Hawaii SWARS Workshop Day 2

Did I say yesterday that the workshop was gonna get more intense? There you had it! What say you?!

Congratulations to everybody! If you survived today, it's gonna be a very smooth ride for the rest of the week. Indeed, all of our participants did a great job. Congratulations!

I have repeatedly said that, for this SWARS analysis, the actual model building and running are probably the least you should be concerned about. Those tasks are very mechanical and straightforward. Tools are already provided by ArcGIS. All you need to do is to fill in the blanks. The real challenge, for us the GIS specialists, is to get the input layers ready.

And getting all the input layers ready, from whereever the starting point is, can be a very long, boring, tedious, and frustrating march. You should see why that is now already after today. Did you not?

Having a thorough understanding of this data preparation process as well as all the techniques involved is absolutely critical. Only by truly understanding the process can you then be able to conduct the actaul SWARS analysis on your own, and handle any but definitely will occur "variations" along the way.

Let's review a few key points here:

• The ISSUE will dictate what layers should be used and how they should be ranked. Therefore, having a clear and well refined definition of the issue is the first crucial step!
• How a layer should be used to address a particular issue should be carefully studied. The same layer could be used completedly differently depending on the issue. For example, the same slope layer, should you give the maximum value to areas with slope greater than 45 degrees or areas with slope <>
• You could use the exact same set of input layers for different issues but have not only very different layers weights but also different value classification/scheme for the individual raster layers.
• Selecting the raster value classification/scheme is a quite subjective task. Say you need 5 values. You can use (1, 2, 3, 4, 5), or you can use (10, 20, 20, 40, 50). But because what we are pursuing is the relative importance, these two seemingly very different value classifications/schemes should produce the same result, as long as we apply the same value classification/scheme across all the input layers.
• To go from the original vector shapefile or feature class to the model ready raster layer with the corerct value classification/scheme, you have more than just one way to achieve that. You could run the feature class to raster conversion first and then reclassfiy the raster into the slected value classfication/scheme; or you could first add the correctly value classfication/scheme to the feature class/shapefile's attribute table, and then run the feature class to raster conversion based on the values you already put in.
• It's always beneficial, and preferably, to create a "master" or "boundary" raster first that will cover and only cover your land area. This "master" raster should have the right Cell Size, Projection, and Spatial Extent. You will use this master raster to set your spatial analysis Environment which avoid many unnecessary errors. You can also use this master raster to fill in holes (NODATA cells) as well as mask out ocean cells after the model running.

Well, that's a lot to digest already.