"There is no question in my mind that Sacramento is at the highest flood risk in the nation," said Darryl W. Davis, director of the hydrologic engineering center for the U.S. Army Corps of Engineers (COE) in Davis. There are a few specific factors that contribute to this fact. Sacramento sits at the confluence of two major rivers, the American River and the Sacramento River. Similar to New Orleans, many parts of the city are below the river and rely on levees to keep them dry in times of high water, reports Deb Kollars of the Sacramento Bee (Kollars, 2005). The Sacramento Area Flood Control Agency (SAFCA) points out that much of the current levee system was built 150 years ago by settlers and farmers and they are not up to current engineering standards. Folsom Dam was authorized in 1944 and completed in 1956 to help prevent flooding downstream. It was originally designed to provide in excess of a 500-year level of flood protection. From 1951 to 1964, three record storms hit causing flooding in Sacramento, its protection was downgraded from 500-year to 60-year protection (SAFCA, 2007). The dam itself has not deteriorated structurally, but its flood protection has gone down as storm levels have increased.

A dam protects from flooding by taking the peak off of the inflow hydrograph and storing the water and releasing over a longer period of time. The volume of water is the same, but it is stored and released later. See Figure 1.

Figure 1: Inflow vs Outflow

The problem with the Folsom Dam is that the low level outlets are not able to release water fast enough to draw the reservoir down so there is more room to store the incoming flood. The reservoir fills up and then begins to flow over the spillway.

This problem has caused Folsom Dam to be the

center of concern for the Bureau of Reclamation who operates the dam. Funding was approved and an auxiliary spillway is being designed with a much larger capacity at lower reservoir levels. This will enable the reservoir to be drained faster in anticipation of a large storm event. It will also enable the dam to pass more flow before it would be overtopped and fail releasing its 1,000,000 acre feet downstream causing catastrophic damage. I have been working on a model study for the COE of the auxiliary spillway. The model is completed and is in the final stages of data collection. I became interested in Sacramento and what would happen during a large storm event. GIS can be used to create cross sections of the floodplain which can be imported into the Hydrologic Engineering Centers River Analysis System, known as HEC-RAS. The flood can be modeled in HEC-RAS and the water surface exported to GIS and floodplain maps can be delineated.

Objectives

My project involved:

· Getting the required data sets and layers to be able to produce a floodplain map for Sacramento of a large (500-year or greater) flood. I needed a DEM, the NHDplus flow lines and water bodies.

· Using HEC-GeoRAS and ArcMap to create cross sections to use in HEC-RAS to route the flood.

· Determining what area would be flooded.

· Examine what is being flooded by getting land cover data.

Methodology

To make a floodplain map of Sacramento I needed to start with a digital elevation model (DEM) layer. I downloaded a 1/3 arc second (10 m) DEMs from the National Elevation Dataset at http://ned.usgs.gov/. Figure 2 shows the seamless server from the USGS.

Figure 2: USGS Seamless Server

I had to get several maps because they were too big to download all at once. After downloading them and importing them into ArcGIS I had to clip them using the command Extract by Mask to get the area that I wanted by creating a polygon shapefile and then drawing the shape I wanted by hand using the Editor toolbar. See Figure 3.

Figure 3: Extract by Mask

Figure 4: Mosaic to New Raster

Then I combined the rasters using the command Mosaic to New Raster as shown in Figure 4.

With the shape polygon, I can Clip features from other shapefiles like flowlines, waterbodies, roads, land cover or anything else, as demonstrated in Figure 5.

Figure 5: Clipping Shapefiles

After preparing the data I had to get the programs that I would need to do the flood routing. The Army Corps of HEC- RAS program can be downloaded for free from at their site. At the same site there is a link to download HEC-GeoRAS which was also needed for this project. With HEC-GeoRAS there is a specific set of instructions to install it. Several programs (XML, MSXML, AP Utilities Framework) have to be uninstalled and then reinstalled. Also at the USACE site are user manuals for both programs which can be downloaded. Once these programs are installed, analysis is ready to begin.

