Using USGS streamflow data to design class projects that combine independent and collaborative researchTim Lutz, Department of Geology & Astronomy

West Chester University, West Chester, PA 19383tlutz@wcupa.edu

1. IntroductionStreamflow data have several advantages for engaging

students. 1)Streams are important and visible parts of our dynamic

environment.2)Stream gages are abundant and sample watersheds with a

variety of sizes and differing geologic and land-use characteristics.

3)Basic hydrologic data from stream gages (annual maximum flows and mean daily flows) are available on-line from the USGS and can easily be downloaded and processed in a spreadsheet.

4)The abundance of gages makes it possible to design student projects of some depth that involve both independent and collaborative components.

The two types of projects described in this poster have both been used in classes of up to 24 students at the 200-level, and are suited to any course in which surface water hydrology is a substantial component. The students need to be familiar with basic stream and watershed concepts, and I have used these projects at the end of the semester to provide a culminating experience. The projects have been organized in stages to accommodate short (50-minute) class periods which are used to give instruction, trouble-shoot problems, and discuss results but they could be adapted to longer class or laboratory periods. Students work outside class to complete individual tasks and some group work.

In the first part of a project, each student individually analyzes data from a single gage or watershed. They learn or practice basic skills in Excel such as importing data, applying formulas, creating charts, and adding trend lines. The students assemble and analyze their individual results collectively to discover and discuss new issues that span multiple gages or watersheds. To conclude each project, the results are presented and discussed in class. Each student also submits their individual data and analysis electronically, along with a report on their work and the class’s discussion.

2. Preparation and designI have planned my students’ projects to use flow data from

streams within the geographic region surrounding West Chester, PA– eastern Pennsylvania and adjacent states. Most of our student population is from this area; using data that represent streams that students know well or that makes connections to our regional history of floods and droughts helps to vitalize the project. It is feasible because gaged streams, including those with real-time stations, are fairly abundant. However, other strategies could certainly be pursued, and might be necessary in other parts of the country.

It is important to have a plan to envision the:1) knowledge and skills the project will require2) time span needed to complete it3) outcomes that can be expected and the way that they will be

evaluatedFor the projects described in this poster, students should

understand hydrologic cycle, watershed, and stream concepts along with units and conversions. Excel is used to analyze the data; the skill levels that students bring into the course set the level at which the instructions for the project have to be prescriptive. I usually have some students who have little experience with Excel, and I write fairly explicit instructions.

The time needed for the project depends mostly on the frequency of class meetings and the number of students. My experience is with 50-minute classes that meet three times a week. A good strategy is to allow at least one class meeting between assigning a particular project task, say downloading the data and importing it into Excel: students who discover difficulties by the next class will ask questions and the entire class will benefit before the deadline. I subdivide the entire project into three or four phases, and a minimum of three weeks is required. However, the individual research largely takes place outside of the classroom, so other teaching can occur in parallel.

Outcomes for the individual phase of the project can include some that are fairly specific and that will be reliably achieved by most students. The collaborative outcomes are more sensitive to the direction that the discussion among the students and instructor might take, and what each student takes away from that. I have each student write a report that includes their individual data analysis as well as the insights that came from the collaborative work and discussion.

On this poster sheet, annual maximum flows are used to investigate flow frequencies and the concept of regional regression relationships.

On the right-hand poster sheet, daily streamflow values over decades are used to investigate long-term trends that might originate from changes in land use or climate.

3. Annual maximum flows, flow-frequency analysis, and regional regression analysis.

A. Independent research componentThe objectives of this part of the project are for students to learn how annual maximum series can be used to estimate return flows (e.g., 100-year flood). The main steps that students perform are outlined in this section.

Selecting a gaging station: Students have used real-time data previously in the course; I make the real-time page their starting point for finding data. I have tried both letting students select a stream gage on their own and giving them a list of streams to choose from. If they select their own, I prescribe the geographic region (e.g., Susquehanna or Delaware watershed) and length of record (> 50 years)

Researching the gage: Fundamental information about the station, such as the length of record and the gaged area, are on the real-time page.

Links take students to the station’s home page, a location map, and the peak streamflows that they will download.

Downloading the data: Students follow the link to “Peak streamflow,” display the data in a tab-delimited format, then download the file to their computers:

Importing data into Excel: Students open the file and use the Text Import Wizard to select the data they need and save it as an Excel spreadsheet. They sort the flows and rank them.

3. Annual maximum flows, flow-frequency analysis, and regional regression analysis.

B. Collaborative research component All students working with peak streamflow create a single spreadsheet into which they put all their individual results. A geology major’s computer room facilitates this. Each student also adds the area of their gaged watershed. This spreadsheet is an example of work done in my Environmental Geology class in spring 2003.

Plotting the data and determining return flows: The statistics of fitting a Log-Pearson III or similar distribution to the data are beyond the students; instead, they use a graphical approach and trend-line fitting. The Weibull exceedance probabilities are plotted directly, or they are used to find normal variates [using the Excel function NORMINV()].

