Data Analysis

Problem

Identify shortfalls in the South Dakota Wildlife Damage Management program (location of staff stations, amount of damage staff, hotspots locations, etc.) through identifying drive times and hotspot locations.

Spatial Analysis

Importing Data

All of the downloaded data from the previous update has been added to map project. This includes:

• Wildlife damage management complaint points

  • This data had to be projected to NAD 1983 UTM Zone 14N Projection using WGS_1984_(ITRF00)_To_NAD_1983 transformation

  • Using 'Show XY Data'

    • This populated all of the complaint points onto the map with the XY coordinates provided by the table.

• Wildlife Damage Management regions and districts

  • A shapefile provided by South Dakota in NAD 1983 UTM Zone 14N Projection with WGS 1984 geographic coordinate system

• Wildlife Damage Management Specialist duty locations

  • This was a .png image provided by South Dakota State Game, Fish, and Parks that had to be georeferenced onto the state using the Wildlife Damage Management regions/districts shapefile

  • There were stars to represent the rough duty locations for the specialists. Accurate locations were not provided for confidentiality reasons as many specialists work from home

    • I created a point feature class and used an edit session to create points that represented the duty locations on the map

    • I added a name field to this feature class and input the names of each damage management specialist for each point, this will make it easier for network analysis and another field might be added for the region-district that they are responsible for.

Quality Control

• Complaint Points

  • Some points were outside of the state of South Dakota or had incorrect coordinates causing them to be way off the map. I selected all of the points within the Wildlife Damage Management shapefile, the inversed that selection to have all the points outside of the state selected. Then I created a new feature class with these points and named it 'Complaints out of Range', then deleted those points from the original file.

    • I still wanted to keep these points in case I needed them for some reason, for instance if my boss decided he wanted me to use the ones on the fringes of the state.

Spatial Analysis

• Average Nearest Neighbor Tool

  • I used 'Average Nearest Neighbor Tool' to find the significance of the points and determine if they were in fact clustered (Figure 2).

  • The points were in fact clustered and not due to random chance so I know as that the hotspots and their certain points are significant. This shows a pattern that the state of South Dakota can reference when making decisions pertaining the the Wildlife Damage Management Specialists.

• Kernel Density Tool

  • In order symbolize the complaint hotspots and find their centers I used the 'Kernel Density Tool'.

    • I set the parameters to use square kilometers being such a large area and kept everything else set to default.

  • This provided me with a nice visualization of the clustering and the density of clusters.

• Reclassify Tool

  • I reclassified the raster provided by the 'Kernel Density Tool' into 10 classes of equal interval (class 10 = highest density, class 1 = lowest density).

    • 10 classes was determined through trial and error on what provided the most usable breaks in the data, unlike the 'Coyote Home Range' lab, I was not concerned with the specific chance of finding a complaint, but with the most likely places to find a complaint.

• Raster to Polygon Tool

  • I converted the reclassified raster to polygons so that I could find the centroid of the polygons

• Feature to Point Tool

  • This was a new tool that I had never used before so some research had to be done on how it could be used for what I needed.

  • What this tool does is create a point feature class based on the input feature class. I was interested in using polygons and this tool will find the centroid of those polygons. For further reading click here.

  • After these points were made, I had to do some more quality control because I wasn't interested in every polygon's centroid, only those of the highest density. For instance, the centroid of polygons in 'Class 1' were of little interest to me or center of polygons that were too close together.

    • I deleted unnecessary centers in an edit session by choosing them arbitrarily based on my knowledge of how the duty locations and regions-districts work. I didn't want to do network analysis on points that weren't true hotspots or on centers that were close enough to the center of a denser hotspot.

Progress Update

All needed data has been added at this point. The network analysis that will be completed will use data from arcgis.com to find routes and route times.

Everything has been processed except for the network analysis which I am still trying to figure out the best way to do that. There are multiple options that can be further researched here. I think that using the 'Closest Facility' analysis will prove the most useful, but am still playing with the parameters. I would also like to restrict the routes based on the boundaries of the region-district each specialist is responsible, but am still struggling to find a way to do that easily. It may end up being that I create separate network analysis layers for each region-district, but I am trying to figure out if there is a better way to do that.

Ideally I will be able to show current drive times as well as recommend new routes, duty locations, or district boundaries, but we will have to see if that is within the scope of the time frame allocated to this project. This will depend on how difficult the network analysis proves to be.