Analysis

With the GIS data processed to a more manageable size, I was able to start analysis. I found out after I started trying to get HEC-GeoRAS to give correct cross sections that the map and shapefiles used have to be in a projected coordinate system, not geographic. I used Batch Define Coordinate System to select a projected coordinate system. I used the NAD 1983 StatePlane California III FIPS 0403 (Feet) projection. With the correct coordinates I was able to proceed through the process of creating stream centerlines, flowpaths, and cross sections using Arc-Map and the HEC-GeoRAS tools.

See Figure 6.

Figure 6: GeoRAS Cross section in ArcMap

Once the required layers are created and processed, the data is exported to HEC-RAS. The cross sections are opened and have to be processed by hand. The bank stations have to be specified by hand and a Mannings n for the river channel and the right and left overbanks.

I had to use large cross sections to try and contain the flow, but HEC-RAS only allows 500 points for each cross section. There is a filter tool that will eliminate some of the points that are within a specified horizontal and vertical tolerance. Figure 7 shows a screen shot of a HEC-RAS cross section.

Figure 7: HEC-RAS Cross section 1

It was necessary to use HEC-RAS levees to help define where the water would and would not go. These levees are not actual levees but more like markers to tell HEC-RAS that the area between the levees has to fill up before it can spill out.

After processing the cross sections and getting them prepared, the flow rate is input and a boundary condition of normal depth downstream was used. Just the slope downstream is required and HEC-RAS calculates the normal depth. The return periods and flow rates that I used come from communication with the COE through my work on the model study. The slope was found by creating 5 foot contours using Spatial Analyst à Surface Analysis à Contours, at the downstream section, I measured the distance between two contours that the river passed through and divided five feet by the length between the contours to get a slope of 0.000781 ft/ft. I made three runs in HEC-RAS using three different flow rates. The levees are supposed to contain the 200-year flood of 160,000 cfs so I ran that flow and moved the levees to get it to stay within the banks. Then I ran what would be about the 500-year flood, of 400,000 cfs. I had to adjust the levees again to make the results look right. Then I ran the PMF release of 808,585 cfs and had to adjust the levees again.

I did not export the 160,000 cfs flood to ArcMap because it wasn’t very exciting. The 500-year flood and PMF data were exported to ArcMap. Using the HEC-GeoRAS toolbar, the files have to be converted to XML and then read using RAS Mapping à Read RAS GIS Export File. Then the water surface layer is created using RAS Mapping à Innundation Mapping à Water Surface Generation. Floodplain delineation is performed using RAS Mapping à Innundation Mapping à Floodplain Delineation. Then the Grid Intersection computes where the DEM and water surface meet and a raster of water depths is created. The result is shown below in Figure 8.

Figure 8: Raw water surface from HEC-RAS in ArcMap

The HEC-GeoRAS user manual states:

“Floodplain delineation results should be carefully examined. Spurious inundated areas may be present and should be deleted in the GIS or prevented from occurring by using levees in HEC-RAS. Inappropriately placed cross sections may result in the bounding polygon data incorrectly limiting the floodplain boundary. The floodplain delineation process using GeoRAS is an iterative process that should be used to refine the hydraulic model in HEC-RAS.”

I did some iteration though more would likely be required for an official flood map.

I created a polygon shapefile and traced by hand the flood plain using the contours and some engineering judgment. See Figure 9. Then I added the land cover layer and extracted (Extract by Mask) just the area that was flooded. I had to join the extracted attribute table to the original to get the names of the different land covers. I added a field and used the Field Calculator to calculate the area of each type of land cover that was under water by multiplying the number of cells by the 900 m² cell size and converted it to square miles. I exported the data and plotted it on a pie graph.

Figure 9: Hand drawn floodplain

In the attribute table of the land cover is the area of each type of land use that is submerged. The pie graph is shown in the Results section of this report.

Results

For the PMF release of 808,585 cfs from Folsom Dam, I found that 175 square miles would be flooded. The Figure 10 shows the flooded area and the land cover.

Figure 10: Innundated area land cover

The Figure 11 shows the amount and percent of each type of land cover that would be flooded.

Figure 11: Innundated area land cover in square miles

Of the 175 square miles, approximately 65% is developed. There are a lot of housing subdivisions along the river that would be the first to be flooded. The results for the 500-year flood were almost the same. The floodplain is pretty flat so the water flows to pretty much the same elevations in the model that I developed in HEC-RAS.

