Lab 5: Watershed hydrology

Size: px
Start display at page:

Download "Lab 5: Watershed hydrology"

Transcription

1 Lab 5: Watershed hydrology Objectives In this lab, you will learn the fundamentals of flow routing and channel network extraction, and analyze precipitation and discharge data to explore the control of basin size on the hydrograph. Figure 1. Digital elevation model of the West Branch Susquehanna River, PA. Background Channel networks evolve as a mechanism to efficiently convey water and sediment through the landscape. As sediment transport and bedrock erosion by river channels are ultimately controlled by the magnitude of stream discharge, a detailed understanding of watershed hydrology is essential for geomorphologists. Specifically, we are interested in how rainfall is converted to discharge via the filter of the drainage basin. In this lab, we will be analyzing the watershed of Sinnemahoning Creek, a tributary of the West Branch Susquehanna River just north of State College, PA (Fig. 1). First, we will use the digital elevation model to extract a network of stream channels. We will then analyze precipitation and stream discharge data from a small storm in the fall of 2014, in order to build intuition about how water moves through a forested, relatively undeveloped watershed. 1

2 5.0 Summary of provided datasets You have been provided with a file geodatabase lab_5_data.gdb, which includes a 20 m resolution digital elevation model and hillshade, and a point feature class indicating the locations of 3 of the 4 USGS stream gages used in this lab (Fig. 2). Name Data type Grid size Description Source sinne20_dem raster 20 m Elevation (m) USGS NED sinne20_hillshade raster 20 m Hillshade gage_locations point feature class N/A Location of USGS stream gages USGS Figure 2. Summary of provided datasets 5.1 Generating flow accumulation grids in ArcMap In this section you will use the topography data to generate a map of flow accumulation and extract a network of streams from the landscape. Before modeling flow routing and accumulation across the landscape, it is necessary to run the Fill tool to remove any enclosed depressions in the landscape, thereby creating a hydrologically smooth DEM (Fig. 3). Figure 3. Example of pit filling in one dimension (from ArcGIS online help) Navigate to and open the tool \\Spatial Analyst Tools\Hydrology\Fill\, and name your output raster something like sinne20_filled_dem (Fig. 4). Figure 4. Spatial Analyst Fill tool dialog. Next, we need to generate a flow direction grid, which tells us the direction of steepest descent for each cell in the DEM. While there are a few different approaches to this calculation, we will use the simplest, known as the D8 method. The D8 method looks at the 8 neighboring cells for each pixel, and chooses assigns a value to the direction of steepest descent that corresponds to one of 8 possible directions (Fig. 5). 2

3 Figure 5. Schematic of the D8 flow direction algorithm (from ArcGIS online help) Navigate to and open the tool \\Spatial Analyst Tools\Hydrology\Flow Direction\, and be sure to use the filled DEM as your input raster. Check the box next to Force all edge cells to flow outward and give the output raster a name like sinne20_flow_direction (Fig. 6). Figure 6. Flow Direction dialog box. Next, we want to use the flow direction grid to determine the flow accumulation at each cell, which is simply a count of the number of upstream cells that flow to a given pixel (Fig. 7). 3

4 Figure 7. Schematic of the Flow Accumulation algorithm (from ArcGIS online help). Navigate to and open the tool \\Spatial Analyst Tools\Hydrology\Flow Accumulation\, enter the flow direction grid as the input raster, and name your output something like sinne20_flow_accumulation (Fig. 8). Note that this step may take a few minutes depending on your computer, so be patient. Figure 8. Flow Accumulation tool dialog box. It is important to note that units of the flow accumulation values are in number of grid cells. To make things a bit more convenient, convert the flow accumulation grid to units of km 2 using the raster calculator (see section 4.3 in last week s lab) to make the following calculation: sinne20_flow_accumulation_sq_km = "sinne20_flow_accumulation" * 20 * 20 / where the 20 * 20 corresponds to the grid size and the converts from m 2 to km Extracting stream networks using the flow accumulation grid In this section, we will use the flow accumulation grid to extract a stream network and convert it to a line feature class for easier visualization. 4

