Determining land cover change over time

Question

I am new to GIS and still learning. It’s interesting but becoming too complicated for me. A part of my project is to determine the land cover change in some counties in Texas from 2001 to 2011. So, a friend told me to use this approach using ArcGIS 10.3:

• I downloaded the Texas NLCD data
• I opened ArcMap, used the county shapefile to select the counties I’m interested in, I then overlaid the Texas land cover data (started with the 2006 data; just to try it out) on it.
• After selecting the counties, I created a specific buffer zone around them.
• Using ArcMap tools, I clicked spatial analyst tools →extraction→extract by mask→ it gave the option to select what land cover data I want in the input raster, I chose the Texas land cover data →then I selected the land cover data of my counties of interest into the input raster/feature mask, then selected an output file and ran it. The result was a number of legends with different colors indicating different land cover types. I am interested in specific land cover types (four actually). What can I do to get only the legends of the land cover types I am interested in?
• I also went further to click on the patch grid menu→then the spatial analyst (Fragstat interface), after which I chose the layer I wanted to use (land cover data of the counties of choice), clicked analysis by class and chose only class area under the class options menu, then I clicked run. I got an error saying Fragstat unable to run, and a table with no data output.

Is there anything I should have done? Am I using the right approach for this analysis?

Answer 1

This question actually deserves a much more complicated answer since LCLUC (Land Cover Land Use Change) is a specific discipline in geography that many people dedicate their life to researching.

To briefly contribute an answer, I will state that it sounds like you did an unsupervised classification. The variations in land cover that the tool spit out show differences in the pixels across the raster data you input. These differences can be determined in many ways but the tool you used uses one spatial analysis algorithm. You should look at the colors and what they represent, discover a pattern, determine what each land cover type is represented by which colors. Then be prepared to justify your reasoning.

As for the error - is there more information? Usually they have a description or a error code.

If you would like to learn more about LCLUC and the various tools that can be used to conduct this research, I recommend doing some good googling. For instance, I found this blog by esri.

I personally use a robust remote sensing software for LCLUC analysis - ENVI. They have excellent documentation on the various supervised and unsupervised classification methods available and what logic the software uses to determine its output.