stream_discharge_student_handout.v2
docx
keyboard_arrow_up
School
Georgia State University *
*We aren’t endorsed by this school
Course
6408
Subject
Geography
Date
Apr 3, 2024
Type
docx
Pages
10
Uploaded by ProfCamel4062
Project EDDIE: STREAM DISCHARGE Student Handout
This module was initially developed by Bader, N.E., T. Meixner, C.A. Gibson, C.M. O’Reilly, and D.N. Castendyk. 26 June 2015. Project EDDIE: Stream Discharge. Project EDDIE Module 5, Version 2. http://cemast.illinoisstate.edu/data-for-students/modules/stream-discharge.shtml
. Module development was supported by NSF DEB 1245707. Module was revised in October 2022 by N.E. Bader, C. Carey, and D. Richardson to repair data sources and update for the new USGS data distribution system.
Learning objectives
:
●
You will understand how to download, organize and analyze streamflow data.
●
You will learn about major climate impacts in their region.
●
You will analyze streamflow data to detect and quantify climate change impacts on water
quantity in their region.
●
You will learn about flood events and how to predict the likelihood of big flood events.
Why this matters:
Fresh water is a fundamental resource for our society. A synonym for streamflow is “discharge,” which is the volume of water passing by a point on a riverbank per unit time. Discharge is also a way to measure the quantity of water that is available for various uses. For example, water rights are often measured in units of discharge. Fish might require a certain discharge in a reach of stream in order to thrive or move through the reach. Conversely, the highest discharges result in floods that may harm people and structures. Discharge is fundamentally connected to our hydrologic cycle. Streams are fed by water that originally fell as rain and snow. Changes in the quantity of rain and snow, their distribution, and their timing, can be expected to cause changes in stream discharge. Clearly, it is important to be able to understand the way discharge varies. How can we measure these changes? We must start with data. In this activity we will use data from the United States Geologic Survey
(or USGS) network of stream gaging stations. The USGS is a governmental organization established in 1879, as part of the Department of the Interior. Originally tasked with the classification and mapping of United States public lands (and assessment of their mineral resources), the USGS has since expanded their role as a provider of impartial information on the status of ecosystems in the United States. (See http://www.usgs.gov
for more details.) Outline:
1.
Discussion of papers read for class and quick PowerPoint introduction
2.
Activity A: Variability in stream flow
3.
Activity B: Changes in discharge over time
4.
Activity C: Peak discharge and flood hazard
1
Activity A
: Variability in real stream data
Viewing and accessing data:
1.
Navigate to the USGS Water Dashboard, at
https://dashboard.waterdata.usgs.gov/
2.
Notice the colored dots, depicting real-time conditions at stream gages nationwide. The dots are colored to show you if the streamflow is unusually high or low for this day of the
year. Red colors are unusually low; blue colors are unusually high. You can click on “legend” at the upper right to see what the colors mean. For example, the dark blue color
indicates that the flow is at or above the 90
th
percentile, which tells you that at least 90% of the flows measured on this day in the past were lower. Put another way, flow is high enough to earn the dark blue dot only about 10% of the time.
3.
You can click on any of the colored dots to bring up a pop-up window with a plot of discharge over the past week. Try this now, then close the pop-up window when you are finished.
4.
Let’s look at a specific stream gaging station, on the Neversink River in New York. In the search box at the upper left, type “neversink,” then select the Neversink Reservoir in Sullivan County, New York. This reservoir is one of several that supply New York City with drinking water. You can scroll to zoom into the map and see the reservoir. Note the
gaging stations on the inflow and outflow of the reservoir. Click on the dot upstream (north) of the reservoir – you want the one called “Neversink River near Claryville, NY (Monitoring location USGS 01435000).” There are several similarly named gaging stations, so be sure you have the correct one. Click on the dot to open the popup window.
5.
Notice that there is a lot of data available for this gaging station. That is because it this station was formerly part of a smaller network of intensively monitored stations in relatively undisturbed locations. These monitoring sites are collectively called the Hydrologic Benchmark Network. (A “benchmark” is something against which other things are measured. In this case, an undisturbed stream is a benchmark against which we can measure a stream modified by human activity.) 6.
