Researching Deer Using Gis Software

RESEARCHING DEER USING GIS SOFTWARE

Brian Foley

Computer Science Department

University Wisconsin-Platteville

Abstract

Global Information Systems (GIS) are revolutionizing the study of wildlife animals.

GIS Software allows researches to see patterns and visualize data to give them a better understanding of the facts. Software that will be covered in this paper will be Location Of A Signal (LOAS) and ArcView. Data calculated from LOAS is used directly by ArcView to make comparisons of information that otherwise could not have been made. In this paper, the University of Wisconsin-Madison CWD White-tailed Deer Research Project has specific routines that are done to get the needed information for LOAS. Other facts such as how LOAS is used and how locations are calculated are also covered. The UW-Madison Research Project also uses software called ArcView, however this will only be covered briefly. Together, LOAS and ArcView form a powerful tool for the study of wildlife.

Introduction

Computer Technology is allowing experts to learn about white-tailed deer in ways that could have never been possible in the past. Software such as LOAS (Location Of A Signal) and ArcView, are packages currently used by the ecology department at the University Wisconsin-Madison, to study topics that could not have been effectively studied by traditional pre-technological ways. For example, UW-Madison is currently involved in the Chronic Wasting Disease White-tailed Deer Research Project. Chronic Wasting Disease (CWD), is a disease infecting deer, very similar to “Mad Cow Disease” in cattle. CWD was originally located in western states such as Colorado and Wyoming. However in February 2002, it was discovered in southwestern Wisconsin [1]. The Research Project is studying deer dispersal, movement, and home ranges of the deer. This will help determine the rate at which CWD may spread across the area. LOAS is only used for the location of the animals (specific x,y coordinates) and ArcView is used to make comparisons of data which is then graphically represented.

CWD Area

Figure1. Wisconsin locations of CWD deer [2]

Gathering of Data

To determine exact location of deer, VHF radio collars are placed on deer. The collar consists of a neoprene impregnated cotton duck which is double layered around the antenna which exits out the top of the collar. Silver oxide batteries from three to six volts are used and can send a signal up to 400 yards [3]. Each collar is set to a unique frequency. Through radio telemetry, deer are tracked three to ten times per week and at various times per day. The times of day signals are recorded range from 4am, 8am, 12pm, 4pm, and 8pm. If a deer moves out of its calculated home range, it is then monitored everyday. In addition to location, type of movement, rate of dispersal and distance traveled are also recorded.

To locate a specific collared deer, technicians are sent to at least 4waypoints. Waypoints are areas along roads that are easily recognizable. Theyare oftenroad intersections, end of driveways, or simply next to a road sign. When a place is determined to be a waypoint, GPS (Global Positioning Satellite) is used to determine the UTM (Universal Transverse Mercator which is discussed below) x,y coordinates. All of the waypoints are transferred into a master list and carried with the technicians. When a technician arrives at a waypoint, they take out an antenna and radio receiver and quickly dial-in to the frequency of the deer they are trying to locate. When the signal is detected, several things need to be recorded. The first is the direction that the signal is coming from. This is recorded in degrees and is determined by a compass. This direction in degrees is called the bearing. At least 4 bearings,one at four separate waypoints are used. Often if signal intersection due to signal bounce is not encountered, more bearings are taken. This increases the accuracy of the deer’s location. Other things that are noted are the strength of the signal. This is a guess based on the technician’s opinion. This is rated anywhere from 0 to 5. If the signal is close (meaning the deer is close), it will be rated closer to 0, and if far, a rating closer to 5 is given. Signals with a rating of 4 to 5 are not used. Using bearings with a week signal can affect the accuracy of the deer’s location. Also noted, is whether the deer is active or inactive. This is determined by the interval of beeps emitted from the collar. Slow steady beeps indicate a deer not moving, while faster beeps means a deer is moving. Once the4 bearings are recorded LOAS is able to determine the location (specific x,y coordinates) of the animal. LOAS requires at least 2 bearings, but more will give a more accurate location. Other data such as temperature, precipitation, date and time of bearing, relative wind speed and weather there was precipitation.

