_{\text{actual}} - X_{\text{calculated}})^2 + (Y_{\text{actual}} - Y_{\text{calculated}})^2}\]
  • Mean Error (RMSE): The average displacement error across all GCPs.
  • Interpretation: A high residual error on a specific point indicates that the point was placed inaccurately or that the coordinate was mistyped. Use the Move GCP Point or Delete Point tool to adjust or recreate points with high residuals until the overall RMSE is minimized (ideally under \(1.5\) to \(2.0\) pixels).

4. Significance in Hydrology

Georeferencing plays an essential role in compiling historical and baseline data: * Historical Runoff Analysis: Digitizing drainage networks and sub-basin boundaries from historical topographic maps to analyze decadal changes in stream alignment and channel shifting. * Siting Hydromet Stations: Finding the exact coordinates of legacy gauge stations that were active before GPS coordinates were recorded, matching them with local coordinates written in historical commission reports. * Land-Use Trend Modeling: Aligning old aerial photography layers to map forest cover dynamics and calculate runoff coefficient shifts over time.