Announcements‎ > ‎

Weekly update 24

posted Jul 25, 2014, 7:04 AM by Austin Milt   [ updated Jul 25, 2014, 7:19 AM ]
Other than having my second chapter rejected from Conservation Letters (woo), I have continued working on documentation for my GIS software's methods. Below is a series of figures from the latest methods I completed, which describes the single most important algorithm in my entire software. The algorithm is a least-cost path algorithm like you'd find in ArcGIS (in fact, it's really similar to Path Distance), but modified for my own particular situation.



This first figure shows a hypothetical set of rasters that define the optimization parameters for the least-cost path algorithm. I wont explain these in detail, but if you're familiar with Dijkstra's algorithm, then most of the surfaces should make sense. My algorithm allows for a secondary cost to be tracked which invalidates paths that are more expensive than a certain budget.


This second figure shows the last iteration (12th) in the optimization before it finishes. You can see in the rightmost rasters how the feasible least-cost path is being tracked from the top-right pixel.


And finally the coolest part of my variation on Dijkstra's algorithm: you get a different answer based on how much you're willing to spend on your path. Importantly for my GIS software, you may have to accept a worse level of impact (here "sediment mobilized") in order to satisfy the lower cost threshold, and in some cases, you may end up with no path at all.

Next week I'm off on vacation for about a week and then going to the Ecological Society of America's annual conference in Sacramento. I hope it's not a waste of time...