Is there an equivalent of ENVI's Change Detection Statistics in ArcGIS Pro?

1349
2
09-09-2020 09:50 AM
OliviaSpencer
New Contributor

I am looking to see LULC between two TIFF in ArcGIS Pro. I have 4 classes in each image that are equivalent between the two. I want to do a pairwise comparison to see percent change between each class pairing. This is easy in ENVI using the Change Detection Statistics tool, but I no longer have access to ENVI. I have already tried utilizing Raster Calculator and Cell Statistics, but that does not allow me to see what the actual landcover change. Is there a way to achieve this in ArcGIS Pro?

0 Kudos
2 Replies
by Anonymous User
Not applicable

Hi Olivia. With the latest release of ArcGIS Pro 2.6, you can compare two categorical rasters and generate a table to get information about the amounts and types of change. The Compute Change raster function can compare two land cover datasets and generate an output containing all the possible land cover transitions that occurred.

Because this is a raster function, you have to save the layer as a new raster before you can get pixel count information and transform that into area. 

1. Run the Compute Change raster function
2. Export the function layer as a new raster dataset (use the Copy Raster tool or the Export Raster pane)

3. Run the Build Raster Attribute Table tool to add the "Count" field to the dataset (do not choose to Overwrite the existing table)
4. Add a field to the dataset and use Field Calculator to generate whatever information you want. You can get the total pixel count and divide the count for each transition by the total to get percentage, or you can multiply the count by the pixel size to get the area. 

This is currently a multi-step process, but we are working on making this whole process much more streamlined in ArcGIS Pro 2.7.

MervynLotter
Frequent Contributor

What would then be really sweet is if you could then have the option to view the change detection results in a sankey chart ...

Creating Sankey Chart in Tableau! - The Data School Australia

0 Kudos