Annual and Seasonal Trends in Discharge of National Capital Region Streams

National Park Service U.S. Department of the Interior

Natural Resource Stewardship and Science

Natural Resource Technical Report NPS/NCRN/NRTR—2011/488

ON THE COVER Potomac River near Paw Paw, West Virginia Photograph by: Tom Paradis, NPS.

Annual and Seasonal Trends in Discharge of National Capital Region Streams

Natural Resource Technical Report NPS/NCRN/NRTR—2011/488

John Paul Schmit

National Park Service Center for Urban Ecology 4598 MacArthur Blvd. NW Washington, DC 20007

September, 2011

U.S. Department of the Interior National Park Service Natural Resource Stewardship and Science Fort Collins, Colorado

The National Park Service, Natural Resource Stewardship and Science office in Fort Collins, Colorado publishes a range of reports that address natural resource topics of interest and applicability to a broad audience in the National Park Service and others in natural resource management, including scientists, conservation and environmental constituencies, and the public.

The Natural Resource Technical Report Series is used to disseminate results of scientific studies in the physical, biological, and social sciences for both the advancement of science and the achievement of the National Park Service mission. The series provides contributors with a forum for displaying comprehensive data that are often deleted from journals because of page limitations.

All manuscripts in the series receive the appropriate level of peer review to ensure that the information is scientifically credible, technically accurate, appropriately written for the intended audience, and designed and published in a professional manner.

This report received formal peer review by subject-matter experts who were not directly involved in the collection, analysis, or reporting of the data, and whose background and expertise put them on par technically and scientifically with the authors of the information.

Views, statements, findings, conclusions, recommendations, and data in this report are those of the author(s) and do not necessarily reflect views and policies of the National Park Service, U.S. Department of the Interior. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by the National Park Service.

This report is available (http://www.nature.nps.gov/publications/nrpm/).

Please cite this publication as:

Schmit, J. P. 2011. Annual and seasonal trends in discharge of National Capital Region streams. Natural Resource Technical Report NPS/NCRN/NRTR—2011/488. National Park Service, Fort Collins, Colorado.

NPS 800/109866, September 2011 ii

Contents

Page

Figures...... v

Tables ...... vii

Executive Summary ...... ix

Acknowledgments...... x

Introduction ...... 1

Factors Influencing Discharge ...... 1

Precipitation ...... 1

Change in Land Use ...... 2

Direct Human Alterations to Streams ...... 3

Scope of this Report ...... 4

Methods...... 5

Data Sources ...... 5

Discharge Data ...... 5

Precipitation data ...... 5

Analysis ...... 6

Results ...... 9

Precipitation ...... 9

Discharge ...... 13

Seasonal Pattern ...... 13

Annual Trends ...... 16

Seasonal Trends ...... 19

Antietam Creek ...... 20

Potomac at Cumberland ...... 21

iii

Contents (continued)

Page

Potomac at Paw Paw ...... 22

Potomac at Hancock ...... 23

South Quantico Creek ...... 24

Difficult Run ...... 25

Rock Creek ...... 26

North East Branch of the Anacostia River ...... 27

Piscataway Creek ...... 28

Potomac at Point of Rocks ...... 29

Potomac at Little Falls ...... 30

Monocacy River ...... 31

Shenandoah River ...... 32

North West Branch of the Anacostia River ...... 33

Discussion ...... 35

Precipitation ...... 35

Discharge ...... 35

Rural Streams ...... 36

Urban Streams ...... 36

Large Rivers ...... 37

Northwest Branch of the Anacostia ...... 37

Limitations of this Study ...... 37

Implications ...... 38

Literature Cited ...... 39

Figures

Page

Figure 1. Map of gaging stations used in this study...... 6

Figure 2. Mean precipitation for Maryland, Virginia, West Virginia and Washington DC...... 9

Figure 3. Annual Precipitation for Maryland, Virginia, West Virginia and the District of Columbia...... 10

Figure 4. Seasonal change in precipitation in Maryland, Virginia, West Virginia and the District of Columbia...... 12

Figure 5. Comparison of older and more recent precipitation in MD, VA, WV, and DC by season...... 13

Figure 6. Monthly median discharge rates for sixteen stream gages in the NCRN...... 15

Figure 7. Trends for gages with increasing discharge at low percentiles...... 16

Figure 8. Trends for gages with increasing discharge at high percentiles...... 17

Figure 9. Trends for gages with no significant trends...... 18

Figure 10. Trends for the stream gage on the Northwest Branch of the Anacostia River...... 19

Figure 11. Seasonal trends for the stream gage on Antietam Creek at Sharpsburg ...... 20

Figure 12. Seasonal trends for the stream gage on the Potomac River at Cumberland ...... 21

Figure 13. Seasonal trends for the stream gage on the Potomac River at Paw Paw ...... 22

Figure 14. Seasonal trends for the stream gage on the Potomac River at Hancock ...... 23

Figure 15. Seasonal trends for the stream gage on South Quantico Creek ...... 24

Figure 16. Seasonal trends for the stream gage on Difficult Run ...... 25

Figure 17. Seasonal trends for the stream gage on Rock Creek ...... 26

Figure 18. Seasonal trends for the stream gage on the North East Branch of the Anacostia River ...... 27

Figure 19. Seasonal trends for the stream gage on Piscataway Creek ...... 28

Figures (continued)

Page

Figure 20. Seasonal trends for the stream gage on the Potomac River at Point of Rocks ...... 29

Figure 21. Seasonal trends for the stream gage on the Potomac River at Little Falls ...... 30

Figure 22. Seasonal trends for the stream gage on the Monocacy River at Jug Bridge ...... 31

Figure 23. Seasonal trends for the stream gage on the Shenandoah River at Millville ...... 32

Figure 24. Seasonal trends for the stream gage on the North West Branch of the Anacostia River ...... 33

Tables

Page

Table 1. Gaging stations used in this analysis...... 5

Table 2. Trends in precipitation in the National Capital Region Network by Season ...... 11

vii

Executive Summary

The following report examines long term precipitation and stream discharge data relevant to the parks of the National Capital Region Network (NCRN). Precipitation and stream flow data has been compiled from public agencies. Precipitation data covers all of the parks in the NCRN, but streamflow data is not available for Catoctin, Manassas or Wolf Trap. The analysis shows that there has been a long term increase in fall precipitation in the NCRN. Based on this observation, it would be expected that stream discharge would also be increasing in the fall months. In the more rural areas of the region, this prediction was borne out. However, in urban areas, and in areas where the streams are used for hydropower and municipal water, it is not. In urban areas, seasonal trends are muted, but discharge rates are becoming more variable, with higher highs and lower lows. In areas where water is diverted for municipal or hydroelectric use, there was no trend in discharge. An important conclusion from this report is that human alteration of the environment can interact with climate changes to determine trends in discharge. Streams that are geographically close, but have differing land use in their watersheds, may have very different trends in stream discharge.