At the top of the pop-up window, click the link to open the Site Page, which is a more detailed description of the site. Take a look at the data available here: o
Notice that there is a camera view showing the river over the past few days, and even an infrared camera showing the relative temperature of the water and land.
o
The map shows the drainage basin for this gaging station, which means the area “upstream” where water that originally falls as rain or snow will eventually flow past the gaging station.
o
At the bottom, open the Location Metadata. You can see that the water draining past this point has been collected from a drainage area of 66 square miles (sorry, but the U.S. Government works in Imperial units. That’s 171 square km.)
2
7.
Scroll back to the top to see the different types of data that we can show. Leaving the range at 7 days, select “Discharge, cubic feet per second” to see a plot showing how discharge has changed over the past week.
Questions
:
1.
Look at the discharge data. How variable was discharge over the past week? You can answer this in several ways. First, what was the range (the maximum minus the minimum) of the data? Another good way to think about variability is to think about percent change. First, estimate the mean value of the data by looking at the plot. Approximately how much higher (as a percent) are the highest discharges? 2.
Where you clicked the box to select discharge data, check the box that says “Select data to graph on second y axis,” and choose water temperature. Look at the temperature graph. Based on this graph, what probably drives temperature changes in the Neversink River? Do you think there is any relationship between temperature and discharge? If so, what do you think it might be?
[Aside: don’t try to calculate percent change using temperature data because your answers won’t be meaningful. Why does percent change work with discharge but not temperature? Because of what “zero” means. For both the Fahrenheit and Celsius scales,
zero is an arbitrary temperature. When it is three degrees C outside, that doesn’t mean that there is “50% more heat” than when it was two degrees C outside. In contrast, zero discharge really does mean that there is no water movement, and a discharge of three cubic meters per second really is
50% more water movement than a discharge of two cubic meters per second.]
3.
When discharge is high enough, flooding occurs. Is the discharge you observe here unusually high? Unusually low? Typical for the region? Can we answer these questions with only a week’s worth of data?
Seasonal variation in streamflow:
We might expect flow in a stream to change seasonally. After all, most (or all) of the streamflow that you observe originated as rain and snow falling in the watershed, and precipitation in most places is seasonally variable. Let’s take a look at how streamflow changes over an entire year to see what happens.
Still looking at both discharge and temperature data, go back to the top and change the time range to 1 Year. It may take a minute to load all of the data.
4.
Question
: Temperature is high in summer and low in winter, as you might expect – but what month was the warmest? What about discharge? Was it the same all year? What months had the highest discharge? The lowest?
3
Your preview ends here
Eager to read complete document? Join bartleby learn and gain access to the full version
- Access to all documents
- Unlimited textbook solutions
- 24/7 expert homework help
Activity B: Changes in discharge over time
B.1: Change through time on the Neversink River
Let’s go back to an important question we had at the beginning of this activity. Is stream discharge changing through time? How might we answer this question, given the seasonal variability of the data?
One solution is to reduce the variability in our data by considering summer and winter data separately. In order to do this, we will need to manipulate the data ourselves using Excel.
1.
First, export the data. We are looking for historical data (going back farther than one year), which is not available from this page. To find historical data, return to the popup window that opened when you first clicked on the map, and click on the link at the top that says “Data.” (If you closed this popup window, that’s ok: simply click on the map again to reopen it - but be sure that you click on the correct dot.). From the statistics page you opened, find and select Monthly Statistics. In the new page, check the box next to discharge, Select discharge, and leave the date range blank to get the entire date range. Choose tab-separated data in YYYY-MM-DD format, and save to file. Once you click Submit, a text file called “monthly” will be saved to your computer.
2.
Now import the data into Excel. Open Excel, and select File > Import > Text file. Navigate to your “monthly” file and open it. (If it is grayed out, you may need to change “Text Files” to “All files” in the drop-down menu under the dialog box.) The text file is delimited by tabs, so check the appropriate boxes to tell this to Excel, accepting the remaining defaults.
3.
You only need three columns: year, month, and mean discharge. Take a close look at your columns to see if you can figure out which is which. When you are confident, type “Year”, “Month”, and “Discharge” in the cells above the appropriate columns. Now you can clean up by deleting the remaining columns, as well as the rows above your labels.
4.
Now is a good time to save your Excel spreadsheet. Call it something sensible such as “Neversink monthly discharge” and save it in a folder where you can find it again. You should periodically re-save your spreadsheet so that you don’t lose your work.
