Use space in your favour

From geometries to geometries

Data comes linked to geometry A

Analysis happens on geometry B

Interpolation techniques

Polygon to polygon

Point to point

Areal interpolation and dasymetric mapping

Data are on one set of polygons

Analysis happens on another set of polygons

proportionally transfer data
from one set to the other

Areal interpolation

Make this Notebook Trusted to load map: File -> Trust Notebook

Overlay geometries

Get proportions taken by each source geometry

Redistribute values proportionally to area

Dasymetric interpolation

areal interpolation with ancillary information

Overlay geometries

Get proportions taken by each source geometry weighted by ancillary data

Redistribute values proportionally to weight

Pycnophylactic interpolation

no sharp boundaries exist

Create a smooth volume-preserving contour map from source data

Redistribute values from the contour map to target geometries

pycnophylactic interpolation does
not work for intensive variables

Point interpolation and kriging

Data are on a marked point pattern

Analysis requires values in between

estimate values in unobserved locations
based on the spatial distribution of
values on observed locations

Assign the value of the nearest point

Model value based on neighbors

K-nearest neighbours

Distance band

Use distance (or not)

uniform (binary) weights

(inverse) distance-weighted

Ordinary Kriging

linear combination of observations that are nearby

which takes into account geographical proximity,

spatial arrangement of observations,

and pattern of autocorrelation

create experimental variogram

fit theoretical variogram

use theoretical variogram to model values

import tobler
import pyinterpolate