Thursday, April 6, 2017

Never Mind: One More Post

I forgot that I had one more assignment to post before I'm done with my blog for good.  I had to organize or attend a "GIS Day event," and since GIS Day was back in November, I had to organize my own.  I set up a temporary museum exhibit at a library in Elmwood, Illinois, where I had some educational materials pertaining to both archaeology and GIS.  I had previously conducted an archaeological survey on a farm in Illinois, and found about 360 artifacts over a 78-acre survey area, so I set up some of those artifacts at the library, along with maps of the survey and illustrations of prehistoric life.  The maps were made with ArcGIS for Desktop, but the data was collected with a Trimble GNNS receiver outfitted with TerraSync software.  I invited a fair amount of people, but nobody in town came to see the exhibit, which is about what I was expecting (and sort of hoping for), so I got to sit in the library and read all day.

Sunday, March 5, 2017

Final Blog Post

I'm nearing the end of the GIS certificate program at UWF, and this should be my final blog post.  I've attached a link to my online GIS portfolio here, which includes my resume and some samples of the maps I've posted on my blog.  I've also posted the audio introduction to my portfolio here.  Please don't listen to it; it makes my voice sound funny.

Monday, November 28, 2016

Final Project

My final project was to create a predictive model for locating Scythian kurgans in southeastern Kazakhstan.  I looked at the locations of known kurgans on the Talgar Alluvial Fan, and from that, I tried to extrapolate the probable locations of other kurgans in a wider study area, based on the known kurgans' relationships with variables such as slope, elevation, and the availability of water.  The map below shows the locations of the 187 known (or possible) kurgans on the Talgar Alluvial Fan.

The next map shows the extent of the new study area that I defined, laid over a satellite image of the region.  The known kurgan locations are also shown within it.

The next map shows the kurgans laid over a U.S. Geological Survey DEM (Digital Elevation Map).  All of the known kurgans are located on the alluvial fan, at the base of the mountains to the south.

Using the U.S.G.S. DEM, I created a topographic contour map, shown below.  This shows the exact elevation at which the kurgans occur (generally below 1000 meters).

Using this data, I reclassified the DEM and clipped it to the study area, creating a map that shows probability by elevation:

Next, I tried to evaluate the relationship between slope and the locations of kurgans.  I used the DEM file to create the following slope map:

And using this data, I again reclassified the DEM and clipped it to the study area, to create a map that shows probability by slope:

The final variable that I consulted (and chose to include in the final model) is the availability of water.  The next map shows the locations of kurgans within a two-kilometer buffer of the streams that run across the alluvial fan:

After finding that 92% of the kurgans are within two kilometers of a stream, I created the following reclassified image, clipped to the study area, which shows probability by availability of water:

And finally, after inputting all of the above data into the "Weighted Overlay" tool in ArcMap, I created the final predictive model, shown below:

According to the model, the green areas within the study area should have a high likelihood of containing additional kurgans.  The orange areas should have medium probability, and the red areas should have low probability.

Saturday, November 5, 2016

Module 10: Supervised Classification

The map above shows a supervised land cover classification, made in ERDAS and ArcMap.  I started with a satellite image of Germantown, Maryland, and I took "signatures" of eight specific types of land cover (shown in the legend above).  In ERDAS, I examined the mean plot for each signature, in each of six bands.  The three bands that showed the most difference among the various signatures were 4, 5, and 6, so I used those three bands to display the image before classifying it (using the following scheme--red: 4, green: 5, blue:6).  Then I used "Maximum Likelihood Classification" to create a supervised classification of the image, by pixel color.  It took me roughly forty attempts to create the image above, because it was very difficult to distinguish roads from other features.  The inset map in the bottom left shows the "distance file"--in this image, the lighter areas indicate spots where the classification is more likely to be inaccurate.

Sunday, October 30, 2016

Module 9: Biscayne Shipwrecks (Analyze Week)

The map above shows the benthic (ocean floor) types in a half-kilometer buffer around each of five shipwrecks in Biscayne National Park, off the southeast coast of Florida.  It also shows an outline of the whole park, laid over a satellite image of the area.  The label for each benthic type (such as reef terrace, pavement, etc.) is color-coded with the colors on the map.
This next map, above, shows the reclassified versions of both bathymetric and benthic images of the park.  In the bathymetric image, the green areas are shallower, and the red areas are deeper.  The green areas have a higher weight in the legend, because shallow waters are more likely to contain shipwrecks.  In the benthic image, to the right, the different sea floor types (described above) are classified according to depth.  Again, the green areas are shallow, and the red areas are deep.
The final map shows a predictive model of the eastern part of the park.  This predictive model is meant as an aid in surveying for underwater shipwrecks.  The green areas have a high probability of containing shipwrecks, due to the presence of reefs or the shallowness of the water.  The red areas have a low probability.

Monday, October 24, 2016

Module 9: Unsupervised Classification

The map above shows an unsupervised classification of an aerial image.  Each pixel in the image was classified according to its exact color shade into five categories:  trees, shadows, grass, buildings/roads, and mixed.  The red pixels are mixed:  they indicate places where the grass was the exact same color shade as rooftops or roads.  This image was classified in ERDAS Imagine, originally using 50 color classifications.  The 50 colors were narrowed down to five.

Module 8: Biscayne Shipwrecks (Prepare Week)

The map above shows the location of five shipwrecks within the boundaries of Biscayne National Park, off the southeast coast of Florida.  Each ship was wrecked on one of the park's many reefs.  The top left data frame shows the shipwreck locations (and ship names) over an NOAA ENC (Electronic Navigational Chart).  The bottom left data frame shows a bathymetric DEM image, and the bottom right data frame shows an 1856 navigational chart of the Florida reefs, georeferenced in ArcMap and clipped to the park's boundaries.