Methods

Digital Elevation Model (DEM) data of our desired study area was downloaded from the USGS National Elevation Dataset. When using the hydrologic features in ArcMap, sinks are often considered to be errors and were therefore filled. The ArcMap’s toolbox to was used to fill any sinks or depressions to account for this error. In order to derive hydraulic flow networks and accurate watershed divides from the DEMs, each pixel was assigned a flow direction. In short, every pixel was assigned a numerical value which was used to delineate the direction in which water in that pixel would flow. Now that we had the direction for each pixel, we determined all accumulated cells flowing into each downsloping cell.

We then decided on a Pour Point (Oroville Dam Location) which would grid the highest flow accumulation values within a given distance. Once completed, a basin layer illustrating the possible watershed boundaries was extracted. The desired watershed boundary was then clipped to emphasize the specific extent of our proper trout creek. Once completed, a hydraulic flow network was generated to illustrate a stream network from the affected areas if water were to be expelled from the dam.

Once we had our results from the hydraulic feature extraction model, we found our results to be sub-par due to a multitude of various accuracy reasons. The largest error we found was that due to the levels of water, the expulsion size of the water cannot be delineated from streamlines but instead more of a flood type model. We also found the streamlines to cut across streets and tall building, which we found to be inconclusive with our results since the level of water (in ft.) would not be enough to completely inundate these buildings, specifically when we analyze highly developed areas including Yuba city and Sacramento.

We then decided to conduct an analysis using and two dimensional unsteady flow model using the HEC-RAS software. Once our Digital Elevation Models were stitched together in ArcGIS, we extracted this geometric information for the two dimensional unsteady flow simulation of the dam breach. The two dimensional flow area was computed along with accurate computation points aimed at properly delineating the water extent of the model. Using this information, the boundary conditions were manually set to reduce the computation extent per model. The delineation of upstream and downstream manually using a colorized overlaid DEM image. In order to better account for the roughness (or lack of it) in the extent of the affected area, we estimated that the Manning roughness coefficient would fall at a value of .06 due to the high levels of cropland in the estimated affected area. In addition, a normal depth downstream boundary was set along with the estimated peak flow conditions.

In order to properly estimate the flow of the model, we estimated the cfs values, we used information from a previous Dam break and used an identical slope in the rate of water being expelled. In order to diversify the levels of water, we decided to include various cfs extents to better account for different scenarios. These are depicted by the maps illustrated in the Maps section. Lastly, we ran the simulation with an 8 hour interval to reduce computation time and due to the unexpected crashing that would occasionally arise.

Once the simulation was finished, flood hazard maps were generated by exporting the HEC-RAS model output results to ArcGIS where they were processed to identify the flood prone areas. Additional USGS Development and cropland information was used to outline effected urban and farmland areas.

In addition to the flood hazard maps, we created two NDVI maps using Landsat imagery obtained from the USGS EarthExplorer website.