I have a temperature image (interpolated from climatic point data) and NDVI data at the monthly interval.

I need to do correlation analysis between temp and NDVI.

Is it impossible to do this using ArcGIS Desktop, and if so, can you tell me how to do it?

  • 1
    are you looking to do a spatial correlation (spatial autocorrelation) or a regression analysis? – artwork21 Jun 14 '11 at 11:37

It is usually a mistake to estimate correlations between grids when one or both have been interpolated: the estimate often reflects the interpolation method rather than the data themselves. Moreover, you will always grossly over-represent the degrees of freedom, typically leading to falsely high estimates of confidence and precision.

When one dataset is a grid of measurements, or derived from measurements (as NDVI is) and the other is a point dataset, it's both better and simpler to extract the grid values at the point locations, forming an (x,y) dataset. In this case x would be the temperature at one of the points where temperature actually is observed and y would be the NDVI value at that same location. There will therefore be one (x,y) datum for each temperature datum available to you.

ArcGIS is not the software to perform a rigorous analysis of an (x,y) dataset: as @Mike Toews suggests, use statistical software. There are Web apps that will compute correlations, perform ordinary least squares regression, draw scatterplots, and give you confidence intervals for the parameters. Even handheld calculators will do this... In general, though, for anything except self-instruction, you're best off using well-tested software such as R or a commercial statistics package.

| improve this answer | |
  • GeoDa is a good, simple, geographical statistics package that would do this nicely for you. – Chris Strother Jan 8 '15 at 15:03
  • @ChrisStrother but one of the main dis-advantage of geodata is it does not support raster datasets – SIslam Nov 17 '16 at 7:40
  • @whuber How about if the interpolated surfaces have consistent parameters used in creating them, just the underlying variables in the point dataset has changed (e.g. different time period)? – Simon Jun 6 '17 at 11:04
  • @Simon It hinges on what you mean by "consistent parameters." Consider two IDW grids where only the values at the data points (but not their locations) have changed from x_i to y_i, i=1,2,...,n. That gives three "correlations": the true underlying correlation between the grids; the correlation coefficient of the data (x_i,y_i); and the correlation coefficient computed from paired grid cells. The latter are data of the form (A_jx,A_jy) where A_j is the coefficient vector at cell j. Those data are highly interdependent and redundant. At best, using them would overcomplicate things. – whuber Jun 6 '17 at 14:20

First, make sure your rasters have the same shape (rows×cols; bounds). You can perform a multivariate regression analysis with the Spatial Analyst extension for ArcGIS. I'd start with this approach.

If you want to get really fine-tuned/serious, you can read both data sets into R (via rgdal), then perform a suite of statistics on the co-located grid locations or rows of a combined SpatialPixelsDataFrame/data.frame object (e.g. GLM, ANOVA, etc.). This approach will confuse most people, so I would only go in this direction if you are brave or curious.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.