USRI Summer 2021 Research Project
A research project on the interaction of deer and hunters in Oklahoma
This project analyzed data taken during the months of November and December of 2008, and 2009 in the state of Oklahoma. The goal was to determine whether or not an interaction of a deer and a hunter was categorized as a GPS interaction or a visual interaction. In other words, based off of proximity, elevation data and land use data, did a hunter potentially see a deer?
Data clean up
Hunter Data
First and foremost, the most important part of any research study is the data. In order to properly analyze the data I was given, some organizing was needed. The first half of my summer was spent cleaning up the data to make it possible for further analysis.
Essentially the data was all combined into one field and the 'categories' were separated by a TAB. I had to convert each category to its own field so that the data could be more easily understood by the software. To do so, some coding was needed. There was an excel file for each day in the study period each containing an extremely large number of GPS points. In ArcGIS Pro, you can work on python scripts within the software and connect the code to the maps you are working on.
Deer Data
Luckily the data for the deer was already in a format that was compatible with the software.
Here is the outcome of simply adding the csv files for the hunter data. The GPS tracked these hunters every minute for both November and December of 2008 and 2009.
This is the result of simply importing the csv files of the deer data. During the same time period, a set of tagged deer were tracked by a GPS every 8 minutes. Now, these blobs of dots don't tell much of a story... this is where my analysis truly started.
Next, I was given a approximately 30 pdf files each containing an average of 70 scanned cards from the hunters in Oklahoma. An example can be seen below.
Every time a hunter stated they saw a deer and at one time, I loaded their information along with the information of the deer they saw into an excel file. Essentially, I was creating my own data file to compare with the GPS file.
Each time a hunter went out and the GPS tracked their path, I refer to this as an outing. If the hunter saw a deer during an outing they filled out a card as shown above.
My second python code went though and parsed out all of the data where we had an outing that contained a filled out hunter card. Therefore I focused on the outings in which the hunter had a sighting.
It is important to note that when a hunter filled out an observation card they may not necessarily have been observing a GPS tracked deer, therefore we may have positive observation cards for which we have no GPS tracked deer. Another note could be that the hunters may or may not have filled out the card at the exact time they saw a deer. (we just have to trust them)
This Map represents the hunters data with a leg speed below 10km/hr indicating the hunters were walking at these points. The squares represent a 'heat map' of where the most concentration of points occurred over the study period.
Here are the points and heat map of the hunters data that has a leg speed of over 10km/hr indicating that the hunters were driving.
Finally, this is the heat map and points of the hunters data with a leg speed value of 0km/hr, indicating that they were standing still.
For the purpose of simplicity and logic, I decided to focus more on the hunting data of which the hunters were walking and standing still as this is when they are most likely to have an interaction with a deer.
Lets take a look at one outing. Here we have the points and path line for a hunter on November 23, 2008 during the midday hours.
I obtained the land use data from Esri's Living Atlas.
I used another python script to extract the deer points that fell within 2 hours of each of the outings. Here are the deer points associated with the time of our example.
Here is a buffer to help visualize the 150m distance that I assumed the hunters could see.
I then searched for high resolution elevation data and ortho imagery (aerial photography) OKmaps.org was where this data was found. I found the elevation data derived from the LiDAR - a way of using laser data to get elevation data (x,y,z). Basically I extracted for my study area, 2m elevation data as rasters and 2m ortho imagery in order to create a viewshed. This is a polygon representation of what the viewshed created.
Here we have a side by side comparison of the buffer and viewshed. They both span 150m all the way around the hunters path but represent something completely different. If we were to use the buffer (left) as our assumption of what the hunter could see, this would neglect to consider factors such as forest cover as well as elevation. Using the viewshed (right) this allows for a more accurate visualization of what the hunter can see at ground level. I chose to use the first return DEM to base the viewshed off of.
In comparison to a first return DEM (Digital Elevation Model) we have the bare earth DEM. The bare earth DEM uses only the elevation of the ground while the first return DEM considers forest and shrub to be a part of the DEM. I felt that this would allow for the most realistic viewable areas for the hunter.
Finally I used the path of the deer within the same time period with the polygon created from the viewshed, and ran the intersect tool. This resulted in showing exactly where the deer's path went right through the viewable areas for the hunter. Therefore rendering this a confirmed visual sighting of a deer. The light pink highlighted areas represent this.
A final note. Each of the geoprocessing tools and steps I took were done in what is called batch mode. This allowed me to simultaneously complete the analysis for multiple hunters paths at one time. But, only one was displayed for the purpose of simplicity.
This screenshot contains an example of what the software I have been using looks like. On the right hand side you can see my three python scripts under the Notebooks Tab and how easily they can be integrated into the project at hand.
For the remainder of the summer I will be creating organized files that will allow anyone to replicate the work I have done and see step by step how everything was achieved. I will also be analyzing a data file that contains a list of contacts solely based off of GPS proximity. That way, I can compare the results of the stated contacts versus the GPS contacts.
