This repository contains a docker image for geostatistical data processing using standardized input and outputs.
This is based on the template for a generic containerized Python tool.
Currently, there are three tools implemented:
- Geostatistical variogram estimation
- Kriging interpolation
- Geostatistical simulation
In order to run an interpolation or simulation,you need to provide a valid variogram first. The variogram tool is accompanied
by extensive plotting to validate its geospatial robustness.
Variograms can be calculated for spatially distributed samples, or spatially dense fields. You need to provide either of
both inputs. If a field is used, it is highly recommended to subsample it instead of resolving all coordinates into a sample.
The tool defaults to a maximum sample size of 1000, which can be disabled by setting the sample_size parameter to -1.
To identify the spatial coordinates, the tool will look for standard column names, namely:
x, y, lon, lat, longitude, latitude. It is recommended to use projected data, but in case of field information, the
tool will reproject to the most likely UTM zone. Samples are never reprojected.
Samples with more than one non-spatial column will be estimated as cross-variograms. Fields with more than one variable
will yield multiple variograms. If many variograms are estimated, the kriging and simulation tools can iterate over
many inputs.