Climate Change Indicators in the United States Technical Documentation Streamflow

Streamflow

Identification

1. Indicator Description

This indicator describes trends in the magnitude and timing of streamflow in rivers and streams across
the United States. Streamflow is a useful indicator of climate change for several reasons. More
precipitation is expected to cause higher average streamflow in some places, while heavier storms could
lead to higher peak flows. More frequent or severe droughts may reduce streamflow in some areas.
Changes in the amount of snowpack and earlier spring melting can alter the size and timing of peak
streamflows.

Components of this indicator include trends in four annual flow statistics from 1940 through 2014:

• Magnitude of annual seven-day low streamflow (Figure 1).

• Magnitude of annual three-day high streamflow (Figure 2).

• Magnitude of annual mean streamflow (Figure 3).

• Timing of winter-spring center-of-volume date (Figure 4).

2. Revision History

December 2012: Indicator published.

May 2014: Updated indicator with data through 2012. Added Figure 3 to show annual

mean streamflow and renumbered original Figure 3 (winter-spring center of
volume) as Figure 4.

August 2016: Updated indicator with data through 2014.

Data Sources

3. Data Sources

Mike Kolian of EPA developed this indicator in partnership with Michael McHale, Robert Dudley, and
Glenn Hodgkins at the U.S. Geological Survey (USGS). The indicator is based on streamflow data from a
set of reference stream gauges specified in the Geospatial Attributes of Gages for Evaluating Streamflow
(GAGES-II) database, which was developed by USGS and is described in Lins (2012). Daily mean
streamflow data are stored in the USGS National Water Information System (NWIS).

4. Data Availability

EPA obtained the data for this indicator from Michael McHale, Robert Dudley, and Glenn Hodgkins at
USGS. Similar streamflow analyses had been previously published in the peer-reviewed literature (Burns
et al., 2007; Hodgkins and Dudley, 2006). USGS provided a reprocessed data set to include streamflow
trends through water year 2014.

Technical Documentation: Streamflow

-------
Streamflow data from individual stations are publicly available online through the surface water section
of NWIS at: http://waterdata.usgs.gov/nwis/sw. Reference status and watershed, site characteristics,
and other metadata for each stream gauge in the GAGES-II database are available online at:
http://water.usgs.gov/GIS/metadata/usgswrd/XML/gagesll Sept2011.xml.

Methodology

5. Data Collection

Streamflow is determined from data collected at stream gauging stations by automated devices that
record the elevation (or stage) of a river or stream at regular intervals each day. Intervals vary from
station to station—typically every 15 minutes to one hour. USGS maintains a national network of stream
gauging stations, including more than 7,000 stations currently in operation throughout the United States
(http://pubs.usgs.gov/fs/1995/0066/report.pdf). USGS has been collecting stream gauge data since the
late 1800s at some locations. Gauges generally are sited to record flows for specific management goals
or legal mandates, typically in cooperation with municipal, state, and federal agencies.

Streamflow (or discharge) is measured manually at regular intervals by USGS personnel (typically every
four to eight weeks). The relation between stream stage and discharge is determined, and a stage-
discharge relation (rating) is developed to calculate streamflow for each recorded stream stage (Rantz et
al., 1982). These data are used to calculate the daily mean discharge for each day at each site. All
measurements are made according to standard USGS procedures (Rantz et al., 1982; Sauer and
Turnipseed, 2010; Turnipseed and Sauer, 2010).

This indicator uses data from USGS stream gauges that are part of the Hydro-Climatic Data Network
2009 (HCDN-2009) network (Lins, 2012), which is a subset of the USGS GAGES-II network. HCDN-2009
gauges have been carefully selected to reflect minimal interference from human activities such as dam
construction, reservoir management, wastewater treatment discharge, water withdrawal, and changes
in land cover and land use that might influence runoff. The subset of HCDN-2009 gauges was further
examined on the basis of length of period of record (75 years) and completeness of record (greater than
or equal to 80 percent for every decade). Figures 1, 2, and 3 are based on data from 192 stream gauges.
Figure 4 relies on 56 stream gauges largely because it is limited to stream basins that receive 30 percent
or more of their total annual precipitation in the form of snow. This additional criterion was applied
because the metric in Figure 4 is used primarily to examine the timing of winter-spring runoff, which is
substantially affected by snowmelt-related runoff in areas with a large annual snowpack. All of the
selected stations and their corresponding basins are relatively independent—that is, the analysis does
not include stations with watershed areas that overlap by more than 30 percent.

