Title

Salmon and the City of Sacramento Levees Analysis of Salmon Habitat in Sacramento River Urban Region
 

Author

Candy Williams
American River College, Geography 350: Data Acquisition in GIS; Fall 2013

 

Abstract

California salmon travel up the Sacramento River to Folsom for spawning. There are certain types of land cover (i.e. riparian overbank habitat) that they need for making this trip safely. By placing Flood Control levees close to the river banks, almost all riparian overbank habitat has been lost. What little is left is threatened by the official US Army Corps of Engineers policy requiring removal of all vegetation from levees. In many areas, this would require removal of all vegetation between the riverbank and the top of the levee. This policy is in direct opposition of the US Department of Fish and Wildlife and the National Marine Fisheries Service, both of which are trying to restore the riparian habitat along the river. Classifying areas of poor habitat is necessary to help direct restoration efforts and this process is discussed in this report.
 

Introduction

This study will assess the presence of riparian habitat along the Sacramento River through the Sacramento urban area from Freeport to the American River. The goal will be to map healthy riparian habitat and areas with little or no vegetation to identify areas for further study for habitat restoration. An unsupervised classification and then a supervised classification in ArcGIS will be used to determine the usefulness of the two classifications. Areas of poor quality riparian habitat were identified.

 

Background

The Sacramento River is the primary drainage pathway for the Sacramento Valley and drains both the Sierra Nevada and the Coast Ranges. The river is also a vital avenue for spawning Chinook salmon and provides vital habitat for wildlife. Levees were constructed along the river by the US Army Corps of Engineers in the early part of the 20th century to channel flood waters through the valley and lower the chance of flooding. In an effort to use flood waters to remove large amounts of sediment deposited by hydraulic mining performed for gold mining during the gold rush, these levees were purposefully placed close to the river banks. This causes an elevated water surface during floods, which results in increased river velocity and erosion of the river bottom. However, by placing the levees close to the river banks, almost all riparian overbank habitat has been lost. What little is left is threatened by the official corps policy requiring removal of all vegetation from levees and any area within 15 feet of the waterside or landside levee toe. Habitat restoration efforts along the river are underway to replace what has been lost and to return the Sacramento River to a beneficial place for salmon.

Like all species of Pacific salmon, Chinook salmon are anadromous. They hatch in fresh water, rear in main-channel river areas for one year. The following spring, Chinook salmon turn into smolt and migrate to the salt water estuary. They then spend anywhere from 1-5 years feeding in the ocean and return to spawn in fresh water. All Chinook salmon die after spawning. Chinook salmon may become sexually mature from their second through seventh year, and as a result, fish in any spawning run may vary greatly in size. For example, a mature 3-year-old will probably weigh less than 4 pounds, while a mature 7-year-old may exceed 50 pounds. Females tend to be older than males at maturity. In many spawning runs, males outnumber females in all but the 6- and 7-year age groups.
Fresh water streams and estuaries provide important habitat for spawning Chinook, and they also serve as nursery grounds for developing eggs, fry, and juveniles. Chinook salmon often make extensive freshwater spawning migrations to reach their home streams on some of the larger river systems. Chinook salmon do not feed during the freshwater spawning migration, so their condition deteriorates gradually during the spawning run as they use stored body materials for energy .

Once numbering perhaps 300,000 to 400,000 in the Sacramento River drainage, they began their upstream migration when the snowmelt waters rose and swelled the rivers in spring. They held through the summer in deep pools in foothill streams above 1,500 feet in the Sacramento Valley, and travelled far up into the high elevation streams to spawn. As late as 1945, more than 50,000 spring-run salmon went up the rivers, and they numbered perhaps in the hundreds of thousands before 1850. After 1920, low river flows due to canal diversions, which created warm waters in the rivers, became detrimental to the salmon. In addition, water put back into the river from irrigation contained residues of selenium salts and chemicals, which were detrimental to riparian habitat.
Cover is crucial for the survival of both adult and juvenile salmon. Fish that migrated in summer needed cold pools for holding, often with bedrock ledges or curtains of bubbles where water splashes from falls. Juveniles need slow-moving backwater and edgewater areas, often with branches or vegetation in the water for cover. An alder or Doug fir falling into the stream caused the flow to scour a pool around it, creating good habitat. So important is this type of cover that fishery biologists will carry out “Large Woody Debris surveys” in salmon streams. In restoration projects, elaborate methods have been developed to bolt and wire up logs and boulders to make “LWD” structures that will increase cover and pool habitat. “Small Woody Debris” -- branches and twigs -- is also important in creating cover for small juveniles and the aquatic insects they feed on. Recent river bank repair sites on the Sacramento River often use woody debris.

