‹ back to GIS Glossary

Interpolation Techniques

Definition

Interpolation techniques are mathematical methods used to estimate values at unsampled locations from known observations. In geospatial work they convert scattered point measurements—temperatures, pollutant concentrations, soil properties, well levels, cellular signal strength—into continuous surfaces that can be visualized and analyzed. Common approaches include inverse distance weighting (IDW), spline interpolation, natural neighbor, radial basis functions, and a family of geostatistical methods such as ordinary, universal, and co-kriging. Each method makes assumptions about spatial smoothness, stationarity, and directional trends. Good practice starts with exploratory analysis—checking for anisotropy, outliers, and non-stationarity—then picks a method aligned to the phenomenon and decision context. Cross‑validation, where a subset of points is withheld and predicted, provides unbiased error estimates. Analysts should also consider barriers (ridges, rivers), support mismatch (point vs. area), and the dangers of extrapolating beyond the convex hull of data. The output is not the final story: uncertainty maps and explanation of assumptions are as important as a pretty surface. Crucially, interpolation is not causation; it simply spreads measured information in space according to rules we set, so business or policy claims should be tempered accordingly.

Application

Environmental agencies interpolate air quality and heat to target interventions. Hydrologists estimate rainfall between gauges, then drive runoff models. Agronomists map soil nutrients for variable-rate fertilization. Telecoms interpolate signal strength to predict coverage gaps. Public-health teams interpolate clinic catchments to anticipate crowding. Because results can vary widely by settings, teams publish methods, parameters, and diagnostics alongside the surface so others can reproduce or challenge conclusions.

FAQ

How do you pick between IDW and kriging for a project?

Choose IDW for quick, deterministic smoothing when you lack a strong model of spatial structure; choose kriging when you can model variograms and want both predictions and quantified uncertainty. If computing time is tight, test both with cross-validation and pick the one that generalizes better for the domain.

What role does anisotropy play in interpolation?

If correlation decays faster in one direction than another—common with wind or slope-driven processes—account for anisotropy via directional variograms or anisotropic kernels. Ignoring it can smear features and misplace hotspots, especially along valleys or coastal fronts.

Can interpolation respect hard barriers like ridgelines?

Yes. Use barrier enforcement (cost-distance, hydrologic conditioning, or masked neighborhoods) so values do not unrealistically ‘leak’ across mountains, reservoirs, or walls. For coastal work, restrict neighborhoods to same-side littoral cells.

How should uncertainty be communicated to decision makers?

Publish maps of prediction error or confidence intervals, and annotate regions outside the data hull. Provide simple language about what the surface can and cannot support, e.g., “rank neighborhoods” vs. “set legal thresholds.”