Plotting discharge by month
1.
Currently your data is organized by year, and within each year it is organized by month. In order to make a plot of a particular month, you should arrange your data by month instead.
Click one of the cells in the month column and use Sort and Filter > Smallest to Largest to sort by month - you will see all of the January data first, followed by February, etc.
Select all of the data cells from February, in all three columns. (Hint: you can click once to select the top left cell, then scroll down and shift-click on the bottom right cell to select the data you want.) 4
Graph the data with a scatter plot (go to Insert and look at the chart options). You should see a cloud of points with discharge on the Y axis and year on the X axis. I recommend moving this plot to its own tab using Chart Design > Move Chart so that you can see it better.
Excel will also plot all the integers representing the month in another series on the plot. This is obviously not very useful. You can click on a point in the series to select
it, then delete the series to get it out of the way.
2.
Take a look at the plot you made. This shows the mean monthly discharge for February, over the entire period of record. How variable is it? In which year was the highest mean discharge? What was it?
3.
Now let’s compare this data to mean discharges from a summer month. Let’s use August.
Right-click on the chart and Choose Select Data from the drop down menu.
In the “Name” box, rename your series “February data.”
Add a new series, and name it “August data.”
Click on the small button next to the “X values” field. You can scroll down using the bar on the right and select the corresponding years next to the August data. (Alternatively, you can note the cell names at the top and bottom of the range, and type them into the field.)
Do the same thing with the “Y values” field, selecting the August discharge data.
4.
Examine the plot you made. How is August mean discharge different?
Change in time
1.
Can you see any change in time? It may be difficult to see, against such a variable background. We can use a regression line to visualize and measure this change through time.
Right-click on a point from one of your data series, and select “Add trendline.”
In the dialog box that pops up, make sure that you have selected a linear trendline and
that you show the equation and R-squared on the screen.
NOTE: The “
R-squared
” is the fraction of the variance in the Y-axis that can be explained by the X variable. For example, in your plot of Neversink discharge against
time, an R-squared of 0.20 means that 20% of the variance in discharge can be explained by the trend. The other 80% of the variance in discharge would be due to other factors. 2.
Is there a trend? Are August and February the same, or are they different?
B.2: Try the analysis on a new watershed
1.
This analysis tells us something about the drainage basin in Neversink, New York. Now let’s try this again in another location. Select another watershed in the Hydrologic Benchmark Network and repeat the analysis. When you are finished, convene with the rest of your class to share your results.
2.
It is very likely that the watersheds that you analyzed are not the same. These watersheds may be from a variety of geographic regions, with different patterns of precipitation, 5
different precipitation regimes (e.g., rain vs. snow), and different topography and soils. Why do you think the discharge pattern in your area is different from other places? Come
up with a hypothesis to explain some of the differences that you observe. How would you test this hypothesis - what additional data would you need?
Activity C
: Peak discharges and flood hazard
C.1 Extreme discharge events
When we are discussing water as a resource, we are often concerned with scarcity. For example, when water resources are allocated to water users, often the units are in discharge (e.g., cfs). Conversely, when discharge is high, we begin to worry about floods. A flood can be defined in several ways, but a good rule of thumb is: a flood is a discharge that is large enough to overtop the banks of the river channel. If a flood occurs in a populated area, it can endanger lives and structures. Floods are one of the most common types of natural disasters, and can occur in rainforests, deserts, or any landscape in between.
We have already developed most of the skills we need to understand flooding. What is different about flooding? The main difference is that we must focus our attention on extreme events, not average events. Sure, you can calculate that your stream has an average discharge of 1000 cfs, but this information is of no interest to you when a 10,000 cfs peak discharge event is carrying away your house. Thus, in order to understand floods, we will need data on peak discharges. A peak discharge is simply the highest recorded discharge over a period of time, often over one year.
We can use peak data provided by the USGS to ask questions about the likelihood of flood events. Because we are thinking about the interactions between rivers and humans, it makes sense to think about rivers in inhabited areas. Start by navigating to the USGS Water Dashboard,
at https://dashboard.waterdata.usgs.gov/
C.2 Floods on the Mississippi
1.
Add a new worksheet to your Excel spreadsheet, and rename it “Peak data.”
2.