All watershed characteristics, including basin area, station latitude and longitude, and percentage of
precipitation as snow were taken from the GAGES-II database. GAGES-II basin area was determined
through EPA's National Hydrography Dataset Plus and supplemented by the USGS National Water-
Quality Assessment Program and the USGS Elevation Derivatives for National Applications.

Technical Documentation: Streamflow

-------
6. Indicator Derivation

Figures 1, 2, and 3. Seven-Day Low (Figure 1), Three-Day High (Figure 2), and Annual Average (Figure 3)
Streamflow in the United States, 1940-2014

Figure 1 shows trends in low-flow conditions using seven-day low streamflow, which is the lowest
average of seven consecutive days of streamflow in a year. Hydrologists commonly use this measure
because it reflects sustained dry or frozen conditions that result in the lowest flows of the year. Seven-
day low flow can equal zero if a stream has dried up completely.

Figure 2 shows trends in very wet conditions using three-day high streamflow, which is the highest
average of three consecutive days of streamflow in a year. Hydrologists use this measure because a
three-day averaging period has been shown to effectively characterize runoff associated with large
storms and peak snowmelt over a diverse range of watershed areas.

Figure 3 shows trends in average conditions using annual average streamflow, which is the average of all
daily mean streamflow values for a given year.

Rates of change from 1940 to 2014 at each station on the maps in Figures 1-3 were computed using the
Sen slope, which is the median of all possible pair-wise slopes in a temporal data set (Helsel and Hirsch,
2002). The Sen slope was then multiplied by the number of years in the period (74 years: the last year
minus the starting year) to estimate total change over time. Trends are reported as percentage
increases or decreases, computed from the value for the last year of the regression line relative to the
value of the first year of the regression line.

Figure 4. Timing of Winter-Spring Runoff in the United States, 1940-2014

Figure 4 shows trends in the timing of streamflow in the winter and spring, which is influenced by the
timing of snowmelt runoff in areas with substantial annual snowpack. The timing of streamflow also can
be influenced by the ratio of winter rain to snow and by changes in the seasonal distribution of
precipitation. Figure 4 shows trends in the winter-spring center-of-volume (WSCV) date, which is
defined for this indicator as the date when half of the total streamflow between January 1 and July 31
for sites in the western United States, or half of the total streamflow between January 1 and May 31 for
sites in the eastern United States, has passed by the gauging station. These regionally different time
periods were selected to best represent the period during which snowmelt occurs in each region, based
on snowmelt-related streamflow interpreted from long-term average seasonal hydrographs. Trends in
this date are computed in the same manner as the other three components of this indicator, and the
results are reported in terms of the number of days earlier or later that the WSCV date is occurring. For
more information about WSCV-date methods, see Hodgkins and Dudley (2006) and Burns et al. (2007).

Indicator Development

For the 2016 update to this indicator, EPA and USGS adjusted the analysis in two ways:

• The analysis of all four metrics changed from calendar years (January 1 to December 31) to
water years (October 1 to September 30). Water years are defined such that "water year 2014"
runs from October 2013 through September 2014.

Technical Documentation: Streamflow

-------
• The seasonal window for the timing of winter-spring flows changed from January through June
for all gauges to January through July for western gages and January through May for eastern
gauges.

EPA and USGS made these adjustments to harmonize this indicator with other publications based on the
same data and to reflect best practices for characterizing WSCV.

