Geog 483 Final Project
Sachin Chand
This project presents us with the challenge of performing site selection analysis suitable for a biological reserve system in Centre County, PA. For the first portion of the project I performed my analysis on vector data. To create the final results, I converted the vector data to raster to perform raster analysis. The first criteria calls for a selection of study areas that have a combined bird and mammal species count of greater than 70. In order to satisfy this requirement, I performed a join between the studyareas layer attribute table and the speciesrich table on the BLOCK_ID field. I then created a new field in the joined table that calculated the sum of the birds and mammals fields. I queried this new field for values greater than 70 and exported the results to a new shapefile. I was able to perform this analysis without creating any new, intermediate layers. The results left me with 44 study areas with a bird and mammal species count greater than 70.
Click on images for larger view.
Figure 1. Attributes of the 44 study areas selected in the first analysis step. The screenshot shows the new "Total" field that was calculated as a sum of the mammal and bird fields. The total field is sorted in ascending order and only values greater than 70 were selected.
An additional requirement on the study areas is that less than 10% be occupied by buffered roads, highways and interstates. For this, I added a new distance field to the roads layer. I populated this field with the specified distances by road type (roads = 20 meters, highways = 50 meters and interstates = 100 meters). I then performed a buffer on the roads layer based on the distance field I had just added. At this point, I took the shapefile of study areas with a combined population of mammals and birds greater than 70 (created in the first step) and used the erase function to erase study areas occupied by the buffered roads. This created a new layer with gaps/empty spaces where the roads would be. I calculated a new area field for this new layer. The original layer already had an area field so I now knew what the original area was and what the new area (after removing road coverage) was. Once again, I created a new field and calculated its value to be the new area/old area * 100. This gave me a percentage of what the new coverage area was in relation to its original area. The criteria called for study areas less than 10%, so I queried the percetage field for values greater than 90%. The results were exported to a new shapefile. This shapefile satisfied having a bird and mammal total greater than 70 as well as less than 10% being covered by roads.
Figure 2. The green study areas are the original 157. The pink study areas show ones that have a bird and mammalian count greater than 70 and have less than 10% road coverage. The pink study areas also show where road areas, based on road type, were erased to show road coverage.
Figure 3. A close-up screenshot of study areas and the road network buffer. The illustration shows the results of the erase function on the study areas. You can see the white areas that have been erased/removed from the study areas. The amount of area removed is a result of the road type. Interstates show the highest area removed, followed by highways and then roads.
The next two criteria called for land that is publically owned and has a high habitat potential. For these criteria, I queried the habitat layer on the "habitatpot" field for "high". The results were exported to a new shapefile. A similar operation was performed on the ownership layer where I queried "public" land and exported the results to a new shapefile.
Figure 4. Screenshot of publically owned land in Centre County. This screenshot is a result of the query performed on the ownership layer where publically owned lands were queried and exported to a new shapefile. The blue represents publically owned land and the red outline represents the Centre County political boundary.
At this point, I was done with my vector analysis and ready to work with raster data. Before converting the newly created, "trimmed down" shapefiles to rasters, I added a new field with a value of 1 to their attribute tables. The reason for adding this new field was so that I could classify the data on this field when converting to raster. This would save me the step of reclassifying the data later because all the areas that satisfied the criteria would be classified as a 1. I then performed a reclassify operation on the provided landuse raster data. Forested areas were set to a value of 1 and all other values were set to 0.
Figure 5. A screenshot of forested landuse in Centre County, PA. The screenshot was created using raster landuse data that was reclassified to have all landuses except forests equal 0. The green shows the forest landuse and all other areas have been set to null in display properties.
The criteria called for the selection to have a slope of less than 10%. For this, I used the elevation raster to calculate slope percentage. I then reclassified the raster output to have slope percent 0% - 9.99% = 1 and 10% - 100% = 0. The results of all analysis performed thus far left me with four raster datasets. Using the raster calculator, I multiplied each of these datasets. The results show that grids with a value of 1 across the board were classified as a 1 while grids where a 0 was in the cog or where all grids didn't intersect were classified as a zero. The 1s' were the areas that satisfied all our criteria.
Figure 6. Potential biological reserve sites in Centre County, PA. The potential sites are represented in red and aggregate to 25 square miles. These areas have a combined mammalian and bird species count greater than 70, less than 10% of the study areas are occupied by a road network, have a high habitat potential, are publically owned, are classified as a forest landuse and have a slope of less than 10%.
Figure 7. Screenshot of the attribute table of the final raster calculation. The 25859 grids classified as a 1 are the areas appropriate, per the criteria, for a biological reserve system. Each cell has a size of 50 which means its dimensions are 50x50. The units used for this project were meters.
Figure 8. Inset map of Centre County, PA with potential biological reserve sites. The inset map serves the audience to get a perspective of the size of Centre County and its location relative to the state of Pennsylvania. With the perspective of the inset map, we can make a good guess on why Centre County got its name.
Figure 9. Screenshot of the potential biological reserve sites displayed in the Google Earth Program. In order to display the data in Google Earth, I first converted my final results from raster to vector data using spatial analyst tools. I then downloaded, installed and executed the "Export to KML" script available at http://arcscripts.esri.com/details.asp?dbid=14273 to convert the vector data to a kml format. KML stands for Keyhole Markup Language and is an XML-based language for defining the display of three-dimensional spatial data in Google Earth (ESRI ArcGIS 9.2 Help, 2006). In addition to the screenshot, I have made the kml file available for download in the link below. If the client has google earth installed, executing this file will open up the potential biological reserve sites in google earth.