Use the search bar or the map to find the gaging station on the Mississippi River at St. Louis, in Missouri (Monitoring location USGS 07010000). 3.
Find the dropdown menu next to the words “Available data for this site.” Currently this is probably set to “SUMMARY OF ALL AVAILABLE DATA.” Click the menu and select “Peak Streamflow.”
4.
In the Output Formats section, elect "Tab-separated file." Select all, then copy the text to the clipboard. Now, in Excel:
1.
In a new sheet, paste the data, then use "Paste special" in the Paste menu at the upper left and choose “text” to get the data properly into your new Excel worksheet.
6
Your preview ends here
Eager to read complete document? Join bartleby learn and gain access to the full version
- Access to all documents
- Unlimited textbook solutions
- 24/7 expert homework help
2.
Clean up the spreadsheet as you did with the daily data. You should keep the date, gage height, and discharge columns. Data from before 1933 is not reliable, so delete these rows also.
3.
Using the Excel functions AVERAGE() and STDEV(), calculate the mean and standard deviation of discharge for this dataset. To use formulas in Excel, you must first type `=' and that you can refer to cell contents by their position. An example formula to calculate the mean of some numbers in column B between rows 3 and 7 might therefore be: =AVERAGE(B3:B7)
Question 1:
Examine the data. What years are covered by this dataset? Find the maximum discharge on record during this time. When did it occur? What was the discharge in cfs? What was the gage height during this discharge?
Question 2:
What is the mean peak discharge across all of the years (always include units)? What is the standard deviation of peak discharge? Approximately how many standard deviations above the mean PEAK discharge was the highest peak discharge?
Flood probability
We can use peak data such as this to calculate the probability of floods of a particular size. This is how we figure out that a particular discharge is e.g. "a 100-year flood" for a given stream.
Discharge probability
is the probability of experiencing a flood of this size or larger in any year.
Long-term recurrence interval
is a misleading term. It is equal to 1/
p
where p
is the discharge probability. A flood frequency of 100, for example, means that a discharge this high or higher is a "100-year flood," which means roughly that over a the period of record, a flood of this size or larger occurred once per 100 years of record.
Questions 3 and 4:
Think carefully about what causes floods.
3.
If there was a 100-year flood two years ago, how long would you have to wait for the next 100-year flood? (Careful, this is a trick question!)
4.
Since probability and recurrence interval are related, is it more realistic to predict a flood event next year with probability or with recurrence interval?
Make sure you understand these questions before you go on.
Calculating flood probability with ranked data
The simplest method for estimating the probability of flood events uses peak discharges organized by rank - the largest peak discharge on record has rank "1", the second-largest has rank
"2", etc. Use Excel to organize your peak flow data by rank like this:
1.
Copy the all of the Peak Discharge data to a new worksheet (just the discharge). Name the worksheet "Floods."
2.
Now use the "Sort Z to A" function to sort the data from the highest discharge at the top to the lowest at the bottom.
7
3.
Add a new column, named "Rank." In the top row, with the largest flood discharge on record, put a "1" in the Rank column. Then rank the remaining peak discharges. In the Rank column, in the cell just below the "1," type a formula adding one to the cell above it, and hit return. There should now be a "2" in the cell. Fill Down to calculate rank in the
remaining cells.
4.
Now start a new column, called "Probability." For this column, we can calculate the probability of a discharge this large or larger, based on the data. We will calculate this as the total number of peak events this high or higher divided by the total number of events on record, or p
= R
/ (
n
+1) where R
is the rank and n
is the total number of events on record.
Note:
we adjust the denominator by one to correct a systematic bias in this method. The explanation for this adjustment is easiest to see in small datasets. Imagine that we had only one peak discharge on record; our calculated probability of exceeding that peak discharge next year should properly be 1/(1+1) = 50%, indicating complete uncertainty, not 1/1 = 100%. (Obviously,
a dataset containing only one peak discharge is not likely to be very useful to us.)
1.
Use the formula to calculate the discharge probability for each discharge.
2.
Make a new column, and calculate the long-term recurrence interval, or 1/
p
.
3.
Plot your results. Make a new X-Y scatterplot with Recurrence Interval on the X-axis and
Discharge on the Y axis. Make both axes logarithmic.
4.
To improve readability, right-click on the numbers along an axis and select "add minor gridlines."