Acknowledgments

I would like to thank Mark Lehman for his help with GIS and data retrieval. I would also like to thank J. Patrick Campbell, Jim Comiskey, Marion Norris, Megan Nortrup and Penelope Pooler for help comments.

Introduction

In recent years there has been increased attention to a variety of potential impacts of climate change. National Parks were established to protect various natural and cultural resources and are expected to develop strategies to adapt to the changing climate to carry out their mission. To adapt, park managers must first understand how the local climate is changing and how this impacts park resources.

To further that goal, this report looks at historical changes in precipitation and stream discharge in National Capital Region Network (NCRN) parks. There are several reasons that precipitation and discharge were chosen for analysis. First, both of these have already been selected as “vital signs” for the NCRN. Vital signs are a small set of indicators that are meant to represent the overall condition of the parks. Precipitation is part of the “Weather” vital sign and stream discharge is part of the “Surface Water Dynamics” vital sign. Secondly, long term data for the NCRN is available for each of these measures. Finally there is a link between these measurements and climate. Precipitation is one of the factors which make up the climate and long term trends in precipitation could have important impacts on the parks. Precipitation is the ultimate source of stream discharge, therefore long term precipitation trends could result in trends in discharge.

Factors Influencing Discharge Discharge is a measure of the amount of water in a stream. It is measured as the volume of water passing a particular location at a given point in time. In the United States, discharge is traditionally measured as cubic feet of water passing a location in one second, abbreviated as “cfs”.

Stream discharge is considered to be important for a number of reasons. It influences the distribution of sediments and nutrients in the water which in turn influences the distribution of stream dwelling species (Vannote et al. 1980, Vannote and Minshall 1982, Statzner et al. 1988). It influences the floodplain dynamics of streams (Leopold et al. 1964, Gore 1996), which is important in providing habitat for flora and fauna. Finally discharge is a measure of water available for human use, such as irrigation, drinking water, or hydropower generation.

The amount of discharge from a stream is determined by a variety of factors. For the purpose of this report, I focus on three factors which are clearly relevant for the parks in the NCRN: amount of precipitation, land use, and human alteration for municipal or other water use.

Precipitation In the NCRN, precipitation is the source of water in streams. Water enters the streams either directly from runoff or via groundwater. Unlike many areas in the western US, typically there is no significant build-up of ice and snow in the winter months, and therefore there is little melt water added to streams in the spring and summer.

The amount of precipitation in a watershed can change both seasonally and over the long term. Long term change in precipitation has been hypothesized as an effect of global warming (Lins and Cohn 2003, Huntington 2006). One hypothesis is that warming of the oceans will result in greater evaporation, which in turn will increase cloud cover and eventually increase

precipitation. This hypothesis has been dubbed the “acceleration of the hydrologic cycle” as it will decrease the amount of time it takes for an individual water molecule to complete the cycle from ocean to rain and back to ocean.

The effect of climate change on discharge is uncertain under this scenario. On the one hand, the increase in precipitation could lead to an increase in stream discharge, and possibly increased flooding (Huntington 2006, Krakauer and Fung 2009). On the other hand, increased temperatures may also bring increased evapotranspiration which would reduce discharge levels (Burns et al. 2007, Krakauer and Fung 2009, Rose 2009). However, recent work (Gedney et al. 2006) demonstrated that increased CO2 concentrations can result in a decrease in plant transpiration. Plants lose water when stomata are open to allow for uptake in CO2, but increasing CO2 concentration reduces stomata opening and thereby reduces water losses. If transpiration is reduced, then plants will require less water, which could result in an increase in discharge.

Empirical studies of historic precipitation and discharge data have shown that at least some parts of the US are getting more rain but not an increase in flooding (Lins and Slack 1999, McCabe and Wolock 2002, Lins and Cohn 2003, Lins and Slack 2005, Krakauer and Fung 2009) however Groisman et al. (2001 and 2004) found evidence of increased high discharge events. Notably, Lins and Slack (1999) found evidence that while there has been no change in discharge when water levels are high; the volume of discharge when water levels are low is increasing in some parts of the eastern US. Or, to put it another way, while there has been no increase in flooding, streams no longer drop as low as they once did.

This result was somewhat unexpected as it is not obvious how there can be more water in streams when they are low, but not when they are high. The apparent paradox was resolved by Small et al. (2006). They analyzed the precipitation and discharge data separately for each season to detect patterns that would be lost in the analysis of annual data. They demonstrated that the increases in both precipitation and discharge are mainly occurring in the fall months. The fall is the time of year when discharge levels are at their lowest in the eastern US. Therefore, increased fall precipitation is raising these low levels of discharge, but having no effect on high discharge levels which generally occur in the spring.

In general, the previous studies examining discharge made use of data from the USGS Hydroclimatologic Data Network (HCDN - Slack and Landwehr 1992). This is a dataset developed by the USGS specifically for climatological studies. The stream gages included in the dataset are those that are not affected by alterations such as dams or municipal water withdrawals and discharges. Additionally, the land in these drainages is primarily natural or agricultural and has remained relatively unchanged for the period of record, typically many decades. Streams in the NCRN however, are subject to dams and municipal withdrawals and discharges. Additionally much of the land around the streams in the NCRN has been urbanized. As a consequence of these factors, it is not clear if the results reported in previous studies would be borne out in the NCRN.

Change in Land Use Change from one land use to another can change stream discharge. Clearing forest for agricultural use can reduce soil infiltration and evapotranspiration, thereby increasing runoff and “flashiness” (reviewed in Allan 2004). “Flashiness” refers to high discharge immediately after a

storm that quickly subsides. Areas that are urbanized have a high level of impervious surface and low soil infiltration. This leads to high flashiness and low stream discharge. The low discharge is due to the fact that precipitation is blocked from entering the groundwater (reviewed in Allan 2004).

Direct Human Alterations to Streams Human alterations can change stream discharge. These alterations can include removing water for irrigation, human consumption, or other use or adding water at the outflows of water treatment plants. Rivers in the NCRN, particularly the Potomac, Monocacy, and Shenandoah Rivers, are used for municipal water supply.

Although many communities remove water from the Potomac, the Washington DC metropolitan area is by far the largest. The Washington metropolitan area is served by three major water suppliers: the Washington Aqueduct, the Washington Suburban Sanitary Commission, and the Fairfax County Water Authority. In 2008, 76% of the water provided by these three suppliers came from the Potomac, with the balance coming from the Occoquan and Patuxent Rivers (Ahmed et al. 2010). In 2008, on average, 354 mgd (million gallons per day) were removed from the Potomac by these water suppliers. This is equivalent to ~548 cfs (cubic feet per second), the units in which stream discharge is conventionally measured. Water production is highest in July, August, and September. During these months, water use can exceed over 600 mgd. If 76% of this comes from the Potomac, then summer usage can exceed 450 mgd, or 696 cfs (Ahmed et al 2010).

