Several institutions have joined effords to implement what is called the GREAT BASIN INTERDISCIPLINARY RESEARCH AND MANAGEMENT PROJECT: Maintaining and Restoring Riparian Ecosystem Integrity. This is a five year project, being implemented in Toiyabe National Forest, close to Austin, Nevada. The project is currently in its second year.

The general goal of the above mentioned project is to archieve a better understanding of the of the structure and functioning of riparian ecosystems and watersheds within the Great Basin and to develop management guidelines for maintaining or restoring riparian ecosystem integrity.

The project comprises a series of components, including videografy of the riparian corridors and valley bottoms, and a GIS of the riparian corridor derived from topographic maps and video images.

Utah State University (USU) is one of the institutions involved in the implementation of the project. As such, the Remote Sensing Services Lab at USU has been directly responsible for the development of the two components mentioned above.

The area of influence of the project is drained by five creeks for which management strategies are being developed. These include: Big Creek, Washington Creek, Kingston Creek, Cottonwood Creek, and San Juan Creek, all of them being tributaries of the Reese River. To develop management strategies for these creeks and their riparian zones, data in a variety of formats, including vidiography, have been collected.

Our class project has two basic objectives:

1. To organize the different layers of information available for Washington Creek using the ARC/VIEW environment.
2. To characterize the distribution of the different vegetation species in the riparian zone of washington Creek.

In order to satisfy the the first objective mentioned above, the following steps were followed:

1. Gathering different layers of available information. These information consist mainly of several raster images (detailled below).
2. Creation of a vector coverage representing Washington Creek, Cottonwood Creek, and part of San Juan Creek, using ARC/INFO.
3. Rectification of the different layers to the same map base and projection so that they could be overlaied.
4. organization of the different layers in a ARC/VIEW project.

The second objective was satisfied as follows:

1. Video imagery covering the area of influence of Wahington creek was obtained. This include images from five different flight lines which cover the whole area.
2. The video images were then classified using a supervised classification scheme. The ground truth data for the classification was provided to us by the forest service.
3. Once the video images were classified, an area of interest including only the riparian zone was visually created using ERDAS/IMAGINE.
4. The area of interest was used to extract statistics about the vegetation distribution in the riparian zone.

More detailled information about the methodology is presented in the progress reports shown below.

The data required for the project was obtained and organized. After carefully checking the available data it was found that the digital elevation model for Kingston creek was missing. Therefore, it was decided to work with Washington Creek instead, since all the necessary data is available for that creek. The available data for the Washington Creek was transformed to the appropriate LAN format and stored on disk. This includes:

An ARC/VIEW project was then created, including all the available layers.

Since the last report, we have been working on creating a coverage for the creeks, adding the digitized creek coverage to the other layers using the ARC/VIEW environment, and getting started in the classification of the video images.

Creating a coverage

We created a new coverage which includes Washington Creek, Cottonwood Creek and part of San Juan Creek. This was done by digitizing this creeks using a digitizing tablet. After digitizing the coverage, it was cleaned to eliminate dangling and fuzzy nodes problems and to create topology. Then, the coverage was transformed to a UTM projection so that it could be overlaid with the video images and the ortho maps. All these coverages have the same UTM projection. The video images has got the UTM projection by rectifying it to the ortho maps.

Overlaying the coverages

After successfully digitizing the creeks, this coverage was added to the rest of the layers using the ARC/VIEW environment. At this point it was observed that the digitized creek coverage and the rectified video images did not match well. This discrepancy seems to be due to a slight skewness in the ortho maps, to which the videos images were rectified, when it was originally scanned. Therefore, the video images were re-rectified using the digitized creek coverage.

Classification of Video Images

We got started in the process of classifying the video images by first learning the procedure of supervised classification using ERDAS/IMAGINE. For this procedure, ground truth information is needed, which we have already obtained. Therefore, after learning the procedure and obtaining the ground truth data, we started classifying the images.

Classification of Video Images

Since the last report, the video images covering Washington Creek have been classified. The classification process of the video imagery was conducted according to the following procedure:

1. Based upon ground-truth information, obtained for a particular place, a signature file was created.
2. Using that signature file, the Supervised Classification procedure was run applying the Unclassified Rule the first time around. This procedure devided the image into different classes according to their signatures. Some areas which signatures were not accurately represented in the signature file were left unclassified, usually appearing as black spots on the image.
3. The Unclassified portions of the video image were then re-trainned in order to include their signature in the signature file.
4. Then, the classification procedure was repeated, but this time however the Parametric Rule was selected instead of the Unclassified Rule.
5. After this step, there were still some areas which were assigned to the wrong class due to the closeness of the signatures of adjacent classes. Therefore, for instance, SHADOW may have been classified as WATER. To solve this problem, based on the ground-truth information, a manual Recoding of the image was implemented to make sure that each area of the image was assigned to the appropriate class.

The classified image for the entire area is shown below:

This image was then visually cut along the Riparian Zone. The Riparian Zone is Shown Below:

To satisfy our first objective, an ARC/VIEW project was create which included several converages as detailled previously. The figure below shows a satellite image covering the entire watershed. In this image, the other coverages are too small to be seen at that scale.

The other coverages displayed at the proper scale are overlaid below:

To comply with the second objective, video imagery was classified. After classifying the video images, statistics were extracted for both, the entire area and for the riparian zone. the coverage distribution for the entire area of the creek is presented in the figure below.

The coverage distribution for the Riparian Zone is shown below.

A comparison of the vegetation distribution of the riparian zone with respect to the entire area of the creek is shown in the following figure:

From the results of the classification analysis, the following is concluded:

1. The entire area covered by the video images was classified into twenty different classes. Vegetation species comprise fifteen of the total number of classes.
2. As shown above, the dominant species for the entire area are Sagebrush, which cover 24.57% of the area, and Mountain Mohogany, covering 12.36%.
3. For the riparian zone, the dominan specie is still Sagebrush, covering 14.5% of the area, and several Willow Species, accounting for 13.7%.
4. Other species which exist mostly in the riparian zone are the Quaking Aspen, Ryegrass, Cheatgrass, Kentucky Blue Grass, Wood Rose, Wildrye, and Smoth Brome.
5. Finally, it is concluded that classified video imagery, along with other GIS components, are a valuable tool for enhancing the management of riparian zones to preserve ecosystem integrity.