7. Quality Assurance and Quality Control

Quality assurance and quality control (QA/QC) procedures are documented for measuring stream stage
(Sauer and Turnipseed, 2010), measuring stream discharge (Turnipseed and Sauer, 2010), and
computing stream discharge (Sauer, 2002; Rantz et al., 1982). Stream discharge is typically measured
and equipment is inspected at each gauging station every four to eight weeks. The relation between
stream stage and stream discharge is evaluated following each discharge measurement at each site, and
shifts to the relation are made if necessary.

The GAGES-II database incorporated a QC procedure for delineating the watershed boundaries acquired
from the National Hydrography Dataset Plus. The data set was cross-checked against information from
USGS's National Water-Quality Assessment Program. Basin boundaries that were inconsistent across
sources were visually compared and manually delineated based on geographical information provided in
USGS's Elevation Derivatives for National Applications. Other screening and data quality issues are
addressed in the GAGES-II metadata available at:

https://water.usgs.gov/GIS/metadata/usgswrd/XML/gagesll Sept2011.xml.

Analysis

8. Comparability Over Time and Space

All USGS streamflow data have been collected and extensively quality-assured by USGS since the start of
data collection. Consistent and well-documented procedures have been used for the entire periods of
recorded streamflows at all gauges (Corbett et al., 1943; Rantz et al., 1982; Sauer, 2002).

Trends in streamflow over time can be strongly influenced by human activities upstream, such as the
construction and operation of dams, flow diversions and abstractions, and land-use change. To remove
these non-climatic influences to the extent possible, this indicator relies on a set of reference gauges
that were chosen because they represent least-disturbed (though not necessarily completely
undisturbed) watersheds. The criteria for selecting reference gauges vary from region to region based
on land use characteristics. Therefore, a modestly impacted watershed in one part of the country (e.g.,
an area with agricultural land use) might not have met the data quality standards for another less
impacted region. The reference gauge screening process is described in Lins (2012) and is available in
the GAGES-II metadata at: https://water.usgs.gov/GIS/metadata/usgswrd/XML/gagesll Sept2011.xml.

Analytical methods have been applied consistently over time and space.

Technical Documentation: Streamflow

-------
9. Data Limitations

Factors that may impact the confidence, application, or conclusions drawn from this indicator are as
follows:

1. This analysis is restricted to locations where streamflow is not highly disturbed by human
influences, including reservoir regulation, diversions, and land cover change. Changes in land
cover and land use over time, however, could still influence trends in the magnitude and timing
of streamflow at some sites.

2. Reference gauges used for this indicator are not evenly distributed throughout the United
States, nor are they evenly distributed with respect to topography, geology, elevation, or land
cover, due to variations in the availability of long-term data.

3. Some streams in northern or mountainous areas have their lowest flows in the winter due to
water being held in snow or ice for extended periods. At these sites, low flow trends could be
influenced by climate factors other than reduced precipitation or otherwise dry conditions, such
as long stretches of cold weather.

10. Sources of Uncertainty

Uncertainty estimates are not available for this indicator as a whole. As for the underlying data, the
precision of individual stream gauges varies from site to site. Accuracy depends primarily on the stability
of the stage-discharge relationship, the frequency and reliability of stage and discharge measurements,
and the presence of special conditions such as ice (Novak, 1985). USGS has published a general online
reference devoted to the calculation of error in individual stream discharge measurements (Sauer and
Meyer, 1992).

11. Sources of Variability

Streamflow can be highly variable over time, depending on the size of the watershed and the factors
that influence flow at a gauge. USGS addresses this variability by recording stream stage many times a
day (typically 15-minute to one-hour intervals) and then computing a daily average streamflow.
Streamflow also varies from year to year as a result of variation in precipitation and air temperature.
Trend magnitudes computed from Sen slopes provide a robust estimate of linear changes over a period
of record, and thus this indicator does not measure decadal cycles or interannual variability in the metric
over the time period examined.

While gauges are chosen to represent drainage basins relatively unimpacted by human disturbance,
some sites may be more affected by direct human influences (such as land-cover and land-use change)
than others. Other sources of variability include localized factors such as topography, geology, elevation,
and natural land cover. Changes in land cover and land use over time can contribute to streamflow
trends, though careful selection of reference gauges strives to minimize these impacts.