The city of Frederick draws over 98% of its municipal water from the Monocacy River and its tributaries. In 2009, the city produced 5.56 mgd, of which ~4.48 mgd (8.5 cfs) was from the Monocacy River (City of Frederick 2010). Although seasonal data is not readily available, it is likely that demand for water is greatest in the late summer and early fall, as seen in the Washington DC area.

The Shenandoah River and its tributaries are the source of municipal water for a number of small communities along its length. Due to the number of water systems involved, it is difficult to estimate the amount of water removed annually for water use or the amount returned via waste water treatment.

Dams alter discharge by storing and later releasing water. In particular, it has been shown that dams reduce peak discharge, increase minimum discharge, and reduce flashiness and variability of discharge (Poff et al. 2006). On the Potomac River, there is a dam with a controlled spillway associated with the large Randolph Jennings Reservoir, upstream from Cumberland Maryland. Also on the Potomac are the Little Falls dam in Montgomery County and Dam 4 and Dam 5, which are run of river hydroelectric dams in Washington Country. These three dams are “run of river dams” (water is not stored in a reservoir behind the dam).

The Shenandoah River and its North and South Fork tributaries have several dams on them. These include the Luray, Chapman, Burnshire, Newport, and Shenandoah dams. These run of river dams are used for hydroelectric power generation.

Linganore Creek, a tributary of the Monocacy River, has a dam that creates Lake Linganore which serves as a reservoir for the city of Frederick

Scope of this Report In recognition of the importance of discharge and other stream characteristics, “Surface Water Dynamics” was selected as a “vital sign” for the National Capital Region Network Inventory & Monitoring program (NCRN). Measurements of stream flow and discharge are made monthly when stream conditions allow at 41 streams throughout the NCRN parks (Norris et al. 2007). These monthly measurements can be combined with other data to determine the amount of nutrients, such as nitrogen and phosphorus, in the water. However, single monthly measurements are less suited to discovering long term trends in stream discharge.

In order to determine long term trends in discharge, I analyzed existing data collected by the U.S. Geological Survey (USGS). Starting in 1889, the USGS established a network of stream gaging stations which record stream levels and discharge, as well as other stream data. Data from USGS gaging stations has previously been used to determine trends in steam discharge (Lettenmaier et al. 1994, Lins and Slack 1999, 2005, Groisman et al. 2001, 2004, McCabe and Wolock 2002, Poff et al. 2006, Small 2006, Kalra et al. 2008, Krakauer and Fung 2009, Rose 2009). This report focuses solely on stream gages that are in NCRN parks, or are near to a park and are on streams that flow through NCRN parks. Due to the numerous land use changes and alterations to stream flow, these streams are not ideal for detecting the effects of climate change in general. They are, however, the streams that are managed by, or exert influence on, NCRN parks.

In this report I examine the data to determine if the previously documented precipitation and discharge patterns are also found in streams in the NCRN parks. Furthermore, the data is examined to see if there are any trends in stream discharge that may be related to urbanization effects or to alterations related to municipal water use.

Methods

Data Sources

Discharge Data Discharge data is from US Geological Survey gaging stations located in or near the parks in the NCRN (Figure 1) and was downloaded from the USGS National Water Information System website (http://waterdata.usgs.gov). Stations were selected if they were located within a park or were near to a park on a stream that went through the park. Only stations that are currently active were included. These criteria resulted in sixteen gaging stations (Table 1). Two stations, Piney Run and Watts Branch, have less than 20 years of continuous data. Due to the relatively short time span of data available, these stations were not considered in some analyses because the shorter time frame was not informative. These sixteen stations are not part of the USGS Hydrologic Climate Data Network.

Table 1. Gaging stations used in this analysis. Gaging Station State Park USGS Station Date of earliest Mean ID # data Discharge (cfs) Antietam Creek MD ANTI 01619500 Jan 1, 1929 208.0 Anacostia NE Branch at MD NACE 01649500 Aug 1, 1938 44.0 Riverdale Anacostia NW Branch at MD NACE 01651000 Jul 1, 1938 24.0 Hyattsville Difficult Run VA GWMP 01646000 Apr 1, 1935 38.0 Monocacy at Jug Bridge MD MONO 01643000 Jan 1, 1930 476.0 Piney Run near Lovettsville WV HAFE 01636690 Oct 1, 2001 7.8 Piscataway Creek MD NACE 01653600 Oct 1, 1965 24.0 Potomac N Branch at MD/WV CHOH 01603000 May 24, 1929 684.0 Cumberland Potomac at Hancock MD/WV CHOH 01613000 Oct 1, 1932 2200.0 Potomac at Little Falls MD/VA CHOH/GWMP 01646500 Mar 1, 1930 6540.0 Potomac at Point of Rocks MD/VA CHOH 01638500 Feb 1, 1895 5420.0 Potomac at Paw Paw MD/WV CHOH 01610000 Oct 1, 1938 1820.0 Shenandoah at Millville WV HAFE 01636500 Jan 1, 1929 1600.0 Quantico, S Branch VA PRWI 01658500 May 1, 1951 2.7 Rock Creek at Sherrill Drive DC ROCR 01648000 Oct 1, 1929 37.0 Watts Branch DC NACE 01651800 Jun 19, 1992 2.0

Precipitation data Precipitation data (Table 2) was retrieved from the National Climate Data Center (NCDC) “U.S. Climate at a Glance” website (www.ncdc.noaa.gov/oa/climate/research/cag3/cag3.html). Statewide precipitation data was downloaded for Maryland, Virginia, and West Virginia and city wide data was downloaded for Washington DC. Data for MD, VA, and WV is NCDC data. The DC data is from the Automated Surface Observing System (ASOS) station. Data was available from 1895-2009 for the states, and from 1963-2009 for the District of Columbia.

Figure 1. Map of gaging stations used in this study.

Ideally, precipitation data would be used that corresponds to the watershed upstream of each gaging station. To do this would require a determination of which weather stations to use and then compiling the data in a way that takes into account the geographic location of each weather station, which is beyond the scope of the current report. Given the similarity in the rain data for each of the four jurisdictions covered, it is reasonable to assume that they represent the general pattern in precipitation in all of the NCRN streams.

Analysis Precipitation data was analyzed to determine trends in annual and seasonal precipitation for each of the four jurisdictions. For the analysis, the winter season started in December of one year and ended in February of the next year. The other three were defined as follows: spring (March- May), summer (June-August) and fall (September-November). For all analysis, the data for December is always included in the following calendar year.

Like the precipitation data, discharge data was analyzed on an annual and seasonal basis. Unlike the precipitation data, the discharge data was first analyzed to determine the level of discharge that represented the 10th, 20th, 30th, 40th, 50th, 60th, 70th, 80th and 90th percentile of discharge for each year and for each season within each year. The 10th percentile of discharge is the level at which 10% of the discharge values are lower and 90% are higher, the 20th percentile of discharge is the level at which 20% of discharge values are lower and so on up to the 90th percentile of

discharge where 90% of discharge values are lower and only 10% are higher. Maximum and minimum discharges were not analyzed as these are highly variable year to year and are often the results of rare events (e.g. which years had tropical storms).