They also needed hiding places from the prowling predatory Sacramento pike minnows. Root wads and boulder also clusters provided good cover, the latter helping to stabilize gravel piles that were slowly moved downstream by current.
The riparian areas along the Sacramento River once totaled more than 500,000 acres; today only about 10,000 acres (5%) remains. Much of it consists of restored stretches, and there is also a significant amount of artificial wetland in the watershed. River control has prevented the Sacramento River from its natural flooding, braiding and course-changing patterns, which are important for the maintenance of existing wetlands and the creation of new ones. Since the 1860s, the river has been mostly locked in its channel, which once could shift hundreds of feet or even several miles in a year because of floods. These wetlands originally flooded every winter and spring, but levee construction, agricultural encroachment and the construction of dams upstream have also eliminated the flooding process. Today about 100 miles (160 km) of the river’s riparian forests are undergoing active restoration.

Methods

I first went to http://atlas.ca.gov/ (CA Atlas) and downloaded and pyramided 19 Tiff. files along the Sacramento River. I then tried to Mosaic the tiles. After trying 3 times I sorted the tiles into blocks of 4 and mosaicked those then mosaicked the 4 blocks into 1 image. The Mosaic was pixelated and I lost one of my color bands. (i.e. only a 3 band.) I then created a shapefile and tried to run an unsupervised classification. I think the image was too high a resolution and it didn’t run. The next technique I tried was Interactive Supervised Classification. This method requires creating spectral signatures in ArcGIS™. I drew thirty polygons to represent samples of five land use sets: water, barren, urban, grass and agriculture. The spectral analysis page in ArcGIS™ is shown below.

I carefully digitized the polygons, trying not to cross into other land use areas and make them as homogeneous as possible. This is different from the unsupervised classification in that I needed to create a signature file (.gsg) in order to classify my image. The spectral signature file contains small areas within the image that represent the “spectral characteristics” of a known cover type (i.e. one of the land cover classes in a land cover classification schema). Spectral signatures are added to the Training Sample Manager. Several tools exist to review individual signatures (such as the histogram, scatter plot, and statistics). In addition to the tools in the Training Sample Manager, the Dendrogram routine in the Spatial Analyst Toolset was also be used to evaluate how similar or different spectral signatures are from each other. The dendrogram looks at pairwise combinations of spectral signatures and then provides a “tree” diagram graphically showing this similarity.

I then reclassified the training samples by hiding all classes except one to get a better idea of what that spectral class represents. I used the symbology tab of the classified image and assigned “No color” to all of the classes except one. This was very time consuming and I finally decided I had too many land cover types. I was really only interested in the Barren, which represented grass, low shrubs, and non-vegetated areas (i.e. rock, soil, concrete). This represents my main area of concern for lack of salmon habitat.
I then digitized the non-vegetated area along the river between the levees. This was named the Areas of Concern.
I then went to the field to look at what was actually there and looked for specific criteria referenced above. (Shade, protection and food source.)


Results

 
The first site I looked at was prefect. A lot of shade and debris.


">
This is an erosion repair site designed by the US Army Corps of Engineers


 
Unfortunately this is mostly what I found. Rip rap Lining placed on Sacramento River East and West levees.


 
Rip rap lining placed on Sacramento River East levee. Rip rap composed of old concrete blocks used to stabilize levees.


 
Concrete Lining placed on Sacramento River East levee undercut by river and breaking apart and rip rap placed on West levee. Rip rap placed on Sacramento River levees in this area is a death trap for spawning salmon.
At this point I went back to my classification.


Analysis

The following photo is the final project in ArcGIS. The bright yellow areas show the Non-Vegetated areas that still need to be repaired for the preservation of the salmon. The analysis was used to classify poor riparian habitat; however, due to image and time limitations, the result was generic. The results can be used a screening tool to help direct additional studies and target areas for conservation and/or restoration. By identifying areas of concern, poor habitat areas can be identified for analysis and restoration. This is especially important in the current climate of restricted private and government funding.


Areas of Concern for Salmon Safety

Conclusions

My main issue was the amount of time the project took. Every step was extremely time consuming. Selecting training sets, mosaicking, reclassifying, and the processing by the computer were all time consuming. Cropping or masking the image would have helped at reducing some of the processing time; maybe by limiting the analysis to just the area between the two levees. I think the supervised classification I developed for this project was a good start and is somewhat helpful in picking areas that are likely detrimental to spawning salmon. If more time were available I would have selected better training sets, spent more time reclassifying, made more adjustments and even rerun the classification until the results were what I wanted.

 

References