I am working with a collection of Landsat 8 images with pixel-wise NDVI computations over a period of time and I'm hoping to pin point areas within the region of interest where NDVI change is greatest. I have looked into different Hot Spot Analysis techniques and ArcGIS has a Hot Spot Analysis tool that seems like it may be a viable route to go. However, I am not sure how to move forward with my Google Earth Engine derived data set to compute hot spots of change between two image periods and further down the road, hot spots of change over a full grazing period for the region of interest.

Does anyone have experience with this type of change detection analysis and would you be able to offer guidance? Thanks!

  • A "hot spot" local autocorrelation approach is not appropriate for time series data. There is a bivariate version of the statistic but, it only accounts for 2 time periods. I think that what you want is the slope from a trend analysis. Take a look at the Kendal Tau statistic. – Jeffrey Evans May 22 at 22:43
  • Here is a tutorial on non-parametric trend analysis in Earth Engine - per-pixel time series slope and significance: developers.google.com/earth-engine/tutorials/community/… – Justin Braaten Jun 2 at 22:03

Your Answer

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

Browse other questions tagged or ask your own question.