Statistical analysis of the data follows Khaliq et al. (2009b). Each series of data was first examined for significant lag 1 autocorrelations using the “acf()” function in R.10.0 (R Development Core Team, 2009). A significant first order autocorrelation can artificially inflate the significance of a Mann-Kendall test for trend (Hamed 2008, Khaliq et al. 2009a,b). If the first order lag was not significant, I used the “MannKendall()” function from the Kendall package (McLeod 2009) in R to assess direction and significance of trend.

If the first order lag was significant, then the analysis followed (Khaliq et al. 2009b). In order to correct for autocorrelation, a bootstrapping method is used to determine the correct p-value. Bootstrapping resamples the data a large number of times, putting “blocks” of successive measurements in a random order to determine if any trend in the data is merely an artifact of autocorrelation. A “block” is simply a set of measurements from a given number of successive years, e.g. the 20th percentile discharge from five years in a row. Blocks of successive measurements are used as this will preserve the autocorrelation, but by placing them in a random order this should remove any trend present. If the original data shows a trend that is not present in the bootstrapped data, then the trend is not due to autocorrelation alone and a p-value can be calculated to assess the significance of that trend.

In order to perform a bootstrap test, it is necessary to first determine the size of the block to bootstrap. To do this I followed Khaliq et al. (2009b). They suggested repeatedly bootstrapping the data using a variety of block sizes. This is used to determine the block size that minimizes the difference between first serial autocorrelation coefficients observed from the data and those estimated from the bootstrap replicates. However, in the current study, there was no block size which clearly minimized this difference. Rather the difference in the observed and estimated first serial correlation coefficients smoothly decreased with increasing block size. So instead, based on 100,000 bootstrap replicates of each block size from two to ten, I selected the lowest block value where the mean difference between the observed and estimated first serial correlation coefficient was less than one standard deviation. Once the block size was determined, it was used to perform the bootstrap version of the Mann-Kendall test in R using the tsboot() command from the boot package (Canty and Ripley 2010) and the MannKendall() command from the “Kendall” package. Bootstrapping was conducted using 100,000 bootstrap replicates.

While the Mann-Kendall test can determine if a trend is present and determine the direction of the trend, it does not provide a measurement of the magnitude of the trend. Magnitude of the trend was calculated using the Sen estimate of slope. This was done with the zyp.sen() command in the zyp package for R (Bronaugh and Werner 2009). Sen’s estimate of slope is determined by first estimating the slope between all pairs of points. Practically this means getting an estimate of the slope by comparing Year 1 to Year 2, then Year 1 to Year 3, then Year 1 to Year 4 etc. until all possible pairs have been used. Sen’s value is the median of these estimates.

In many cases, changes are presented as % change per decade. The data is presented in this way in order to make it easier to compare changes in streams with very different discharge values. The Sen’s estimate of slope calculates the magnitude of the trend in change per year. This value

was then multiplied by ten to the get the change per decade. The change per decade was then divided by the mean value over the entire time range of the data.

Results

Precipitation Precipitation follows a characteristic pattern in all four jurisdictions (Figure 2). Precipitation is below annual average during the winter months, and increases in March. Precipitation then drops again in April, before rising during the summer. Precipitation then falls to below average values in the fall.

Figure 2. Mean precipitation for Maryland, Virginia, West Virginia and Washington DC. The horizontal lines indicate the annual mean in each jurisdiction, and the bars indicate the means for each month. Data is from 1895-2009 (MD, VA, WV) and from 1963-2009 (DC).

Throughout the NCRN, there is a slight trend toward increasing annual precipitation (Figure 3). However, the trend is not significant (p >0.05) according to the Mann-Kendall test (Table 2). In Maryland, Virginia and West Virginia the annual trend is small, a change of approximately 0.5% per decade. The trend in DC is larger, 2.0% per decade, but it is still not significant (p =0.32).

Figure 3. Annual Precipitation for Maryland, Virginia, West Virginia and the Washington DC. Horizontal lines indicate the overall means and bars indicate precipitation for each year. Note that due to availability of data, the date ranges for the DC data differ from the other three jurisdictions.

When the data is examined by season, however, different relationships emerge. In all four jurisdictions, precipitation is highest in summer, intermediate in the spring, and lowest in fall and winter (Table 2).

Table 2. Trends in precipitation in the National Capital Region Network by season. State (years) Mean Precipitation Trend Mann-Kendall (inches) (inches/decade) p-value Maryland (1895-2009) Annual 43.1 +0.10 0.52 Winter 9.6 -0.03 0.67 Spring 11.1 +0.11 0.13 Summer 12.5 -0.13 0.12 Fall 9.9 +0.26 0.0044

Virginia (1895-2009) Annual 42.7 +0.13 0.35 Winter 9.4 -0.03 0.62 Spring 11.0 +0.06 0.41 Summer 12.8 -0.15 0.061 Fall 9.5 +0.28 0.0021

West Virginia (1895-2009) Annual 44.3 +0.13 0.35 Winter 10.0 -0.08 0.26 Spring 11.9 +0.13 0.053 Summer 13.2 -0.01 0.83 Fall 9.2 +0.22 0.0064

Washington D.C. (1963-2009) Annual 41.2 +0.78 0.32 Winter 8.6 -0.48 0.13 Spring 10.8 +0.73 0.048 Summer 11.6 +0.02 1.0 Fall 10.4 +0.55 0.29

The analysis showed increases in precipitation in the spring and fall. In MD, VA, and WV, the precipitation increases in the fall are significant (MD: p=0.0044, VA: p=0.0021,WV: p=0.0064) and there is a marginally significant (p=0.053) increase in WV in the spring (Figure 4). DC on the other hand, has significant increases (p = 0.048) in rain in the spring, but a non-significant increase in the fall (p = 0.29).

All four jurisdictions show decreasing precipitation in the winter and summer. The only significant decrease is in Virginia in the summer, and that is at the p<0.061 level.

Figure 4. Seasonal change in precipitation in Maryland, Virginia, West Virginia and the Washington DC. Filled bars indicated trends determined by a Mann-Kendall test significant at p≤0.05, cross hatches indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend is Sen’s slope of seasonal total precipitation on year.

Ironically, these changes reduced the variation in precipitation between the seasons in MD, VA, and WV. For illustrative purposes, comparisons were made between the earliest and latest 50 years (MD, VA, WV) or 20 years (DC) of data (Figure 5). Precipitation decreased in the summer months, when it is typically highest, and precipitation increased in the fall, when the lowest levels occurred.

Figure 5. Comparison of older and more recent precipitation in Maryland, Virginia, West Virginia, and Washington DC by season. Blue lines indicate older precipitation and red lines are more recent precipitation patterns. Error bars are 95% confidence intervals. Note that due to availability of data, the date ranges for the DC data differ from the other three jurisdictions.

