In order to complete this project, two different types of imagery must be obtained and analyzed. Satellite images will be used to represent landscape coverage from the 1970s through the present. Aerial photography will be utilized to represent in the UCLA campus in years predating the development of reliable and accurate satellite imagery, covering years 1920-1970. For each of these two types of images, slightly different methods will be utilized.
Aerial PhotographyAerial photographs for this project were obtained from the UCLA Benjamin and Gladys Thomas Air Photo Collection. This collection includes photographs taken throughout the 20th Century, primarily of the Los Angeles County area. The two photographers who have contributed to this collection are Fairchild and Spence. Photographs taken of UCLA from the date of the establishment of the UCLA campus through present were selected and withdrawn from the collection. Images were then more selectively analyzed based on height of oblique, distance from the UCLA campus, and angle. Images were ideally chosen to represent intervals of approximately 7-10 years. However, availability of photographs containing similar angles and direction limited these specifications, and some larger time intervals and some shorter time intervals were used out of necessity based on the material provided. After this selection process, seven photographs were chosen as the optimum representation of the change in landscape from the years 1929 to 1968.
Satellite ImagerySatellite imagery is the most ideal for extensive analysis of greenness and change detection because the image's electromagentic bands can be displayed and manipulated in the remote sensing software ENVI. The process by which the satellite images were displayed and analyzed is outlined below in five extensive steps. First, six satellite images ranging from 1979 to 2005 were downloaded and opened in ENVI. Next, an NDVI was created for each image and using the NDVI, a Density Slice was overlaid on the each image, creating a Greenness Index. The images were then compared in a Change Detection analysis and finally each image was edited and was given proper spatial information in the form of an overlaid vector of UCLA.
Each of the six satellite images were obtained from Maryland Land Cover Facility and downloaded in 4 bands; Blue, Green, Red, and Infrared (bands 1,2,3,4 respectively). Additionally, all six images were from the LANDSAT series of satellites, either LANDSAT 1-3, 4-5, or 7 depending on the year. There were two important criteria considered when choosing which images to download. First, the images all needed to be from roughly the same time of year in order to show accurate greenness changes from year to year. Second, each image had to be produced from the same source and have the same attributes in order to perform the change detection analysis. The first four images downloaded were produced by the USGS and had attributed listed as Ortho, GLS. Theses images included an MSS image from June 8, 1979, a TM image from June 28 1989, an ETM+ image from May 1, 2000 and an ETM+ image from May 15, 2005. The final two images that were downloaded were done so that they could be analyzed using change detection. The only image available from the 1990s was a 1996 USGS image with attribute of L1T, which could not be analyzed with attributes of the previous images. In order to perform a change detection of this time period, an additional image from 1989 was downloaded with attribute of L1T.
Once all the images' bands were downloaded, they were displayed in ENVI. The proper algorithm for displaying images in the visible is to display bands 3, 2, and 1 for RGB respectively. After it is opened in ENVI, a subset of the image including the study area was saved as well as a zoomed image that highlighted UCLA. Since all the images have a relatively large pixel size compared to the area of UCLA, the zoomed image is rather pixelated. Images from Google Earth were collected from roughly the same years in order to better demonstrate the visible landscape of UCLA (obtaining images from exactly the same year was limited by Google Earth's historic image availability).
While Greenness and change can be detected by closely examining the visible images from year to year, ENVI enables the user to actually find greenness values via the NDVI display. NDVIs (Normalized Difference Vegetation Index) use the red and infrared bands of an image and return an image with pixel values ranging from -1 to 1, showing the amount of greenness in a certain area. In order to create an NDVI in ENVI, the image was displayed and saved in the Infrared, using bands 4,3,4 for RGB. Next, using the NDVI tool located under Transform in the power toolbar, the infrared image was converted to the NDVI, producing a grayscale image that was available for analysis of greenness and change detection.
Although the NDVI gave a picture of greenness in the study area, a density slice gives a more user-friendly image that is easier to understand and interpret. A density slice was created for each of the image and was displayed in five different classes ranging from -1 to 1. The lowest class was between 1 and 0.1110, followed by 0.1110 to 0.2816, 0.2816 to 0.4523, 0.4523 to 0.6229 and 0.6229 to 1. Classifications with lower greenness values were assigned colors associated with a lack of vegetation, such as dark brown and tan. The medium greenness value was given a yellow color and the two highest greenness values were assigned a light and dark green color. Each of these images were displayed in the same manner as the visible images, with a larger image of greater Los Angeles and a zoom image of the UCLA study area.
The final step in the process of analyzing the greenness levels and change of greenness of UCLA's campus from the 1970s to 2005 was to create a change detection image, an image that displays the difference between two NDVIs in which a blue color represents a decrease in greenness and red represents an increase in greenness. The ability to compare images with the change detection tool in ENVI was based the attributes and producers of the images downloaded in step one. Change detection images were produced for 1979 to 1989, 1989 to 1996, 1989 to 2000 and 1989 to 2005 and displayed in the same manner as the visible and greenness indices from the previous steps.
Satellite imagery is the most ideal for extensive analysis of greenness and change detection because the image's electromagentic bands can be displayed and manipulated in the remote sensing software ENVI. The process by which the satellite images were displayed and analyzed is outlined below in five extensive steps. First, six satellite images ranging from 1979 to 2005 were downloaded and opened in ENVI. Next, an NDVI was created for each image and using the NDVI, a Density Slice was overlaid on the each image, creating a Greenness Index. The images were then compared in a Change Detection analysis and finally each image was edited and was given proper spatial information in the form of an overlaid vector of UCLA.