5 First, we need to decide on how to delineate between what is a stream or not. There are a multitude of ways to delineate stream networks, but here we will use a simple drainage area threshold based on the flow accumulation grid. Go to the symbology of your flow accumulation grid (the one in units of km 2 ), and choose a classified symbology with 2 classes. You can open the classify window to adjust the lower break value that determines the cutoff drainage area. Experiment here to find a stream network that looks appropriate for the scale of the map (Fig. 9). Figure 9. Flow accumulation grids with cutoff of 0.01 km 2 (too many channels!) and 100 km 2 (too few!) Once you have chosen an appropriate threshold, we need to generate a raster that has values of 1 everywhere with a channel and NoData everywhere else. Navigate to and open the tool \\Spatial Analyst Tools\Conditional\Set Null\ (Fig. 10). For the input conditional raster, choose your flow accumulation grid (the one in units of km 2!). Then, type in the expression that defines the cells you wish to set to NoData, in this case any value less than the cutoff drainage area in km 2 you chose above. Be sure to set the input false raster or constant value to 1, and give the output raster a name that reflects the cutoff area. Figure 10. Set Null tool dialog box. Note that you should choose the cutoff value you determined above. Next, we want to convert this raster to a line feature class for better visualization. Navigate to and open the tool \\Spatial Analyst Tools\Hydrology\Stream to Feature\, and enter your stream raster and flow direction grids as inputs (Fig. 11). For the output, choose a name that reflects the cutoff drainage area, and be sure to unclick the box next to Simplify polylines. You can change the color and thickness of the resulting line feature class to your liking by adjusting its symbology. 5

6 Figure 11. Stream to Feature tool dialog box. One trick I have found for visualizing stream networks is to extract the value of drainage area to the feature class using \\3D Analyst Tools\Functional Surface\Add Surface Information\ (Fig. 12). Be sure to check the output property Z_MIN. Figure 12. Add Surface Information dialog box. Once you have done this, you can edit the symbology so the line thickness reflects the field Z_MIN, which in this case represents the minimum drainage area of each line segment in km 2 (Fig. 13). For smooth transitions, be sure to select more than 10 classification levels or so. If you want to be really fancy, see if you can set the thickness to correspond to the local scaling between channel width and drainage area Experiment! 6

7 Figure 13. Stream network line thickness set by drainage area value using the Add Surface Information tool. Finally, for your overview map you will be asked to draw the drainage basins for each of the 3 USGS stream gages provided in the point shapefile gage_locations. While there are ways to generate watershed outlines in ArcMap, the goal of this exercise is for you to build intuition about the landscape and perhaps sympathize with the pre GIS geomorphologists! Figure 14. Sketching a drainage basin by hand (red) for the outlet at the white dot. A good skill to have! 7

8 It will probably be easiest to use the drawing toolbar to draw a polygon, tracing out the drainage divides starting from the outlet and moving either clockwise or counter-clockwise around the basin (Fig. 14). You can also create a polygon feature class as in the landslide lab, but it s probably overkill. If you want to a check on how well you did, compare the area of your polygon (right click\properties\area) to the value for the flow accumulation at the outlet (use the info button). 5.4 Analyzing hydrologic time-series data You have been provided with an Excel spreadsheet that contains time-series data for precipitation (1 station) and stream discharge (4 stations) for a small storm event in October of 2014 (Fig. 15). Date Time (hours) Rainfall Rate (mm/hr) Stream Discharge (m 3 /s) Emporium, PA (near ) /13/ /13/ /13/ Figure 15. Data table containing time series of hydrologic data for October 2014 storm The Date column contains the date and time for each hourly measurement, while the Time column shows the elapsed time in hours from midnight on October 13, You can make your plots with either column, but just remember to label your axes appropriately! The Rainfall Rate column contains the hourly rainfall rate in mm/hr from a station in Emporium, PA, near the USGS stream gage in the center of the mapping area. The Stream Discharge column contains the hourly water discharge data from the 3 USGS stream gages provided in the feature class gage_locations, along with one additional USGS stream gage about 40 km downstream of your mapping area on the West Branch of the Susquehanna River. For this section, you will plot all of these time series data on a single plot, to investigate how different parts of the river network respond to a moderate precipitation event. Before we start, fill in some of the information for each station in the table provided in the spreadsheet (Fig. 16). Name Waldy Run near Emporium, PA Driftwood Br Sinnemahoning Cr at Sterling Run, PA Sinnemahoning Creek at Sinnemahoning, PA Figure 16. Table to be completed in the Excel spreadsheet. Be mindful of units! West Branch Susquehanna River at Renovo, PA Station ID Drainage area (km 2 ) 7705 Channel width (m) 178 Q base (m 3 /s) Q peak (m 3 /s) Peak lag time (hours) Total storm runoff (mm) Total storm rainfall (mm) Total storm runoff/rainfall (%) 8

9 For Drainage area, use the info button to get the value of the flow accumulation grid in km 2 for each station in ArcMap. For Channel width, add an imagery basemap to your map, and use the ruler tool to estimate the width of the stream channel at each station. Figure 17. Measuring channel width for Station off of the imagery basemap with the ruler tool. For Q base, record the value of the base flow stream discharge prior to the storm in m 3. For example, the value for Waldy Run is m 3 /s (Fig. 15). For Q peak, record the highest stream discharge value for each station in m 3 /s. For example, the peak flow for the West Branch Susquehanna River at Renovo is 73.3 m 3 /s. The Peak lag time is the difference in time between the rainfall peak and the peak stream discharge at each station (Fig. 18). It will be easiest to use the column Time (hours) to calculate this difference. The Total storm runoff TSR can be thought of as the total volume of water discharged by stream in excess of baseflow during the storm V excess, divided by the drainage area A. TSR = V excess/a (Equation 1) To calculate V excess, we want to take the integral of excess discharge during the storm (Fig. 18). To save time, I have calculated these values for you and entered them into your table (Fig. 16). 9