There are constraints that need to be considered when taking bearings. It would be better to have 4 separate technicians located at the 4locations, all getting a bearing on a specific animal at the same time. However, this would be expensive. To save money, The UW has just one group of technicians getting all 4 bearings. This means after gathering the first bearing on the desired animal, they have to quickly drive to the next waypoint to get the next bearing. They then need to repeat the procedure at the third and fourth waypoint. The constraint is all 4 bearings must be taken within a 20 minute time frame to be valid data. Otherwise a deer could have moved too far and LOAS would not be able to estimate an accurate location. Another constraint is at least 6 hours must pass before another set of bearings is recorded on the same animal, otherwise the movement is considered insignificant. Finally, the more bearings taken, the error polygon will be smaller. The error polygon is the area where the bearings intersect. The polygon shape that forms from the bearing intersections is the area that the animal is likely in. The pinpointed location will be within this area. Therefore the smaller the error polygon, the more accurate the estimated location will be. If the error polygon consists of an area greater than 50,000 square meters, the location is not accepted. The error polygon needs to be less then 50,000 square meters to be considered an accurate location.

When using radio telemetry in hilly terrain, it is often difficult to tell exactly where the signal is coming from. This is due to a term called bounce. The signal bounces off a hill from behind the technicians and is re-recorded. Instead, the deer appears to be in the opposite direction. Sometimes beacons are used to test the telemetry accuracy. Beacons are simply radio collars that are not on animals. A beacon is placed in the hills and location tests are performed to see how much the bounce effects the accuracy [2].

Coordinate Systems

The global coordinate system is called Universal Transverse Mercator (UTM) System. This was developed for military purposes by the Dept. of Defense. The UTM divides the world into 60 different north-south longitudinal zones. These are then projected onto a transverse mercator projection. Zones 15 and 16 meet in Wisconsin. This poses a problem for the UW-Wisconsin’s research project. They have radio collared deer in both Iowa and DaneCounty. However the joining of zones 15 and 16 occurs in IowaCounty which divides it into 2 UTM zones. To account for this, the Wisconsin Department of Natural Resources developed the Wisconsin Transverse Mercator (WTM). Since Wisconsin is no longer divided into two separate UTM zones, software such as LOAS and ArcView and other GIS software are easier to use [2].

LOAS SOFTWARE

LOAS is only used for estimating locations. Information such as population numbers, ages, and activity cannot be attained. However, in order for LOAS to estimate locations, at least 2 bearings need to be taken. Other location software can be used, although they may not be able to incorporate graphical representations as LOAS can. Another advantage is that it works hand in hand with ArcView which will be discussed later.

The interface of LOAS is divided into two separate parts. One part is reserved for graphical representation of data, while the other part is for data. The data half is then divided into five additional parts. These parts include the data spreadsheet, results spreadsheet, error log, legend, and the polygon ID.

User Interface

Figure 2. LOAS User Interface

• Data Spreadsheet: Spreadsheet of the data to be used by LOAS. The user may

modify data by double clicking a cell tochange the

contents.

• Results Spreadsheet: This file is read only, however it can be copied into other

programs.

• Error Log: If LOAS encounters any kind of an error in processing, it

is listed here. Examples of this include bearing lines

that do not intersect or data of the wrong type.

• Legend: This will give alternative graphing options such as changing

the way lines and points are displayed.

• PolygonID: Data of the clicked ellipses appears here

LOAS can open database files and text files. If a text file is opened, LOAS will convert the file into a character based database. LOAS allows the user to link databases. That is as long as it is a one to one relationship. This means 1 parent file and 1 child file can be used together to analyze data. For example the parent file may contain bearing data while the child file may contain receiver locations, as long as the child file has fields in common with the parent file. Once the child field that is common with the parent file is clicked, the user can click the Link Button to merge the data.

Using Bearings to Calculate Locations

Once the bearing data (checkpoint coordinates, angle to signal, and signal strength, active/non-active) has been entered into LOAS, the user must select the desired estimator. There are seven estimators to choose from and each differs in accuracy and required data. Every circumstance is different and often the best selected estimator depends on many factors. Some of the factors depend on desired accuracy, number of bearings taken, and the quality of the bearings.

Estimator Selection

Figure 3. Choosing an estimator and delimiters

• Maximum Likelihood Estimator(MLE): This method finds the least angular error between the set of bearings and the location estimate of the signal. An iterative algorithm is used but it will only work if the data has no outliers. This means if a bearing is not within the majority of bearings, then the MLE should not be used. The user must select the desired accuracy and desired number of iterations to stop the recursion when these limits have been reached.

