Thursday, August 9, 2018

GIS5265 - GIS Applications for Archaeology - Final Project


                Camino de Mulas was a long distance trade route stretching from the general location of San Jose, Costa Rica and ran down to David, Panama. The route was operational from around 1601CE – 1730CE, and was eventually destroyed by the building of the Panamerican Highway. While parts of it may still remain, the exact location of the route is no longer known. Historical sources provide us with a list of known locations that intersected the route at the height of its activity, and can be used to recreate the possible location of the route by performing a series of Least Cost Path analyses in ArcGiS. While there are many unknown factors that may have affected the exact direction of the route, a good estimate of its location can be generated using basic known information and a little bit of common-sense speculation in regards to the current nature of the landscape.
                I began my process by creating individual point shapefiles, and placing them at the coordinates given to us in the lab document and labeling each location by its provided name. This step gave me an idea about the size of the area as a whole, which would in turn help me locate an appropriate DEM raster. After retrieving a large DEM from the Earth Explorer website, I used the clip tool to decrease the size of the raster to reach just above the northernmost location and just below of the southernmost location. Doing this would decrease processing times for all tool operations that will need to be run when finally creating a weighted overlay raster. Because the final result would require more data than elevation, I took time to track down a land classification shapefile for the general region of Central America. I clipped the shapefile, and after taking some time to classify the land cover categories based on estimated difficulty in the features attribute table, I used the Features to Raster tool to convert it to a workable raster.
Once I had two workable rasters, I began my work to create a weighted overlay raster. I began by creating a new toolbox, and within it a new model. Much like in the first part of the final project, I began by going into the models properties, and replicating the Values set up in the previous step (Workspace, Raster Analysis and Processing Extent.) I dragged the slope tool and used the provided standard (.00000912) for my Z value. The raster input was, of course, the DEM downloaded from Earth Explorer and had the output raster was set to be created in a whole new folder designated for the second part of the project. I added the reclassify tool to the model and connected the two operations together. I recreated the same setup for my Slope raster as I had done in the first part of the lab. (10 grade categories).
After successfully running the tool, my Reclass raster was successfully created. Now, the time has come to combine the factors from both rasters to create a Weighted Overlay. I dragged the Weighted Overlay tool into my model, and set the Evaluation Scale as: from 1 to 10 by 1, as previously instructed and aligned the evaluation scales so that they match up (this time I set the NoValue field to 9 instead of 1 in attempt to give water more weight.) I added the Land Cover raster, and weighted the previously defined classes as follows: easy – 1, medium – 4, hard – 9. I set the ratio of influence as 80-20 favoring the Reclass raster and ran the tool.
As I examined my overlay product, I noticed that the entire water region showed up within an Easy-Medium range. I found that this may be problematic when predicting the Least Cost path. The lab instructions specify that Camino de Mulas was a mule route, and should not include any sort of water travel. I decided to retrace my steps and give the Land Cover raster more weight in order to contrast the results at the end. I made another Weighted Overlay raster, this time favoring the Reclass raster 60 – 40 to contrast alongside of the first raster. My hope was that by giving more weight to the Land Cover (which did not contain any part of the ocean) would help divert the least cost route away from the water for more accurate results.
I moved on to the next step, and took time to create all of the Least Cost Paths between the points. Because using a single model gave me many problems in the previous part of the lab, I decided to calculate each Least Cost Distance and Least Cost Path individually. This is where I took time to re-name my location point files by adding an order number after their name. (This will make it easier to work in order.) I used the Least Distance tool on my origin point and created a new backlist raster. Then, I used the new products to calculate the Least Distance Path between the first two points. I followed these steps for each individual path, for both of my weighted rasters until the entirety of the Camino de Mulas was mapped.
At the very end, I created two new line features, and traced over the Least Cost Path in order to make it more legible at the large scale. I added a shapefile of Central Americas road systems to account for any overlap, created a new data frame and set up my map layout.
As I examined the contrast between the two different weighted rasters, and it seems as though the 60-40 raster is more likely to be more accurate. I conclude this based on the direction the path takes around the area of Paso Real, Boruca and Palmar Sur in the 80-20 weighted raster. After approaching this area along the coastline (presumably weighed down by the water) the path cuts across Palmar Sur, and moves north to Boruca, then to Paso Real and cuts back to Palmar Sur. This seems to be counter-productive, and the historical document provided clearly states that the Railway Section near Palmar Sur ought to be the third stop of the above mentioned locations. The Least Cost Paths generated using the 60-40 Weighed Overlay Raster seem to provide a much more reasonable vision of the direction that Camino de Mulas had most likely taken.


No comments:

Post a Comment