Although the WSCV date is driven by the timing of the bulk of snow melt in areas with substantial annual
snowpack, other factors also will influence the WSCV date. For instance, a heavy rain event in the winter
could result in large volumes of water that shift the timing of the center of volume earlier. Changes over

Technical Documentation: Streamflow

-------
time in the distribution of rainfall during the WSCV measurement period could also affect the WSCV
date.

12. Statistical/Trend Analysis

The maps in Figures 1, 2, 3, and 4 all show trends through time that have been computed for each
gauging station using a Sen slope analysis by USGS. Because of uncertainties and complexities in the
interpretation of statistical significance, particularly related to the issue of long-term persistence (Cohn
and Lins, 2005; Koutsoyiannis and Montanari, 2007), significance of trends is not reported. Future
updates of this indicator will likely also include details on the significance of the reported trends.

References

Burns, D.A., J. Klaus, and M.R. McHale. 2007. Recent climate trends and implications for water resources
in the Catskill Mountain region, New York, USA. J. Hydrol. 336(1-2):155-170.

Cohn, T.A., and H.F. Lins. 2005. Nature's style: Naturally trendy. Geophys. Res. Lett. 32:L23402.

Corbett, D.M., et al. 1943. Stream-gaging procedure: A manual describing methods and practices of the
Geological Survey. U.S. Geological Survey Water-Supply Paper 888.
https://pubs.er.usgs.gov/publication/wsp888.

Helsel, D.R., and R.M. Hirsch. 2002. Statistical methods in water resources. Techniques of water
resources investigations, Book 4. Chap. A3. U.S. Geological Survey, https://pubs.usgs.gov/twri/twri4a3.

Hodgkins, G.A., and R.W. Dudley. 2006. Changes in the timing of winter-spring streamflows in eastern
North America, 1913-2002. Geophys. Res. Lett. 33:L06402.
https://water.usgs.gov/climate water/hodgkins dudley 2006b.pdf.

Koutsoyiannis, D., and A. Montanari. 2007. Statistical analysis of hydroclimatic time series: Uncertainty
and insights. Water Resour. Res. 43(5):W05429.

Lins, H.F. 2012. USGS Hydro-Climatic Data Network 2009 (HCDN-2009). U.S. Geological Survey Fact Sheet
2012-3047. https://pubs.usgs.gov/fs/2012/3047.

Novak, C.E. 1985. WRD data reports preparation guide. U.S. Geological Survey Open-File Report 85-480.
https://pubs.er.usgs.gov/publication/ofr85480.

Rantz, S.E., et al. 1982. Measurement and computation of streamflow. Volume 1: Measurement of stage
and discharge. Volume 2: Computation of discharge. U.S. Geological Survey Water Supply Paper 2175.
https://pubs.usgs.gov/wsp/wsp2175.

Sauer, V.B. 2002. Standards for the analysis and processing of surface-water data and information using
electronic methods. U.S. Geological Survey Water-Resources Investigations Report 01-4044.
https://pubs.er.usgs.gov/publication/wri20014044.

Technical Documentation: Streamflow

-------
Sauer, V.B., and R.W. Meyer. 1992. Determination of error in individual discharge measurements. U.S.
Geological Survey Open-File Report 92-144. https://pubs.usgs.gov/of/1992/ofr92-144.

Sauer, V.B., and D.P. Turnipseed. 2010. Stage measurement at gaging stations. U.S. Geological Survey
Techniques and Methods, Book 3. Chap. A7. U.S. Geological Survey, https://pubs.usgs.gov/tm/tm3-a7.

Turnipseed, D.P., and V.B. Sauer. 2010. Discharge measurements at gaging stations. U.S. Geological
Survey Techniques and Methods, Book 3. Chap. A8. U.S. Geological Survey.
https://pubs.usgs.gov/tm/tm3-a8.

Technical Documentation: Streamflow

7

-------