Question 5: Based on your graph, what discharge is a 50-year flood for this basin? How about a 100-year flood? Which of these estimates are you more confident in?
C.3: Flood probability, changing with time
Up until now, we have calculated flood probability based on the entire period of record. This is sensible because it uses all of the available data. However, this method implicitly assumes that the probability of a particular flood event is consistent over the entire period of record. What if the probability of a 100-year flood changes with time?
In order to address this question, we need to consider some of the things that affect discharge. If something causes discharge to increase or decrease through time, then the probability of large flood events may also increase accordingly.
Urbanization: the problem with pavement
Urbanization (expansion of cities) may seem to have nothing whatsoever to do with rainfall and streamflow. However, urbanization is accompanied by increased impervious surfaces
- surfaces like pavement and rooftops that restrict the flow of water. When rain falls on pavement, it cannot
infiltrate into the ground; instead, it runs off directly into streams. This causes rain from storm events to enter rivers quickly, resulting in rapid changes in discharge. Streams that respond rapidly and dramatically to rain events are called “flashy” streams, and are more likely to flood. Humans affect floods in another important way: we intentionally design and build flood control projects, such as dams, in order to reduce the impact of flooding on populated areas. We certainly hope that such structures reduce the likelihood of flooding!
8
For the next part of this activity, we will see how the likelihood of flooding has changed in three different watersheds near Seattle, Washington. To analyze this data, you will use exactly the same technique you used to assess flood probability on the Mississippi. The only difference is that you will need to divide the peak flow data into two smaller datasets: one dataset containing the early years of data, and a second dataset containing the recent data. By calculating
the size of the 100-year flood (for example) from each dataset, you can compare the results to see
how the watershed has changed through time.
We will use gaging stations from three different watersheds.
1.
Mercer Creek
(USGS 12120000, now “COB MCF” operated by King County, full name
Mercer Creek near Bellevue, Washington): This is the closest of the three gaging stations to downtown Seattle. We have peak streamflow data beginning in 1957. Rapid urbanization began in this area in the late 1970s. Compare the 1957-1977 peak flow data to peak flow data from 1977 to 2019 to detect the effects of urbanization. Note that the USGS stopped gathering data from this site in 2019, and the site was turned over to King County, which is why we will stop our analysis in 2019.
2.
Green River (USGS 12113000, Green River near Auburn, Washington): Gaging on this stream began in 1937. In 1961, the Howard Hanson Dam was built upstream as part of a flood control project on the Green River. Thus, the 1937-1961 peak flow data may be different from data after 1961.
3.
Newaukum Creek
(USGS 12108500, Newaukum Creek near Black Diamond, Washington): This drainage is in a forested semi-rural setting, which has not changed greatly since peak streamflow data begins in 1945. This data will be a useful point of comparison with the other two drainages, as it has neither a history of flood control nor urbanization during the study period.
How do we get the data? We can get the Green River and Newaukum Creek data from the USGS. The easiest way to locate each watershed is to go to the page of Washington data, then select the Statewide Streamflow Table link to access the list of watersheds. Both gaging stations are in the Green/Duwamish River Basin. The Mercer Creek data is no longer hosted by the USGS. Operation of this gaging station is now
done by King County. The data is still available at https://waterdata.usgs.gov/nwis/uv?
site_no=12120000
Or your instructor may provide this data to you if there is a problem with the website.
Your task is to calculate the size of the ten-year flood event (the event with a 10% chance of occurring in any given year). You will make the following comparisons:
1.
Calculate how the size of the 10-year flood event changed after 1977 at the Mercer Creek gaging station. As a non-urbanized “control,” also calculate the same flood event to see how it changed after 1977 at the Newaukum Creek station. (In other words, you will need to calculate flood frequency four times to answer this question!) A good way to answer this question is to calculate the “before” and “after” events, then calculate the percent change.
9
Your preview ends here
Eager to read complete document? Join bartleby learn and gain access to the full version
- Access to all documents
- Unlimited textbook solutions
- 24/7 expert homework help
2.
Do the same analysis, but this time calculate the change in the size of the 10-year flood event
after 1961 at the Green River gage, compared to the Newaukum Creek station. (Don’t bother recalculating the Newaukum Creek data for this very similar time period; the dataset is good enough to use the values you obtained in the last question.)
10