These analyses indicate an increase in fall precipitation. Consequently, if precipitation is driving trends in stream discharge, then there should be a trend for increasing discharge that is particularly strong in the fall.

Discharge Seasonal Pattern Stream discharge varies in a predictable pattern during the year (Figure 6). During December, at the beginning of the winter, the median monthly discharge is close to the yearly median. Discharge increases in the following months as precipitation increases, until median discharge values peak in March. This peak is consistent with what is seen in other streams in the eastern half of the United States (Groisman et al. 2001, Lins and Slack 2005). Although precipitation increases into the summer months, discharge decreases. This is likely due to increased evaporation and increased uptake by plants as the summer progresses. Discharge decreases into the fall, until median discharge hits a minimum in September. Discharge then increases until returning to median values at the beginning of the winter. It is striking that this pattern is

consistent across all of the stream gages despite vast differences in the volume of discharge and in the land use surrounding the streams.

Figure 6. Monthly median discharge rates for sixteen stream gages in the NCRN. Horizontal line indicates yearly median discharge. Bars indicate monthly median discharge for each stream.

Annual Trends When trends in discharge are examined on an annual basis, several different patterns emerge. First, consistent with Lins and Slack (1999), several stream gages show a pattern of large and statistically significant increases in the low percentile discharges, with no significant change to higher percentile discharges (Figure 7). These gages include Cumberland, Paw Paw, and Hancock on the upper Potomac, as well as Antietam Creek. A similar pattern is found for South Fork Quantico Creek, except that the increases in low discharge are non-significant.

Figure 7. Trends for gages with increasing discharge at low percentiles. Solid bars indicate trends determined by a Mann-Kendall test to be significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear

bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Stream gages in more developed areas near Washington DC show the opposite pattern. Rock Creek and Difficult Run show significant increases in the high discharge percentiles, and small, non-significant decreases in the low discharge values (Figure 8). The NE branch of the Anacostia also shows this pattern, but the increase has less statistical support. Piscataway Creek shows a somewhat similar pattern, but none of the trends have statistical support. The combination of an increase in the high discharge percentiles, with no change in the low discharge percentiles is resulting in a larger range of discharge in these streams, with the possible exception of Piscataway Creek.

Figure 8. Trends for gages with increasing discharge at high percentiles. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Several stream gages show no significant change in any percentile (Figure 9). These gages include Point of Rocks and Little Falls on the lower Potomac, and the Shenandoah and the Monocacy River. All four of these gaging stations show slight increases in discharge but in no case is this change statistically significant. The Little Falls gage on the Potomac shows a pattern

somewhat like those in developed areas with higher highs and lower lows (Figure 8), but all of the changes are small, and none are significant.

Figure 9. Trends for gages with no significant trends. In all cases the Mann-Kendall test is not significant. Magnitude of trend (size of each bar) is the Sens’s slope of each discharge percentile on year.

One stream gage, the Northwest Branch of the Anacostia, shows consistent significant increases for all percentiles of discharge (Figure 10). This stream gage is located in a suburban area with a relatively high density of impervious surface, so it would be expected to have discharge patterns like Rock Creek or the Northeast Branch of the Anacostia. It is not clear why the Northwest Branch of the Anacostia has such a unique pattern of changes in discharge.

Figure 10. Trends for the stream gage on the Northwest Branch of the Anacostia River. Solid bars represent trends determined by a Mann-Kendall test significant at p≤0.05. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile by year.

Seasonal Trends

We can gain insight into these trends by examining discharge on a seasonal basis. Precipitation is generally increasing the most during the fall, which could cause a similar increase in discharge. Some human impacts, such as dams and impervious surfaces, are present year round and would not be expected to show a seasonal pattern. Intake for municipal water use, however, is highest during the summer and lowest in the winter. Water use in the DC area, however, has not greatly increased since 1990 as growth in population has been matched by increasing efficiency in water use (Ahmed et al. 2010). Therefore, impacts to discharge due to municipal water use in the DC area would not be expected to show a long term trend.

Low discharge values generally occur in the late summer and fall (Figure 6). If discharge were to change during these months, it would be reflected as a change in the lower discharge percentiles on an annual basis. High discharge values generally occur in the winter and spring. A change in discharge during these months would be reflected as a change in the higher percentile discharge values.

Antietam Creek Antietam Creek (Figure 11) shows some evidence of increase throughout the year, but increases significant at the p≤0.05 level occur only in the spring and fall. The drainage area for Antietam Creek occurs largely in the state of Maryland and these are the seasons in which precipitation has increased in that state. The increase in discharge in the fall is approximately 3% per decade across all percentiles of discharge. This is reflected in a similar increase in the 10th, 20th and 30th percentiles of discharge on an annual basis (Figure 7).

Figure 11. Seasonal trends for the stream gage on Antietam Creek at Sharpsburg for 1929-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (size of each bar) is the Sens’s slope of each discharge percentile on year.

Potomac at Cumberland The three gaging stations on the upper Potomac—Cumberland, Paw Paw and Hancock— reveal similar patterns. On an annual basis, all three stations showed large, significant increase to the low percentile discharges, and smaller non-significant changes to the higher percentile discharges (Figure 7). For all three stations, discharge is lowest during the late summer and fall (Figure 6). While discharge generally increased for all three gages at nearly all percentiles for all four seasons, there were some differences.

The Cumberland gaging station (Figure 12) showed significant increase in low percentile discharge levels in all four seasons. Fall, summer and to a lesser extent winter, also have increases in high percentile discharge levels. The magnitude of the increases is greatest in the fall, nearly as large in the summer, smaller in the winter and smallest in the spring. The 10-15% increase in the low percentile discharges in the summer and fall is reflected in a similar increase in the low percentile discharge on an annual basis.

Figure 12. Seasonal trends for the stream gage on the Potomac River at Cumberland for 1929-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Potomac at Paw Paw The Paw Paw gaging station shows similar patterns to that of the Cumberland station. While discharge increases across season and percentile, it did not increase consistently (Figure 13). Fall and summer have larger and statistically significant increases, while the increases in winter and spring are smaller and not significant. The 90th percentile had a non-significant decrease in the winter. Increases in discharge are more dramatic in the fall than in the summer. The magnitudes of the fall and summer increases in discharge match that of the low percentile increases on an annual basis (Figure 7). Notably, the increases in discharge at the Paw Paw gage are consistently smaller than the increases at the Cumberland gage. The Cumberland gage is upstream from the Paw Paw station, and has a lower discharge (Figure 6).

Figure 13. Seasonal trends for the stream gage on the Potomac River at Paw Paw for 1938-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Potomac at Hancock The Hancock gaging station shows increases in discharge in the summer and fall that are statistically significant and fairly large in magnitude (Figure 14). The increases in the summer are somewhat smaller than those seen in the fall. Winter and spring show no significant increases, and there is a non-significant decrease in high percentile discharge in the winter. The increase in low percentile discharge on an annual basis (Figure 7) is about the same magnitude as the increase in discharge in the fall. The Hancock gaging station is downstream from the Paw Paw and Cumberland station, and shows less of an increase in discharge than either of those two stations.

