RS 576 PROJECT
The Logan area is developing very quickly along the eastern bench. We as a group decided to study this area in more intense detail, while using the background experience obtained from Remote Sensing. With this knowlegde we were able to set up a reasonable project that dealt with our selected research area. The study is composed of comparing land development by two different methods of analysis. The first analysis method uses aerial photos from two seperate years in Cache County. The second analysis uses SPOT data retrieved through the computer program called IMAGINE.
The research project that we chose provides valuable information for developers and citizens of Cache County and in particular the City of Logan. The developers may use the research information to assist in future development programs. Citizens can study the impact of development on their community and evaluate how they wish to stand on development issues in the future. A research project such as this may be used by others who are looking at similar problems that arise in their areas.
The research problem deals with evaluating land development over a span of 18 years in Logan, Utah. Our research will be accomplished by categorizing land development from different aerial photos, one produced in 1976 and the other in 1994. These photos will then be compared to classified imagery of the same area in Cache Valley. We wish to compare developed land percentages calculated from the computer with the data collected from the aerial photos.
We propose that after the research has been completed there will be strong evidence that development has increased in this area of Logan. The percent of increase highly depends upon the style of the study, and the tools chosen to complete the project. We chose two different methods to complete this study, and in the end we compared the results of the two methods.
We created a standard by which we classified each of the maps (aerial photos) into developed and undeveloped land. Our definition of developed land includes all land that now contains a building, road, sidewalk, or land that has reach its highest developed stage. Undeveloped land includes agricultural land and land that has not yet reached its highest level of develpoment. When the aerial photos were analyzed, a grid system was created to help in the categorization process. If more than fifty percent of a single grid square contained developed land, then that square became developed. However, when the computer categorized the images it used spectral signatures based on individual pixel values.
The research was contained within the Logan, Utah area. The northern boundary fell upon 17th North in Logan. The southern extent of the study reached to approximately 22nd South. The west boundary was designated by 6th East in Logan and the east boundry reached to the base of the mountains.
The following is an image of our area of study and analysis.
The two different research methods used were compared to each other for the end result. The first collection type was manually done, the second type was completed by the computer.
First we analyzed the aerial photos using the manual method. The two aerial photos were obtained from the City of Logan, Planning and Zoning Office. Two grid systems were produced on transparencies to help in the categorization process. The first grid contained one and a half inch squares, the other, three-quarter inch squares. The boundaries on the photos were then determined and the categorization process began. Grids, similar to the ones on the transparencies, were constructed on paper to record the data. Square by square the land was categorized as developed or undeveloped. If more than fifty percent of the land in the square contained development, then that was the label that particular square received. The different labels became the data that was recorded by the three of us manually. After both aerial photos had been evaluated, and the data recorded, tallies of the data were made. All of the developed squares were totalled for each photo year and then a percent of change for each of the two years was calculated. This percentage became the data for the first type of collection.
The data collected from the aerial photos include statistics gathered from the 1 1/2 inch squares, as well as the 3/4 inch squares overlaid onto the photo. From the 1 1/2 inch classification system on the 1976 photo there were 24 developed squares out of 72, this calculates out to be 33%. The 3/4 inch grid system for the 1976 photo held 91 developed squares out of a total 288. The pecentage total for the 3/4 inch grid calculates to 32%. The two grid systems for 1976 were then averaged for an overall percentage of 32.5%. The 1994 ariel photo was classified using the same type of grid systems. The 1 1/2 inch grid produced 36 developed squares out of 72, the percentage coming to 50%. The 3/4 inch grid captured development in 130 squares, this is 45%. The average between the two grid systems for 1994 comes to 47.5%. Between these two sets of aerial photos it was determined that there was an increase of 15% in developed area. The following graph is a summary of the above data using the aerial photographs and our grid system.
The data collected in the lab came from a 1994 SPOT image of Logan. By using this type of imagery we were able to not only manipulate the image, but we found that it was a far more interesting way to come up with data. Our first step with this image was to become familiar with the differences that it had as compared to the maps (photos) that we had been working with previously. Following this we selected the same plot of land that we had on the maps (photos) and corresponded it to our image. It was necessary for us to cut down the large Logan image to just include the East bench that we were working with in order to assure that our statistics would be as accurate as possible. Upon completion of this we took on the project of classifying the selected portion of the image via pixel values. We did this by running the image through the classifier and image interpreter which gave us the texture values for a given pixel. From here we took a lot of time in coming up with a suitable and realistic method to classify pixel values as developed or undeveloped.
In building our developed/undeveloped model we had to recode our original image so that we only had three pixel values to deal with. These values represented developed land, undeveloped land, and vegetation.
In the end we had three coverages that were relatively the same. The first one (Image One) has developed and undeveloped land only. The vegetation values were incorporated with the undeveloped land. The second image (Image Two) incorporates the vegetation with the developed land. We felt that this is a more accurate depiction of the true percentage of developed land in our study area. The third image (Image Three) simply shows the vegetation as green. We have still included the vegetation as part of the developed percentage of land, however we did want to make the distinction between structures and vegetation. The three images follow below.
Image One Image Two Image Three
After texturing and recoding the above images we were able to gain a percentage for developed land and undeveloped land for each of the three images. The following graph shows the percentages for each image.
In Image One the vegetation is being added to the undeveloped land, hence the skewed percentages. Image Two and Image Three are the same thing except that in Image Two the vegetation is red (developed land) and in Image Three the vegetation is green, but it is still being considered developed land.
At this point of our project we were ready to compare the 1994 percentages that we had calculated by the use of our grid system and the percentages that Imagine gave us from our textured and recoded image. The following graph shows those percentages.
When comparing the image and the photo from 1994 there is only a 1.5% difference in percentages between developed and undeveloped land.
Throughout the course of the study we have learned that the development of the Eastern Bench of Cache Valley has indeed increased over the course of eighteen years. By using the aerial photos and the computer generated SPOT imagery we conclude that the two research methods used, produced relatively the same results. In the future, this particular study will help in the understanding of similar research projects. Our group has developed a better understanding of the applications and methods used by modern day Geographers by participating in this project related course.