10 V excess Figure 18. Hydrograph showing definition of Peak lag time and Vexcess Next, we need to make is an estimate of the total storm rainfall over the time period of interest. The total rainfall for the storm R T corresponds to the integral of the rainfall rate RR (area underneath the curve). We can get a rough estimate by summing the rainfall rates and multiplying by the average interval of measurement ΔT: R T = Σ RR x ΔT (Equation 2) You can easily sum the rainfall rates in your excel spreadsheet to get RR, and the time interval is simply 1 hour. Note that this value will be the same for all sites, as we are only using one rainfall station. The last calculation, the ratio of total storm runoff to rainfall, is obtained by dividing the runoff in mm by the rainfall in mm for the storm and multiplying by 100 to convert to units of %. This value gives you a sense of how much of the rain that fell during the storm actually made it out of the watershed as runoff. Think about what controls this number for the discussion in your written report. 5.5 Normalizing discharge values and plotting hydrologic data To visualize the variation in hydrologic response for different drainage basins, it is helpful to normalize the stream discharge to some reference value. In this case, we will normalize the discharge by dividing by the base flow Q base, which is different for each station (you should have already recorded the values in your Excel spreadsheet). 10

11 Open your Excel spreadsheet, and make 4 new columns with the same headings as the stream discharge, but name it Normalized stream discharge (Fig. 19). Figure 19. Excel spreadsheet showing normalized stream discharge calculation Note that the normalized stream discharge is non-dimensional! For your final chart of hydrological data, you should plot the normalized stream discharge for all 4 USGS stations, as well as the rainfall data on a separate axis. Be sure to flip your rainfall axes by selecting Values in reverse order in the Format Axis options. Your final plot should look something like Figure 18, except your discharge values will be normalized and there will be 3 additional discharge curves. 11

12 Lab 5 deliverables, due Monday March 2 before lecture (40 pts total) Please submit as single PDF file to the Lab 5 drop box on Angel (5 pts) An overview map showing the stream network overlaid on the DEM or hillshade, with the 3 visible USGS stream gages labeled and their drainage basins outlined. Make sure to clean up your legend items so that your map is publication quality. There are no excuses for missing items, scale bars, incoherent labels, etc. Make sure your stream network is legible over the color map of your DEM. It may be easiest to keep the rasters black and white. (5 pts) A composite time series plot showing the rainfall data and normalized discharge data for all 4 USGS stream gages. The hourly data are in the Excel file lab_5_hydrologic_data.xlsx Note that the rainfall data is taken from a rain gage near station Be sure to label your axes and scale your axes limits appropriately such that the data are easily interpreted. (5 pts) A completed table of calculations Fill in the table provided in the Excel file lab_5_hydrologic_data.xlsx. Note that these measurements and calculations will go into your plots below. (5 pts) A plot showing baseflow, peak discharge, and channel width as a function of drainage area for each of the 4 USGS stream gages. Plot channel width on the secondary axis, and make all axes logarithmically scaled. Include a power-law regression for each dataset, and show the best-fit equation for each. Label your axes and include a legend! (20 pts) A written report no more than 3 pages long (12 pt font, 1.5 line spacing, 1 margins), which should include the following A brief introduction (no more than 1 paragraph) A description of the observed differences in both site characteristics (drainage area, channel width, position within the basin, etc) and differences in the hydrograph response for each of the four USGS gaging stations (~2 paragraphs focused on synthesizing, rather than simply repeating the measurements and calculations in your data table) A discussion focused on addressing the following questions: 1. How was water partitioned in the landscape during this storm, and what fraction of the total rainfall left the catchment as runoff? Be specific about which processes you think were active during and after the storm event. 2. How might this partitioning of storm water in the landscape vary throughout the year? In different climate zones? Will it be different for developed areas? Again, be as specific as you can, drawing on the processes we learned in lecture. 3. For this exercise, we only have one location where rainfall was measured. In reality, rainfall likely varied across the landscape during this storm, both in the total amount of precipitation delivered, and in the timing of delivery as the storm moved across the basin. Discuss what influence the direction of the incoming storm would have on the shape of the hydrographs. How appropriate do you think our assumption of uniform rainfall is? Note: this is not a trivial question, bur requires you to think about the timescale of the storm and the timescales of the hydrologic response. 12