Figure 14. Seasonal trends for the stream gage on the Potomac River at Hancock for 1932-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

South Quantico Creek The South Quantico Creek gaging station shows large and significant increases to discharge in the fall and smaller increases in the summer (Figure 15). The winter showed non-significant increases in low percentile discharges and small, non-significant decreases to high discharge percentiles. Discharges in the spring were nearly unchanged. Despite the increase in fall discharge, the changes in low percentile discharge on annual basis are not significant (Figure 7).

Figure 15. Seasonal trends for the stream gage on South Quantico Creek for 1951-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Difficult Run Difficult Run shows decreases in the low percentile discharge in all seasons, but these decreases are always non-significant (Figure 16). Significant increases in high percentile discharge occur in spring, summer and fall. Winter also shows an increase in higher percentile discharges, but it is not significant. Peak discharge in Difficult Run occurs in the late winter or early spring (Figure 6). On an annual basis, Difficult Run shows a significant increase of over 5% in the 90th percentile discharge (Figure 8). This is intermediate between the winter and spring increases in higher percentile discharge.

Figure 16. Seasonal trends for the stream gage on Difficult Run for 1935-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05, dense lines indicate 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Rock Creek The Rock Creek gaging station showed significant increases in high percentile discharge in all four seasons (Figure 17). In the winter, spring, and fall, there were non-significant decreases in low percentile discharge. In the summer low percentile discharge was unchanged. On an annual basis, high percentile discharges in Rock Creek have increased over 7% (Figure 8). The seasonal increases in high percentile discharge were generally 6-9%.

Figure 17. Seasonal trends for the stream gage on Rock Creek for 1929-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

North East Branch of the Anacostia River The gaging station on the North East Branch of the Anacostia River showed non-significant increases in the high percentile discharge in all four seasons (Figure 18). As with Rock Creek and Difficult Run it also shows small non-significant decreases in low percentile discharge in all seasons except for summer. This pattern matches the increase in high percentile discharge seen on an annual basis (Figure 8).

Figure 18. Seasonal trends for the stream gage on the North East Branch of the Anacostia River for 1938-2008. None of the trends are significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Piscataway Creek Piscataway Creek shows non-significant decreases in discharge at low percentiles in spring and fall, and at all percentiles in the summer and winter (Figure 19). Non-significant increases are seen in high percentile discharges in the spring and fall. As the highest discharge occurs in the spring, discharge on an annual basis also shows a non-significant increase in the high discharge percentiles.

Figure 19. Seasonal trends for the stream gage on Piscataway Creek (1965-2008). None of the trends are significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Potomac at Point of Rocks The stream gage on the Potomac at Point of Rocks showed small increases in stream discharge in every season except for summer (Figure 20). The increases in the lowest percentile of discharge were significant at the p<0.1 level, but on an annual basis this increase was non-significant (Figure 9). Discharge was lower in the summer, but again, this trend is not significant.

Figure 20. Seasonal trends for the stream gage on the Potomac River at Point of Rocks (1895-2008). Dense lines indicate trends determined by a Mann-Kendall test significant at 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Potomac at Little Falls The stream gage on the Potomac at Little Falls showed small a mixture of small non-significant increases and decreases in discharge (Figure 21). Discharge was generally lower in the summer, higher in the winter and spring, and mixed in the fall. This is consistent with the lack of change in discharge on an annual basis (Figure 9).

Figure 21. Seasonal trends for the stream gage on the Potomac River at Little Falls (1930-2008). None of the trends are significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Monocacy River The gaging station on the Monocacy River at Jug Bridge shows a small trend for increased discharge during all seasons except for summer (Figure 22). These increases were not significant, with the exception of some increases in the winter which were significant at the p<0.1 level. The summer shows non-significant decreases at all percentiles of discharge. This is consistent with the lack of significant trend in discharge on an annual basis (Figure 9).

Figure 22. Seasonal trends for the stream gage on the Monocacy River at Jug Bridge from 1930-2008. Dense lines indicate trends determined by a Mann-Kendall test significant at 0.1≤p<0.05 and clear bars are not significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Shenandoah River The gaging station on the Shenandoah River at Millville shows a small trend for increased discharge during all seasons and at all percentiles of discharge (Figure 23). These increases were not significant. This is consistent with the lack of significant trend in discharge on an annual basis (Figure 9).

Figure 23. Seasonal trends for the stream gage on the Shenandoah River at Millville from 1929-2008. None of the trends are significant. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

North West Branch of the Anacostia River The gaging station on the North West Branch of the Anacostia River shows a unique seasonal pattern (Figure 24). Significant increases are evident for every percentile of discharge in every season. The increases are larger in the summer and fall than in the winter and spring, and are particularly large in high percentile discharge in the fall. This is consistent with the significant increase in discharge at all percentiles on an annual basis (Figure 10).

Figure 24. Seasonal trends for the stream gage on the North West Branch of the Anacostia River from 1938-2008. Solid bars indicate trends determined by a Mann-Kendall test significant at p≤0.05 and dense lines indicate 0.1≤p<0.05. Magnitude of trend (length of each bar) is the Sens’s slope of each discharge percentile on year.

Discussion

There has been long standing concern that climate change may result in more “extreme” events, such as floods and droughts (Lins and Cohn 2003). This is understandable as extreme events are the most noticeable and have the potential to cause significant property damage. Extreme events are by their nature relatively rare, and therefore it can be difficult to detect changes in their frequency.

Relatively less concern has been focused on changes that do not represent extremes, but rather are changes in the typical patterns of discharge. While these changes may be less dramatic, they could still influence natural processes and could potentially pose challenges for resource management. Changes in annual patterns can be easier to detect than changes in extremes, as more relevant data is available.

The analyses presented in this report demonstrate that precipitation, a component of the climate, has changed over the last century in the NCRN. Correspondingly there have been changes in discharge in some streams in the NCRN. However, an important result of this study is that human alterations to streams and surrounding landscapes can influence the direction and magnitude of trends seen in stream discharge, or if a trend is found at all.

Precipitation In general, changes in precipitation followed previously published patterns (Lettenmaier et al. 1994, Groisman et al. 2001, 2004, Huntington 2006). Increases were seen in the spring and fall months with the largest increases generally found in the fall. Decreases in precipitation were seen in the winter and summer (Figure 5). What is most striking about this pattern is that the net effect of these trends is to reduce the seasonal variation in precipitation.

This is somewhat in contrast to predictions for future precipitation made by climate models. Precipitation is predicted to increase in the winter months in Maryland, and some models predict declines in summer and fall precipitation (Boesch 2008).

