Raster geography turns the landscape into a field of numbers—and then calculates with the whole field.
A temperature map can be subtracted from another temperature map. A slope raster can be reclassified into suitable and unsuitable terrain. Distance, land cover and flood risk can be combined cell by cell to create a suitability surface.
Raster algebra is the application of mathematical, logical and conditional operations to raster cells across one or more geographic layers.
Raster algebra is spreadsheet logic extended across space.
Quick Read: The Mechanism
ALIGNED RASTER CELLS → MATHEMATICAL / LOGICAL RULE → CELL-BY-CELL OUTPUT → NEW GEOGRAPHIC SURFACE
1. Each Cell Carries a Value
A raster might store elevation, temperature, land-cover class, travel cost or probability. Raster algebra works by applying operations to corresponding cells.
2. Simple Arithmetic Can Reveal Change
Subtract an earlier elevation model from a later one to estimate surface change, or subtract baseline temperature from a heatwave map to show anomaly.
3. Reclassification Turns Continuous Values Into Rules
Slope below 5 degrees might be scored as highly suitable, 5–15 degrees as moderate and steeper terrain as unsuitable. Reclassification makes decision criteria explicit.
4. Conditional Logic Builds Geographic Decisions
Cells can be selected where several conditions are true: outside flood zones, near roads and on gentle terrain. The output becomes a geographic decision surface.
5. Alignment Is Essential
If raster grids have different cell sizes, origins or projections, cells do not correspond cleanly. Resampling may be required, and resampling can change values.
6. Resolution Controls the Meaning of the Calculation
Spatial Resolution determines how much detail each raster cell contains. A calculation cannot recover sub-cell variation that the inputs never recorded.
7. Raster Algebra Is Not Spatial Overlay
Spatial Overlay owns combining layers conceptually and geometrically. Raster algebra owns the cell-level mathematical operations used when those layers are represented as grids.
8. NoData Is Not Zero
A missing value can mean unknown, outside the study area or not applicable. Treating NoData as zero can create false results and sharp artificial boundaries.
9. Standardisation Matters Before Combining Variables
Temperature in degrees, distance in metres and slope in percent cannot be added meaningfully without transformation. Variables may need to be standardised or reclassified onto comparable scales.
10. Weights Express Priorities
If flood risk receives twice the weight of road proximity, that priority is built into the output. The final map is partly data and partly decision design.
11. Primary Geography: Add the Grids
Give pupils two small number grids and ask them to add matching squares. They immediately see the basic cell-by-cell logic.
12. Secondary Geography: Build a Simple Suitability Model
Students can combine slope and distance to road after first converting both to comparable suitability scores.
13. Advanced Geography: Map Algebra as a Model Language
Complex workflows can chain focal statistics, local operations, conditional rules and neighbourhood functions. The mathematics becomes a transparent model of how evidence is transformed.
14. Singapore Example: Urban Suitability
A planning screen could combine slope, flood exposure, accessibility and existing land use. Raster algebra can make those criteria computationally consistent across the island, provided the inputs share compatible scale and resolution.
15. Environmental Example
Habitat suitability models often combine vegetation, distance to water and human disturbance. The result is useful only if the scoring system reflects plausible ecological relationships.
16. Hazard Example
Flood or landslide susceptibility can combine terrain, rainfall and land-cover rasters. The output is a modelled susceptibility surface, not a direct observation of future failure.
17. Hostile Test: “The Formula Is Objective”
The arithmetic may be exact while thresholds, weights, input quality and resampling choices remain judgement calls. Reproducibility is not the same as neutrality.
18. Where Raster-Algebra Reasoning Breaks
- Alignment blindness: calculating across mismatched grids.
- NoData-zero collapse: treating missing as zero.
- Unit confusion: combining unlike variables without standardisation.
- Weight innocence: hiding judgement inside coefficients.
- Resolution overreach: interpreting results more finely than the inputs allow.
- Formula-equals-truth: mistaking deterministic calculation for empirical certainty.
19. Ten Questions for Raster Algebra
- What does each raster cell represent?
- Are grids aligned?
- Are projections compatible?
- What happens to NoData?
- Are variables on comparable scales?
- What thresholds are used?
- What weights are used?
- How does resolution affect the result?
- Does the model survive alternative rules?
- What claim does the output legitimately support?
20. Where This Fits
Spatial Overlay owns combining layers broadly. Spatial Data owns geographic data generally. This article owns cell-by-cell calculation across raster surfaces.
The Idea to Keep
Raster algebra is powerful because it makes geographic reasoning computable—but every cell still carries the assumptions that created it.