Trend lines: Regression fit (R2) and residual patterns are used to select the best model: a power law fit to log(exceedance) vs. log(flow), or an exponential law fit to normal variate vs. log(flow). Censoring of a few anomalous low flows is permitted to improve fit.

y = 1809.2x-0.6122

R2 = 0.9875

1000

10000

100000

0.01 0.10 1.00

Exceedance probability (per year)

Pea

k flo

w (

cfs)

Estimating exceedance flows: The trendline equation is used to calculate the peak flows which are exceeded with annual probabilities of 0.01, 0.02, 0.05, and 0.1 (corresponding to the 100-yr, 50-yr, 20-yr, and 10-yr peak flows). These values are the students individual contribution to the group.

At this point the students can write up their own individual report on their gaging station and their methods.

Peak flow as a function of area: The students work collaboratively in the classroom to understand the implications of their data. It is clear from the table the students create that there is a correlation between area and peak flow. Students are generally impressed that streams which have no close connection except being in the same region will display such a tight pattern of variation: the trendlines for various exceedance probabilities have similar slopes and fairly good fits over several orders of magnitude in area. Students who remember that the mean annual flow is directly proportional watershed area will be surprised to find that this is not so for peak flow:

The mismatch between expectation and reality concerning dependence on area leads to discussion about the differences in the way annual flows and peak flows differ in terms of watershed characteristics.

Regional regression: The graph also allows me to suggest (if it doesn’t come up in discussion) a possible use for such curves: could they be used to predict peak flow in watersheds which are ungaged, or at locations between gaging stations? This is the concept of regional regression of peak flows, which is one way in which station information is made more generally useful. Regression models for Pennsylvania streams are covered in a USGS publication (WRIR 00-4189). Those models include area as the most important factor, but also include other factors. The need for additional factors comes into discussion easily, the students can fairly accurately hypothesize what it makes sense to consider (%forested area, %urban area, %controlled area, %carbonate area).

The discussion of their combined data and regional regression occurs within a class period. Afterwards, the students expand their individual reports to include the group discussion.

The data can be imported starting with the first row of data; headers are added later in Excel.

4. Daily flow values, cumulative deviation charts, and effects of urbanization and climate change.

The objective of this project is for students to investigate changes in streamflow over decades by examining how gaging stations differ. Changes over time might be expected to result from climate fluctuations and from changes in land use, such as urbanization. Gaging stations can be selected to sample potential differences in response, for example, by selecting gages in urban and non-urban areas. Using daily values allows variability on all scales to be seen; however, thirty years of data correspond to nearly 11,000 values. An Excel file containing these data, along with formulas and charts, will well exceed the capacity of a 1.44 Mb diskette. Therefore, it is important that students have access to machines with hard drives or with CD drives. My students also have access to a Blackboard site for their course and can upload and store files on the Blackboard server (Digital Drop Box). Although the data are numerous, fewer statistical and Excel skills are needed for this project than for the annual maximum project on the left-hand poster board.

A.Independent research componentThe objectives for this part of the project are for students to learn

how to construct cumulative deviation from the mean charts, which are a very direct way of visualizing changes in stream flow over time. The main steps that students perform are outlined in this section.

Selecting a gaging station: Students have used real-time data previously in the course; I make the real-time page their starting point for finding data. For this project there should be no significant gaps in the data and the record also has to be the same length for all stations; I used 52 years of data (water years 1950-2001) with my last class. I screen stations in advance for students to use, but I have them locate their gage and download the data:

Researching the gage: Fundamental information about the station, such as the gaged area, are on the real-time page:

Links to the station’s home page, a location map, and the daily streamflows they will download.

Selecting the data: Students follow the link to “Daily Streamflow,” select the time-span that all students are using, and download in a tab-delimited format:

Importing the data into Excel: Students follow the link to “Daily Streamflow,” display the data in tab-delimited format, then download the file to their computers. The Text Import Wizard makes it easy to import the data:

Determining cumulative deviations: Students use Excel formulas to find the cumulative amount of water that flowed each day and the cumulative amount based on the mean flow over the entire 52-year period. The difference between these is the cumulative deviation from the mean. This value is recalculated from cubic feet or acre-feet to the equivalent rainfall per square mile of watershed (inches/sq. mi.). This is a critical step since it allows different students to compare their results on the same scale:

Graphing the deviations: A chart of the deviations over time shows periods of lower than average streamflow by negative slopes and greater than average flow by positive slopes. In this example, flow was lower than average over a two-decade span, then was generally higher than average until the present. Such graphs immediately raise questions: what happened to change the situation about 1970? A change in climate? A change in land use?

4.B. Collaborative research component

Sharing data: To help answer their questions students share their data to create a single file; only the cumulative deviations (inches/sq.mi.) are transmitted to reduce file sizes. I also give students data on the variation in precipitation rate over the same time range. Students then compare graphs to answer questions such as:

Do all streams show the same patterns?Do small streams and large streams differ?Are streams in urbanized areas different from streams in forested areas?How do streamflows compare to precipitation?

Students create charts which combine data selected from different locations to answer these questions. This chart compares streams of quite different watershed areas

Closure: In class, students complete their experience by discussing their charts in terms of their understanding of hydrologic cycle and watershed processes. They write a final report which includes their individual work on their stream station and their collaborative discussion of everyone’s work.