# Correlation between a raster and a point dataset

I have an interpolated surface (in raster form) with temperature values and I also have the measurements of temperature in certain locations that belong to the same region (with the raster). I want to plot in a scatterplot the relationship between the raster (estimated values) and the measurements (observed values) and compute the correlation between the two datasets and also the line that best describes the relationship among the two datasets.

Any idea how can I do this in ArcMap or QGIS?

• Sounds like more of a programmatic problem. Do you use R or Python?
– Aaron
Commented Aug 13, 2014 at 19:43
• I suggest extracting the raster data to the points, and then use an external program (Excel or any stats package) to do the math and graphing. Commented Aug 13, 2014 at 19:46
• @Aaron No :/ I don't use R and Python Commented Aug 13, 2014 at 19:49
• @Erica is there any way that this can be performed in a GIS software? Commented Aug 13, 2014 at 19:49
• The first step of getting the per-point values, yes. The rest (calculating the relationship) -- I am not aware of one, but have never actually looked into it personally, just always gone straight to Excel. Commented Aug 13, 2014 at 19:58

## 2 Answers

The easiest way as @Erica suggests is to do the correlation analysis in Excel. To do that you need to have the data.

In QGIS this is easy to get with the Point Sampling Tool plugin. Install this through the plugin manager, add both your raster and your points to the map, and the tool can be used to create a new dataset with the associated data. The you can export the attribute table of the points from the table one of two ways: copying them from the table view and pasting them to Excel; or saving the file as a CSV (both ways are described in this GIS SE question).

ArcGIS is will do much the same thing with the Extract Values to Points tool. You can then save the table from the output dataset. Unfortunately it requires a Spatial Analyst license, so you might be better off going with QGIS.

• thanks a lot! I used Extract Values to Points and then got my data into Excel where I made the scatterplot, and computed the best-fit line with r-square as well. Commented Aug 14, 2014 at 20:00

Graph your Extracted dataTo point table, to make XY Graph (view-graph) . Graph has 'Curve fit' option.