Each of the six satellite images were obtained from Maryland Land Cover Facility and downloaded in 4 bands; Blue, Green, Red, and Infrared (bands 1,2,3,4 respectively). Additionally, all six images were from the LANDSAT series of satellites, either LANDSAT 1-3, 4-5, or 7 depending on the year. There were two important criteria considered when choosing which images to download. First, the images all needed to be from roughly the same time of year in order to show accurate greenness changes from year to year. Second, each image had to be produced from the same source and have the same attributes in order to perform the change detection analysis. The first four images downloaded were produced by the USGS and had attributed listed as Ortho, GLS. Theses images included an MSS image from June 8, 1979, a TM image from June 28 1989, an ETM+ image from May 1, 2000 and an ETM+ image from May 15, 2005. The final two images that were downloaded were done so that they could be analyzed using change detection. The only image available from the 1990s was a 1996 USGS image with attribute of L1T, which could not be analyzed with attributes of the previous images. In order to perform a change detection of this time period, an additional image from 1989 was downloaded with attribute of L1T.
Once all the images' bands were downloaded, they were displayed in ENVI. The proper algorithm for displaying images in the visible is to display bands 3, 2, and 1 for RGB respectively. After it is opened in ENVI, a subset of the image including the study area was saved as well as a zoomed image that highlighted UCLA. Since all the images have a relatively large pixel size compared to the area of UCLA, the zoomed image is rather pixelated. Images from Google Earth were collected from roughly the same years in order to better demonstrate the visible landscape of UCLA (obtaining images from exactly the same year was limited by Google Earth's historic image availability).
While Greenness and change can be detected by closely examining the visible images from year to year, ENVI enables the user to actually find greenness values via the NDVI display. NDVIs (Normalized Difference Vegetation Index) use the red and infrared bands of an image and return an image with pixel values ranging from -1 to 1, showing the amount of greenness in a certain area. In order to create an NDVI in ENVI, the image was displayed and saved in the Infrared, using bands 4,3,4 for RGB. Next, using the NDVI tool located under Transform in the power toolbar, the infrared image was converted to the NDVI, producing a grayscale image that was available for analysis of greenness and change detection.
Although the NDVI gave a picture of greenness in the study area, a density slice gives a more user-friendly image that is easier to understand and interpret. A density slice was created for each of the image and was displayed in five different classes ranging from -1 to 1. The lowest class was between 1 and 0.1110, followed by 0.1110 to 0.2816, 0.2816 to 0.4523, 0.4523 to 0.6229 and 0.6229 to 1. Classifications with lower greenness values were assigned colors associated with a lack of vegetation, such as dark brown and tan. The medium greenness value was given a yellow color and the two highest greenness values were assigned a light and dark green color. Each of these images were displayed in the same manner as the visible images, with a larger image of greater Los Angeles and a zoom image of the UCLA study area.
The final step in the process of analyzing the greenness levels and change of greenness of UCLA's campus from the 1970s to 2005 was to create a change detection image, an image that displays the difference between two NDVIs in which a blue color represents a decrease in greenness and red represents an increase in greenness. The ability to compare images with the change detection tool in ENVI was based the attributes and producers of the images downloaded in step one. Change detection images were produced for 1979 to 1989, 1989 to 1996, 1989 to 2000 and 1989 to 2005 and displayed in the same manner as the visible and greenness indices from the previous steps.
While Greenness and change can be detected by closely examining the visible images from year to year, ENVI enables the user to actually find greenness values via the NDVI display. NDVIs (Normalized Difference Vegetation Index) use the red and infrared bands of an image and return an image with pixel values ranging from -1 to 1, showing the amount of greenness in a certain area. In order to create an NDVI in ENVI, the image was displayed and saved in the Infrared, using bands 4,3,4 for RGB. Next, using the NDVI tool located under Transform in the power toolbar, the infrared image was converted to the NDVI, producing a grayscale image that was available for analysis of greenness and change detection.
Although the NDVI gave a picture of greenness in the study area, a density slice gives a more user-friendly image that is easier to understand and interpret. A density slice was created for each of the image and was displayed in five different classes ranging from -1 to 1. The lowest class was between 1 and 0.1110, followed by 0.1110 to 0.2816, 0.2816 to 0.4523, 0.4523 to 0.6229 and 0.6229 to 1. Classifications with lower greenness values were assigned colors associated with a lack of vegetation, such as dark brown and tan. The medium greenness value was given a yellow color and the two highest greenness values were assigned a light and dark green color. Each of these images were displayed in the same manner as the visible images, with a larger image of greater Los Angeles and a zoom image of the UCLA study area.
The final step in the process of analyzing the greenness levels and change of greenness of UCLA's campus from the 1970s to 2005 was to create a change detection image, an image that displays the difference between two NDVIs in which a blue color represents a decrease in greenness and red represents an increase in greenness. The ability to compare images with the change detection tool in ENVI was based the attributes and producers of the images downloaded in step one. Change detection images were produced for 1979 to 1989, 1989 to 1996, 1989 to 2000 and 1989 to 2005 and displayed in the same manner as the visible and greenness indices from the previous steps.
