THE BEAR LAKE WATER PROJECT
The G.I.S. applications were done by Becky Sorensen, Mike Hansen, & Shane Lowry.
Remote Sensing applications for this project were done by Jeff Parkinson, Mike Hansen, & Shane Lowry.
INTRODUCTION:
Our project will be focusing on the different water attributes associated with Bear Lake. Bear Lake is intersected by the Idaho-Utah border. It is located in the northeast corner of the state. At the north end of this natural lake is a pumping plant. This was established in 1909 by the Telluride Power Company to devert water for agricultural and irrigational purposes for the residents of the surrounding area. This pump can maintain a constant lake level by having the ability to pump spring runoff into the lake for future storage. During the drier summer months, water is pumped out of the lake as needed for irrigation. The average fluctuating level is usually around three feet.
The lake plays an important role to many people as a source of irrigation, flood control, hydro-power, and recreation. Using GIS and Remote Sensing we wanted to focus on Bear Lake by mapping and analyzing a few of the features of this unique lake.
Our project will be focused on using GIS & Remote Sensing to look at:
Identifying any water currents effecting sedimentaion in the lake
Determine water depth from satellite imagery
Make a digital elevation map of the lake bed
Make a three dimensional map of the bottom of the lake
This viewing is a 1993 TM image of Bear Lake and partial shore area. Default bands of 4, 3, 2 are shown. We clipped the water from a larger image using the "Seed" command from Imagine. We chose the "Seed" command because it gave us the freedom to clip out only the water area. In dealing with this irregular polygon, any other clipping methods would have been more difficult.
We obtained a 1979 contour map from Ecosystem Research Institute. This map was generated by a petroleum company that charted the depth of Bear Lake. We placed tics in four known locations of latitude and longitude. The actual map displayed contour intervals of 10 feet, however, due to the closeness of some contour lines we digitized every 50 feet. The map's large size made digitizing a two step process. The top half of the map was digitized first, the map was moved, and then the bottom half was digitized. This is why there are a total of 6 tics. Because of this large map, arcs were digitized instead of polygons. We then cleaned it and built topology creating one large coverage with polygons.
Next, we overlayed our two images. The new digitized image was placed on our previous remote sensed image. As you can see, the two shorelines differ slightly. These two images were developed 14 years apart by 2 different companies which contribute to this error. Considering this, we feel that the error is very minimal.
This is the above 1993 image of Bear Lake after running an unsupervised classification. Since specific classes were unknown, we chose the unsupervised classification over the classified. This image was processed to visually clarify differences in water depth, currents, and sedimentation. By changing the colors, it is easier to identify these unknown areas. The areas of interest are the red swirl at the top-left side and the yellow swirl in the middle. These two spots indicate distinct water currents. The swirl on the left is caused from an incoming stream. The cause of the yellow swirl is unknown at this time. These areas contain suspended particles that cause sedimentation in the surrounding areas.
Once sedimentation areas were identified, the correlation with water depth was analyzed. This incorporated the use of RS and GIS tools. Utilizing Imagine's capability, we overlayed the vector-based map of the lake contour lines onto our classified raster image. Patterns in spectral values showed similar patterns to that of the water depth leading to a moderate correlation. However, due to inlets causing high particulate suspension, some areas did not correspond because of the greater amounts of reflectance from these particles.
To idendify spectral signatures for each elevation polygon, we overlayed our polygon coverage onto our original TM image. We did an AOI for each polygon to create the individual coverages for each elevation. After creating these polygons we wanted to generate spectral signatures for each elevational coverage to determine if any differences between layers. Each elevational coverage is shown below with its associated spectral signature on the right.
This image is 0 - 50 feet.
This image is 51 - 100 feet.
This image is 101 -150 feet.
This image is 151 - 200 feet.
This image is over 200 feet.
In comparing the spectral signatures, we found little difference between the images except for the first image. It would appear that after 50 feet, the water tends to absorb the majority of the light, making it difficult in distinguishing elevational depths based on the mean spectral signatures.
In further study of sedimentation occuring in the lake, we used Imagine to produce a Feature Space Plot image. By looking at the two images simultaneously, we were able to move the cursor to any part of the lake and observe the reflectance values. If the lake had any sedimentation, plant material, or was shallow in water depth, the feature space plot cursor would be somewhere besides the botton left-hand corner.