Discharge If changes in precipitation were the only relevant factor, then streams throughout the NCRN would be expected to show an increase in fall discharge. All fourteen streams analyzed showed some increase in fall discharge, although for some streams the increase was apparent only in the higher percentiles of discharge. In eight of the streams at least some of the fall increase was statistically significant. This pattern did not hold true for the other seasons. While winter and summer discharge usually increased, it did not increase for all fourteen streams. Furthermore, there were fewer significant increases in the other seasons compared to the fall. In general, the increases in discharge in the fall for a particular percentile were larger than the equivalent increases seen in the other seasons.

Despite this broad pattern, there are important differences between the various stream gages that are apparently correlated with human influence. Streams in this study differed in land use, in that some are in agricultural or natural areas with relatively little impervious surface. Others are largely in the Washington DC metropolitan area which has a high percentage of impervious

surface. The streams also differ in the extent to which they are used as a source for municipal water. The larger streams in the study are important water sources, whereas the smaller streams are not used for that purpose.

Rural Streams Five stream gages, Antietam Creek, South Quantico Creek, and the Cumberland, Paw Paw, and Hancock gages on the upper Potomac, are in areas with relatively little urbanization. Most of the area drained by these streams is agricultural, but the drainage of South Fork Quantico Creek is largely forested and protected as part of Prince William Forest Park and Quantico Marine Base.

These gages show increases, in some cases dramatic, in fall discharge along with more modest increases in other seasons. For some, but not all, streams increases in summer discharge were nearly as great as increases in fall discharge. Across the NCRN, stream discharge is lowest in the summer and fall. The trends documented here have the net effect of increasing discharge in the months when it is lowest.

This pattern is effectively a homogenization of discharge throughout the year. As discharge increases in the lower percentiles faster than at the higher percentiles, the differences between the high and low discharge is decreasing. This mirrors the pattern in precipitation discussed above, where long term trends are resulting in less variation between seasons.

Given the decrease in summer precipitation it is somewhat surprising to see an increase in discharge during the summer months. One possible explanation is that some of the increase is due to plants decreasing their water use as carbon dioxide concentration increases. As plants will require more water in the summer than in the other seasons, it may be that a decrease in plant consumption during the summer is currently more than compensating for any decrease in precipitation. However, without any direct measures on evapotranspiration in the area, it is not possible to assess if plants are increasing or decreasing their consumption of water. Furthermore it may be that in the future, longer growing seasons or more robust plant growth could lead to an increase in plant consumption of water.

Urban Streams Four stream gages, Difficult Run, Rock Creek, the North East Anacostia, and to a lesser extent Piscataway Creek show a strikingly different pattern. All of these streams are in urbanized areas. With the exception of Piscataway Creek, high percentile discharge is increasing in every season, although this increase is not always significant. Increases in the fall are somewhat higher than increases in the other seasons. Low percentile discharge is decreasing in every season for all four streams. The result of this on an annual basis is that low percentile discharge is decreasing, whereas high percentile discharge is increasing, often significantly.

In contrast to the more rural streams, the trends in these streams are producing a greater range of discharge with higher highs and lower lows. This pattern does not appear to be climate driven, however. In all but Piscataway Creek, the pattern is present year round. If changes in discharge were primarily driven by changes in precipitation, all streams would be expected to show a seasonal pattern. Instead, it is more likely that changes in these steams are being driven by increasing urbanization in their drainage areas, with changes in precipitation playing a much

smaller role. A discharge pattern of higher highs and lower lows has been observed in other urban streams and is associated with increased flashiness and bank-destroying erosion (Poff 2006).

The lack of significant change, and the different seasonal pattern in Piscataway Creek may be due to the fact that the data for Piscataway begins in 1965, whereas for the other streams there is data available starting in the 1920’s or 30’s.

Large Rivers The stream gages that show the least amount of change in discharge are those that are on large rivers which have been dammed and are used for municipal water supply. The gaging stations on the Shenandoah, and the Potomac at Little Falls showed no significant trends on an annual or seasonal basis. The Monocacy River did show a slight trend to an increase in middle percentile discharge during the winter. The Potomac at Point of Rocks showed a slight trend to increase low percentage discharge in every season but summer, which is similar to the rural streams, but otherwise showed no significant trends.

Without detailed daily records of water removals from these rivers, it is difficult to be certain why they show little change in discharge despite the trends in precipitation. One possibility, however, is that increases in water withdrawals over the past decades may mask any increase in discharge due to precipitation. The stream gages on the upper Potomac showed increases in the summer and fall months. These seasons are the times when demand for municipal water is greatest (Ahmed et al. 2010) so withdraws could have canceled out any increases in discharge. However, as noted earlier, the DC region has not significantly increased water usage since 1990 (Ahmed et al. 2010). If municipal water usage is canceling increases in discharge, this effect should be weaker in recent years, as precipitation increases, but intake remains flat.

Additionally the dams on these rivers, particularly those that retain water in reservoirs may also reduce the effect of changes in precipitation. Dams reduce peak discharge and increase minimum discharge (Poff et al. 2006). These changes would tend to reduce or eliminate trends by homogenizing the stream discharge.

Northwest Branch of the Anacostia Unlike all the other stream gages, the gage on the Northwest Branch of the Anacostia showed large increases in discharge in all seasons and across all percentiles. One possible explanation for this is that prior to 1961 water upstream from this gage was regulated by the Burnt Mills Dam. After 1961 this water regulation ceased (USGS Water Information System Web Interface), which may have allowed higher discharge.

Limitations of this Study It is important to note that this study has several limitations. First and foremost, this study is retrospective in nature, i.e. it looks solely at trends which have already occurred. This should not be taken as proof that these trends will continue into the future. There is still uncertainty as to how climate change will impact the Mid-Atlantic, and this is particularly true in regards to precipitation (Boesch 2008). While it may seem reasonable to assume that the near future will be like the recent past, this becomes less reasonable the farther into the future you wish to predict.

Secondly, the streams that were examined in this study are not from a random selection of available stream gages. Rather, the gages selected are in or near national park lands, and therefore are relevant to resource managers. While I do not have any reason to believe that this selection of gages is biased towards any particular trend in discharge, they are not selected to be representative of all streams in the NCRN or the Mid-Atlantic region in general.

Finally, the streams have been grouped as “rural”, “urban”, etc. based on land use and human alteration of the streams. This has been done based on general knowledge of the area, rather than a formal analysis of the landscape. It would be useful to expand this study to additional stream gages and quantify the impervious surface in the drainage of each gage. That would better establish the relationship between urbanization and trends in stream flow, and it would aid in estimating changes that may have occurred in streams that do not have gages.

Implications The most important result of this study is that precipitation (an aspect of climate) and discharge have been changing in the NCRN and this result is consistent with previous research (see Introduction). While we cannot predict if these particular changes will continue, it demonstrates that climate related changes are a current reality and not just a future concern.

