CODEWITHSUNDEEP
pythonrgeospatialdata-generation
Blue Moon
Blue Moon

Reputation: 4651

how to create geo-referenced random data?

I'm writing a dissertation on different parameter settings in GWR (https://cran.r-project.org/web/packages/spgwr/vignettes/GWR.pdf). In order to perform tests I'd need the availability of geo-referenced data that show some sort of spiatial correlation (i.e. values are not independent of their geographic position but vary according to space having zones with high values as well as zones with low values). Datasets should have geographical coordinates (i.e. latitude and longitude) and their density patterns should vary.

as an example:

two centroids

one centroid

uniform

edges denser than center

A good starting point for me was reading this question: Generating multidimensional data and even more this question on http://gis.stackexchange.com https://gis.stackexchange.com/questions/35159/how-can-i-conduct-geographically-weighted-principal-component-analysis-using-arc

However I'm stack in the process of giving these data a spatial autocorrelation. Any idea on how to achieve this? I would appreciate if the code is in either Python or R.

Upvotes: 1

Views: 374

Answers (1)

Gregor Thomas
Gregor Thomas

Reputation: 145775

The spatstat package has lots of functions for generating random "planar point patterns". Lots of details in the docs. Starting on page 26 they have a list of 29 included functions for generating random point patterns by a variety of methods---some cluster-based, some homogeneous, and some other.

With spatstat, you generally define a "window" in which to generate points. It's up to you if the units you use are latitude and longitude (you seem to make a particular point of this in your question, but it's just units---if you care about sampling over a big enough area that the curvature of earth plays a role then you can probably do some coordinate transforms).

Upvotes: 3

Related Questions