Reflecting land-use change

[AlexandraAppelbaum1]

Hi everyone

I am working with land-use data in Johannesburg at the moment. I have created maps for each year (1990, 2000 and 2013), but I now need to find a way to create a map showing land-use change between 1990 and 2013. Any ideas?

Someone who did the same mapping for a project in Durban categorised the data from 11 classes into three simpler classes (economic, residential and urban) and then clearly joined the data tables. Those maps show land that is in, say, the urban category in 2013, but not in 1990.

Firstly, do you have any ideas as to how to best represent land-use change (if not the method I've discussed)?

Secondly, how would you turn the current data set (attirbute table 1 in photos) into the other yes/no by category one (attribute table 2)? What tools, other than a table join presumably, would you need to run?

I can't seem to find a topic like this anywhere on here, but if I've missed it, please let me know.

Thanks!

Alli

[XanderBakker]

Is there any particular land use change that you want to emphasize? For instance like in the change in urban areas (expansion of cities) or change in forest? Would you like to aggregate the data into more general classes as you mentioned (economic, residential and urban). You can also tabulate areas (using some administrative areas) and use pie charts to symbolize for the 3 moments in time the distribution (% area) for each of the 11 classes (3 pie charts side by side for instance). There are many ways to symbolize the change in land use, but it really depends on what you would like to visualize. 

[AlexandraAppelbaum1]

Hi Xander. Thanks for the response. I think the emphasis should be on economic, residential and urban expansion, as in the case mentioned; with a different map for each of those three categories. Any tips on getting the data into a form that I can show that kind of land use change? Never done much work with stats programs, so although it probably seems basic, I'm at a bit of a loss.

The pie charts sound great, but think I should go with the simpler option for now.

Thanks!

[XanderBakker]

One of the things you could do is to use the Lookup—Help | ArcGIS Desktop tool to reclassify the raster and create a raster for each class (make sure to assign for instance 0 to all none relevant classes and not NoData). For this you will have to add new field to the raster attribute tables and fill them with the appropriate values. 

You mention that you want to know the land use change between 1990 and 2013. Is what happened in between relevant (is the raster representing 2000 going to be used)? It might be interesting to distinguish urban expansion between 1990 and 2000 and the part that occurred between 2000 and 2013. If that is relevant you should use all three rasters. 

After obtaining the three categories for each year with the lookup tool, you can combine the rasters using the Combine—Help | ArcGIS Desktop tool, which will create a new raster representing all the different combinations of the input rasters. You can execute it twice to include the 2000 raster too and being able to distinguish when expansion occurred (1990-2000 or 2000-2013). Use the attribute table to understand the new values assigned.  

If possible, please share a small part of the 3 raster to provide a more detailed description of the steps.