Fill NoData by Kriging Interpolation
Overview
Kriging interpolation estimates unknown values from spatial autocorrelation described by a variogram model. It is suitable for producing continuous raster surfaces or filling NoData gaps when spatial structure is meaningful.
Use Cases
- Interpolate environmental, meteorological, soil, or geochemical variables with spatial autocorrelation.
- Prepare spatially continuous raster input for later zonal statistics, threshold extraction, or map production.
- Fill NoData gaps where a variogram-based model is appropriate.
Parameters
| Parameter | Description | Notes |
|---|---|---|
| Input raster file | Raster file that requires kriging interpolation | Required |
| Variogram model | Variogram model used to describe spatial correlation | Required; default: spherical; options: spherical (spherical), exponential (exponential), Gaussian (gaussian), linear (linear), power (power) |
| Nugget | Error term representing microscale variation and measurement error | Required; default: 0.1 |
| Range | Maximum distance of spatial autocorrelation | Required; default: 1.0 |
| Number of lags | Number of lags used to calculate the variogram | Required; default: 6 |
| Weighted interpolation | Whether to weight interpolation points | Optional; default: False |
| Exact-value interpolation | Whether to use exact known point values for interpolation | Optional; default: True |
| Anisotropy scaling | Scaling factor in the anisotropy direction | Required; default: 1.0 |
| Anisotropy angle | Anisotropy direction angle in degrees | Required; default: 0.0 |
| Coordinate type | Coordinate type used for distance calculation | Required; default: euclidean; options: Euclidean distance (euclidean), geographic distance (geographic) |
| Use pseudo-inverse | Whether to solve with a pseudo-inverse matrix | Optional; default: False |
| Pseudo-inverse type | Method used to calculate the pseudo-inverse matrix | Required; default: pinv; options: Moore-Penrose pseudo-inverse (pinv), singular value decomposition (svd) |
| Verbose output | Whether to output detailed interpolation process information | Optional; default: False |
| Plot result | Whether to plot interpolation result charts | Optional; default: False |
| Output statistics | Whether to output interpolation statistics | Optional; default: False |
| Output raster file | Output raster file after kriging interpolation | Required |
Steps
- Start the tool: Open Geoprocessing Toolbox → Spatial Analysis Tools > Raster Synthesis > Fill NoData, then start the Fill NoData by Kriging Interpolation tool pane.
- Prepare input: Select the Input raster file and confirm that the data is complete and readable.
- Set core parameters: Configure Variogram model, Nugget, Range, and other kriging settings based on the analysis goal.
- Set output: Specify the Output raster file and confirm that the output path, format, and naming rules meet downstream requirements.
- Run and inspect the result: Click Run, then check whether the output value range, distribution, and raster extent meet expectations.
Notes
- Kriging results depend strongly on the variogram model and parameters. Validate the model when using results for quantitative analysis.
- Use a projected coordinate system when Euclidean distances are used.
- Sparse valid cells, uneven distribution, or outliers can reduce the reliability of the interpolated surface.