The last goal in our project was to develop a 3-D image. We began by deleting the inner contour intervals from our previously made contour image. Next we digitized 600 points that the petroleum company obtained during its extensive surveying for gas and oil. Each of these points represents where the company drilled. The water depth of each location was added to the point-attribute table for the coverage. This information supplied the X,Y, and Z coordinates that is used to develop a Digital Elevation Model (DEM) of the lake.
The following is our DEM. It was produced by using the Arc grid command called IDW, Inverse Distance Weighted. From our 600 point coverage that contained known depths, this program used the distance from one point to another and estimated the depth between the points.
After creating our DEM, we uploaded it into Imagine. We wanted to produce some type of a 3D image. By using the "Spatial Profile" feature, the computer generated this image from our DEM.
After the DEM was produced, we used the "Spatial Profile" command in Imagine to look at a particular cross section of the lake bed. The first image is a cross section from west to east. The second is looking at it cut from south to north.
Here is another type of 3D image that is possible from Imagine. The red dot on the left is where we are standing taking the picture. The green dot is where we are looking. The mouse allows us to place these two points anywhere on the image we want. Once the dots are placed, the snapshot is developed on the screen on the right side.
F.Y.I. by using special SPOT satellite images, we were also able to zoom in on this unique feature locate near the east side of the lake. Remoste Sensing may be the final link in solving the mystery of the Bear Lake Monster!
CONCLUSIONS:
In starting our project, we could not locate a DEM for the lake bed. So, we produced a contour map for the bed using ARC/INFO. We transformed the coordinates into real world coordinates (latitude and longitude) such that a user could identify his or her location.
We also created an image which seperated pixels into ten classes based on the spectral values (unsupervised classification). This image allowed us to identify major water currents within the lake that were not clearly visible in our unclassified image. This classification scheme showed a moderate degree of correlation with the contour map that we had created. However, sole classification of water depth based on the unsupervised classification could produce a high error due to variable levels of particles suspended within similar areas of the lake. This would tend to support out findings where we compared mean spectral signatures of the different depth classes. The spectral signature for the depth class of 0-50ft. was very different from the other four depth classes. These other four classes showed little variation in their spectral signatures. This may indicate that after a certain water depth the waves of the electro-magnetic spectrum are absorbed and reflected similarly. Using spectral signatures for more shallow water bodies (< 100 feet) may result in greater accuracy in identifying water depth.
This information could be very useful when compared with other years of imagery. Sedimentation patterns and depth near the pumping station could be monitored. One could determine if or when dredging may need to occur to maintain the effectiveness of the pumping station as well as maintaining the volume of the lake.
Along with the contour map, unsupervised classifications, spectral signature images, and DEM, we were able to create surface and spatial profiles and a perspective image from our DEM data. These can be generated in Imagine for any specified location. This would be useful to anyone interested in Bear Lake. The surface profile shows the surface of the lake bed indicating the elevation and length for any given cross-section. The spatial profile creates a 3-dimensional image that allows one to see clearly the depth dimension of the lake. The perspective image also provides a 3-dimensional image, only the user specfies the view point and the area to be seen.
All of this information can be useful to managers of Bear Lake and those who are and might be studying this unique water body. The information could serve as an educational tool to students as well as the general public.
FUTURE RESULTS:
This project provided us insight to many capabilities and options in GIS and Remote Sensing that we were unfamiliar with at the beginning of the quarter. As we proceeded in our project, we thought of other ideas or further research that could be done with Bear Lake or other water bodies. These could include:
MAKE COMPARISONS BETWEEN PREVIOUS YEARS IMAGERY TO IDENTIFY ANY CHANGES IN WATER CURRENT PATTERNS AND SEDIMENTATION.
STATISTICAL ANALYZATION BETWEEN SEDIMENTATION PATTERNS AND SPECTRAL SIGNATURES OF WATER DEPTH.
OVERLAY A MINERAL COVERAGE OF THE WATERSHED TO DETERMINE SOURCES AND AMOUNTS OF ADDITIONAL MINERALS EFFECTING WATER QUALITY.
USING MULTIPLE YEARS OF IMAGERY, OBSERVE SHORE EROSION AND LAKE VOLUME DUE TO SEASONAL FLUCTUATIONS.
A special thanks goes to the many people who diligently volunteered their help in working with us one-on-one to produce this project.
Back to the Geography Home Page!