Watershed

Watershed maps show us how water travels across an area, where it accumulates and where it comes from. Watersheds are the systems of rivers and streams, combined with lakes, ponds and other water features that ultimately provide life and sustainability for an area and all its inhabitants, including humans.

Everything upstream comes downstream. Some of the smallest changes upstream can have cascading effects down below including a decrease in available water, a decrease in water quality, among others. By understanding the way that water travels across an area, local government and natural resource agencies can make decisions to preserve the availability and quality of that water.

The purpose of this lab was to create a watershed area map using spatial analysis tools in ArcMap. The first step was to choose an area and download the data. I chose an area in Humboldt County, California, just east of the Redwoods National Forest. This watershed ultimately pours into the ocean, but for my map I chose a pour point near the airport and highway, a 'last stop' if you will before reaching the ocean. After choosing a location, I downloaded elevation data (DEM) from the National Elevation Database. The data came in raster format and I had to check its coordinate system then project the raster to the appropriate projected coordinate system (in this case NAD 1983 UTM Zone 10N).

As with any data, there is a chance for error and when wanting to find drainage routes, its important not to have large changes in elevation between cells that aren't actually there. For that, I used the 'Fill tool' which goes through and 'fills' those in for me. Then using another spatial analysis tool, 'Flow Direction', I was able to find the direction that water would flow within my DEM. The 'Flow Direction' tool does this by finding the steepest downhill neighbor to the cell (at least when using the D8 option). Each cell is then given a value and these values tell you what direction of flow is for the cell.

The next tool I used was the 'Flow Accumulation' tool. This tool uses the data within the raster created by the 'Flow Direction' tool to determine how many cells 'flow' into each cell. Each cell is then given a value and this value tells you how many other cells ultimately 'flow' into that one. Once you know what direction the water would flow and where it would accumulate you can choose a 'pour point', which is a common outlet for the water upstream. This could be the ocean, the mouth of a canyon, honestly wherever the water meets up. Using the draw toolbar in ArcMap, I created a point and 'Converted it to a Graphic'. Now with my flow accumulation and pour point, I could find my watershed boundary.

Using the 'Watershed' tool, I inputted my 'Flow Direction' data and my pour point. This creates a boundary for me that includes all the waterways that would lead to my pour point. After turning the resulting raster into a polygon, I was able to use the 'Extract by Mask' tool to clip my 'Flow Accumulation' data to just that area. So now I've got my 'Flow Accumulation' data nicely clipped to just the watershed area that I am interested in, but how do I know if all those silly lines on the 'Flow Accumulation' raster are really streams. Experts, I tell you, and maps made by experts. I referenced my 'streams' against a topography basemap provided by the USGS through ESRI. I found there were a ton of streams that my accumulation data wasn't picking up, which told me that my threshold was off.

Using the symbology classification options available I experimented to find a 'breaking point'. A point where I determined what had a high enough value to be a stream and what didn't. The lower the threshold, the more data got to be considered streams, the higher this threshold, the less. I messed with this until it matched up pretty well with the USGS topography map, which for my area was around 5500. I felt like that was a little high, but any lower than that and I had 'streams' try to jump over hills and areas that didn't make any sense. Once I got the threshold set, I used the 'Reclassify' tool to create a raster that only had cells where streams should be an 'NoData' elsewhere.

From the 'Flow Direction' and 'Flow Accumulation' data I used the 'Stream Order' tool to categorize the streams by how many streams flowed into them. Stream order is a method of classifying stream networks and can help us better understand habitat diversity, the geomorphic features of the landscape, and provides a nice summary of the watershed area. After getting the stream order, I used the 'Raster to Polyline' tool to turn that stream data into a polyline that I could then use to better symbolize the streams and measure the length of the whole stream as well as the length of each individual segment.

View and download PDF below or download here.

Another neat thing that can be done within ArcMap and ArcPro is 3D analysis. This can help you find the steepest slope, the elevation along a line, a 'line of sight', and a contour line. For this map I drew a 3D line and found the height change across it and create a profile graph to show the change in elevation.

Lab Questions

Q1: When projecting elevation data, what is the appropriate choice for the resampling technique? Bilinear? Nearest Neighbor? Cubic?

When projecting elevation data it is appropriate to use either bilinear or cubic, this is because the nearest neighbor option will resample creating averages of the cells in the area.

Q2: If a cell in your flow direction raster has a value of 8, what does that mean?

If a cell in the flow direction raster has a value of 8 then it means that the flow through that cell is moving southwest. The numbers in a flow direction raster show direction.

Q3: If a cell in your flow accumulation raster has a value of 8, what does that mean?

If a cell in the flow accumulation raster has a value of 8, it means that there are 8 cells that flow into that one cell, the higher the number the more cells are flowing into that cell.

Q4: Where are the higher elevations in the image above? The white areas? The dark areas? In between? Why?

In a flow accumulation raster, granted that you haven’t inverted the symbology, the white lines are the lower elevation areas because they are the cells with the highest flow values, generally those with lower elevations.

Q5: Imagine you are working with a 2m lidar DEM and a 10m USGS DEM of the same watershed. You want to create a stream network from the flow accumulation raster.

You have determined reasonable ‘threshold’ value to use to reclassify the flow accumulation into ‘stream’ and ‘not stream’ binary cells using the 10m USGS DEM.

You can use this threshold value on the 2m lidar DEM to get the same stream network results. True or False? Why?

You couldn’t use the same threshold because the resolution on the 2m LiDAR would be higher, therefore whatever value you chose for the 10m would probably be too low and too much data would be considered a stream.