• Huber and Andrew’s Estimators: Also known as the M methods, these areoptimalwhen bearing outliers exist. Different weights are assigned to each bearing, depending on the estimated location. Less weight is given to outliers and more weight is given to the bearings closer to the majority of bearings. Similar to the Maximum Likelihood Estimator, recursive iterations are performed and limits of accuracy and number of iterations need to be specified by the user. However, to use these methods, a minimum of five different bearings are needed.

• Best Biangulation: This estimator should be chosen if only two bearings are available. This method picks the angle closest to 90 degrees after calculating all intra-bearing angles.

• Arithmetic Mean: When all other estimators fail, the arithmetic mean method is usually chosen. This will add all the coordinate points and calculate the average. However, if outliers exist, the results can be skewed.

• Geometric Mean: LOAS uses the logarithmic method to calculate the geometric mean which eliminates overflow errors. Although this method is more accurate then the arithmetic mean, it is still significantly sensitive to outliers.

• Harmonic Mean: Similar to arithmetic and geometric means, but much less sensitive to outliers.

LOAS GRAPHICS

Most radio telemetry software does not have the graphics capability that LOAS has. LOAS allows the user to put labels such as receiver, ellipses, and estimated animal locations on certain files. The font type and color of these labels can be adjusted. Specific point and line properties can also be modified by changing the styles and color.

When LOAS processes the original data file, other files are automatically created. For example, from the original data spreadsheet a file with an extension (.bearings) is made which is used to plot bearings lines graphics. Another file with an extension (.intersections) is created and used for plotting bearing intersections. Finally a (.ellipse) file and (.locations) file is created and used for plotting ellipse and location graphics. To graphically represent any of these files, the user simple checks the box next to the desired file.

Manipulating LOAS Data

All results are sent to the results spreadsheet. If the user wants to edit the results spreadsheet, the user needs to copy it into some Windows application, and then make changes to that file. LOAS will not allow editing of results.

LOAS allows the filtering of data. Each file may only have one filter in use at a time. First, a file is selected. Then a variable is chosen to base a comparison on. Next, an operator is chosen. The operators are the basic greater than, less than, greater than or equal to, less than or equal to, equal to, and not equal to. Filtering allows the user to limit the amount of records needed.

Sorting is another key capability of LOAS. The user need only specify the sorts in the order to be executed. LOAS is even able to sort linked databases however indexes need to be specified [4].

From LOAS toArcView

LOAS is able to pinpoint location within a few meters, as long as accurate bearings were taken. However, LOAS is just a package to determine location. Other software is useful in determining other kinds of relations after LOAS has determined the appropriate locations. The data locations (x, y coordinates) created by LOAS is used directly by another Geographic Information System (GIS)ArcView, to make special comparisons with other data. Most data to be analyzed is directly tied to space such as an address, GPS location, or a specific region. ArcView creates shape files or maps that enable users to visualize data in a way that is easy to see relationships and patterns that could not be seen with traditional analysis. ArcView comes with 8 CD-ROMS of national maps and the tools to create your own maps. ArcView uses databases such as the (x, y coordinates) of LOAS and compares it with other databases of other interests and combines them into a shape file for visual analysis. In addition, multimedia links to add sounds, pictures, and video may be incorporated [5].

ArcView has many extensions that can be added to enhance what can already be done. Some of these extensions include:

• ArcView Spatial Analyst
• ArcView 3D Analyst
• ArcView Image Analysis
• ArcView Tracking Analyst
• ArcView StreetMap
• ArcView Movement Analysis

The UW-Madison uses the Spatial Analyst and Movement Analysis extensions in their CWD Research Project. The Spatial Analyst allows users to analyze specific cell-based raster maps, and to infer new information based on current data, along with building sophisticated spatial models [6]. The Movement extension is used to analyze animal movement data.

Movement Extension

While using the Movement Extension, UW-Madison CWD Research Project determines a deer’s home range (area where a whitetail spends most of its time). Movement allows for several different views. Two views that are used by the Research Project are Kernel Home Range and Minimum Convex Polygon Home Range.

The Kernel Home Range calculates a home range and outputs three things. The first output is a grid based on utilization distribution. Next, a shapefile is created for each probability chosen. Lastly, calculations for each probability are shown in a message box. This home range is calculated by using a least squares cross validation method, an ad hoc smoothing parameter, or an inputted smoothing parameter. The Research Project uses 95% default probabilities. The advantage to using the Kernel Home Range is that a percentage of outlier points are not used. A more accurate and concentrated visual graphic is created. Only areas of heavy concentration are included.