Another important finding is that the trends in discharge varied dramatically between streams, despite being subject to a similar precipitation pattern and being close geographically. As a consequence, not all streams in a park may respond in the same way to climate change. This will be particularly true in parks such as the C&O Canal, and National Capital Parks East, which are spread out over large areas with a variety of land uses. Furthermore, in some cases, but not all, the trends had a strong seasonal component. Resource managers will need to consider these issues as they plan adaptations to climate change. Within a single park some areas may require no adaptation, some may have issues on a seasonal basis, and others may require adaptation year round.

Literature Cited

Ahmed, S.N., K.R. Bencala and C. L. Schultz. 2010. Washington Metropolitan Area Water Supply Reliability Study. Part 1: Demand and Resource Availability forecast for the Year 2040. Interstate Commission on the Potomac River Basin, ICPRB Report NO. 10-01.

Allan, J. D. 2004. Landscapes and riverscapes: The influence of land use on stream ecosystems. Annu. Rev. Ecol. Evol Syst. 35: 257-284.

Boesch, D.F. (ed) 2008. Global Warming and the Free State: Comprehensive Assessment of Climate Change Impacts in Maryland. Report of the Scientific and Technical Working Group of the Maryland Commission on Climate Change. University of Maryland Center for Environmental Science, Cambridge, Maryland.

Bronaugh, D. and A. Werner for the Pacific Climate Impacts Consortium (2009). zyp: Zhang + Yue-Pilon trends package. R package version 0.9-1. (http://www.r-project.org).

Burns, D. A., J. Klaus and M. R. McHale. 2007. Recent climate trends and implications for water resources in the Catskill Mountain, New York, USA. Journal of Hydrology 336:155-170.

Canty, A. and B. Ripley. 2010. boot: Bootstrap R (S-Plus) functions. R package version 1.2-42.

Gedney, N., P. M. Cox, R. A. Betts, O. Boucher, C. Huntingford and P. A. Stott. 2006. Detection of a direct carbon dioxide effect in continental river runoff records. Nature 439: 835-838.

Gore, J. A. 1996. Chapter 3: Discharge measurements and streamflow analysis. pp. 53-74 in R Hauer, and G. A. Lamberti. Methods in Stream Ecology.. New York, Academic Press. 674p,

Groisman, P. Ya., R. W. Knight and T. R. Karl. 2001. Heavy precipitation and high streamflow in the contiguous United States: Trends in the 20th century. Bulletin of the American Metrological Society. 82:219-246.

Groisman, P. Ya., R. W. Knight, T. R. Karl, D. R. Easterling, B. Sun, and J. H. Lawrimore. 2004. Contemporary changes of the hydrological cycle over the contiguous United States: Tends derived from in situ observations. Journal of Hydrometeorology 5:64-85.

Hamed, K.H. 2008. Trend detection in hydrologic data: The Mann-Kendall trend test under the scaling hypothesis. Journal of Hydrology 349:350-363.

Huntington, T. G.2006. Evidence for intensification of the global warming cycle: Review and synthesis. Journal of Hydrology 319:83-95

Kalra, A, T. C. Piechota, R. Davies, and G. A. Tootle. 2008. Changes in U.S. streamflow and western U. S. snowpack. Journal of Hydrologic Engineering 13: 156-163

Khaliq, M. N., T.B.M.J. Ouarda, and P. Gachon. 2009a. Identification of temporal trends in annual and seasonal low flows occurring in Canadian rivers: The effect of short- and long- term persistence. Journal of Hydrology 369:183-197.

Khaliq, M.N., T.B.M.J. Ouarda, P. Gachon, L. Sushama, and A. St-Hilaire. 2009b. Identifiction of hydrological trends in the presence of serial and cross correlations: A review of selected methods and their application to annual flow regimes of Canadian rivers. Journal of Hydrology 368:117-130.

Krakauer, N. Y. and I. Fung. 2009. Is streamflow increasing? Trends in the coterminous United States. Hydrol. Earth Syst. Sic. Discuss 5:785-810.

Leopold, L. B., M. G. Wolman and J. P. Miller. 1964. Fluvial processes in geomorphology. Freeman. San Francisco CA.

Lettenmaier, D.P., E. F. Wood and J. R. Wallis. 1994. Hydro-climatological trends in the United States 1948-1988. Journal of Climate 7:586-607.

Lins, H.F. and T. A. Cohn. 2003. Floods in the greenhouse: Spinning the right tale. In Applications in Flood Risk Assessment. Proceedings: PHERA International Workshop, Barcelona 16-19 October, 2002, pp. 263-8.

Lins H.F. and J. R. Slack. 1999. Streamflow trends in the United States. Geophysical Research Letters 26;227-230.

Lins, H. F. and J. R. Slack. 2005. Seasonal and regional characteristics of U.S. streamflow trends in the United States from 1940 to 1999. Physical Geography 26: 489-501.

McCabe, G. J. and D. M. Wolock. 2002. A step increase in streamflow in the conterminous United States. Geophysical Research Letters 29: 2185-2188

McLeod, A.I. 2009. Kendall: Kendall rank correlation and Mann-Kendall trend test. (http://CRAN.R-project.org/package=Kendall).

Norris, M. E., A. Cattani, and J. M. Pieper. 2007. National Capital Region Network Inventory and Monitoring Program Water Chemistry Monitoring Protocol. National Park Service, Washington, District of Columbia.

Poff, N.L., B.P. Bledsoe, and C. O. Cuhaciyan. 2006. Hydrologic variation with land use across the contiguous United States: Geomorphic and ecological consequences for stream ecosystems. Geomorphology 79: 264-285.

R Development Core Team. 2009. R: A language and environment for statitistical computing. (http://www.R-project.org).

Rose, S. 2009. Rainfall-runoff trends in the south-eastern USA: 1938-2005. Hydrological Processes 23:1105-1118.

Slack, J.R. and J. M. Landwehr. 1992. Hydro-climactic data network: A U.S. Geological Survey streamflow dataset for the study of climate variations, 1974-1988. USGS Open File Report 92-129.

Small, D., S. Islam, and R. M. Vogel. 2006. Trends in precipitation and streamflow in eastern U.S.: Paradox or perception? Geophysical Research Letters 33:L03403

Statzner, B., J A Gore, and V. H. Resh. 1988 Hydraulic stream ecology: observed patterns and potential applications. Journal of the North American Benthological Society. 7:307-360.

Vannote, R.L. and G. W. Minshall. 1982. Fluvial processes and local lithology controlling abundance, structure and composition of mussel beds: PNAS 79:4103-4107.

Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell, and C. E. Cushing. 1980. The river continuum concept. Canadian Journal of Fisheries and Aquatic Science 37:130-137.

The Department of the Interior protects and manages the nation’s natural resources and cultural heritage; provides scientific and other information about those resources; and honors its special responsibilities to American Indians, Alaska Natives, and affiliated Island Communities.

NPS 800/109866, September 2011

National Park Service U.S. Department of the Interior

Natural Resource Stewardship and Science 1201 Oakridge Drive, Suite 150 Fort Collins, CO 80525 www.nature.nps.gov

EXPERIENCE YOUR AMERICA TM