Discussion

A finer resolution DEM than 10 meters would have been given me better results, especially for my river cross sections, but that was all that I could find. I looked at the California Spatial Information Library (CaSIL) where they had a lot of data, but I had a hard time navigating their site and finding what I wanted.

At this point I do not claim to be any sort of HEC-RAS wizard. I stumbled through the flood routing with a lot of help from fellow student Kameron Ballentine who has a little experience from his work this summer. There are a several things that could be improved on. I probably should have used more cross sections. The cross sections cannot cross each other. When I started I didn’t realize that the cross sections can have a kink in them which would have enabled me to get more cross sections in. Kameron said that seven cross sections per mile is usually required, but I had 26 miles of river to route which would have been very time consuming to make and process that many cross sections. Even if I had done a lot more cross sections, I did not have data for the 10 bridges and two railroad bridges that cross the river so my results still would not have been super accurate. Usually when using HEC-RAS a cross section just upstream and just downstream of any bridges or constrictions is required. Without having any accurate survey measurements of the bottom slope and Mannings roughness factor, my guesses were all I had to use.

As can be seen on the map of the HEC-RAS generated water surface, there seems to be something funny going on at the third to last cross section. See Figure 12.

Figure 12: HEC-RAS cross section 2

Because the water is not able to be contained in the main section of topography, it flows out to the left. The DEM shows the elevation decreases in this direction as it runs toward the ocean. I wouldn’t be able to draw a cross section long enough to contain the flow. In HEC-RAS, it puts a “wall” up at the zero station and contains the water inside the cross section making the water artificially deep in that section. The water in reality would flow downhill away from the river and the depths would not be a great. The next cross section, shown in Figure 13, is able to contain the flow so it makes a weird shape on the flood map. See Figure 14.

Figure 13: HEC-RAS cross section 3

Figure 14: Errant water surface in ArcMap

The water would probably still flow to the south as shown on the map, though the depths would be less. A couple more iterations of changing and adding cross sections and importing them into HEC-RAS and running the model and then following the steps in ArcMap would have produced better results.

I had used the contours to draw the flood map, but they are hard to follow, so later I tried to change the symbology of the DEM and added the maximum number of classes and then changed the class intervals to show more elevation change at the lower elevations. See Figure 15.

Figure 15: Changing the symbology of the DEM

This shows a little better the contours at the lower elevations showing the water would drain to the west and south.

I ignored the area to the west of the Sacramento River, for the sake of easing computation and saving time. Also, if the American River was flowing this high it is likely that the Sacramento River would be flowing very as well and flooding through Sacramento.

Summary and Recommendations

I was able to find the layers and data needed to make a highly simplified model of the American River through Sacramento. The results are not highly accurate and I would not recommend that they be taken for a real flood map. The PMF flood that was modeled would be highly unlikely and the map generated does not give a lot a helpful information. It was more an exercise in using HEC-RAS, GeoRAS and ArcMap together. I can see now that I need more cross sections and perhaps bigger in the sections on the flat part of the floodplain. I would spend more time in calibrating the model to stay within the levees for the 160,000 cfs before trying higher flows and then I would use a smaller flow of say 200,000 cfs to give more realistic results.

I learned some important things about how to get the programs to all work together. I feel that I would be prepared to take a real project and be able to make some actual progress and produce some more meaningful results. Though I would still need some assistance, I feel that I could be an asset in this regard.

Sources

Cox, Nathan, Personal Communication, 16 Oct 2007, Hydraulic Engineer for U.S. Army Corps of Engineers

Kollars, Deb, 2005, Oct 30, Tempting Fate: Are we next?, Sacramento Bee

SAFCA, Sacramento Area Flood Control Agency, accessed 25 Oct 2007, Sacramento Flood Risk, http://www.safca.org/floodRisk/index.html

U.S. Army Corps of Engineers, September 2005, HEC-GeoRAS An Extension for support of HEC-RAS using ArcGIS Version 4, http://www.hec.usace.army.mil/software/hec-ras/hec-georas_downloads.html

Data Sources

http://ned.usgs.gov/

http://landcover.usgs.gov/

http://www.hec.usace.army.mil/software/hec-ras/hecras-hecras.html

http://www.horizon-systems.com/nhdplus