ABSTRACT 
Title of Thesis: A COMPARATIVE HYDROLOGIC ANALYSIS OF 
SURFACE MINED AND FORESTED WATERSHEDS 
IN WESTERN MARYLAND. 
Degree candidate: Timothy Lee Negley 
Degree and year: Master of Science, 2002 
Thesis directed by: Dr. Keith N. Eshleman 
University of Maryland Center for Environmental Science 
This thesis presents the results of a hydrologic analysis conducted as part of a larger, 
multi-faceted, collaborative effort to study ecosystem function of a watershed subjected 
to surface mining and reclamation in the Appalachian Region of the United States. The 
primary goal of this study was to determine whether a small watershed subjected to 
surface mine reclamation practices (MAT 1) displayed a stormflow response to rain 
events that was different from those displayed by a young second-growth forested 
watershed (NEFl). A secondary goal was to investigate whether intensive surface 
mining in the Georges Creek basin has altered stormflow response at a larger river basin 
scale when compared to a similar, but predominantly forested basin (Savage River). At 
the small watersheds, MATl produced greater a) runoff coefficients (2.5x); b) total 
runoff (3x); and c) peak runoff rates (2x) compared to NEFl. Total rainfall explained 
63% of the variation in total runoff at MAT I compared to only 21 % of the variation in 
total runoff at NEF I. Regardless of a recent 13% increase in surface mine reclamation 
in the Georges Creek basin, little difference in stormflow response was observed for 15 
storms analyzed across the two larger basins. Georges Creek on average responded 3 hr 
more quickly than Savage River, However the hydrological response characteristics of 
the two basins were similar. In addition, hydrological response characteristics for 
Georges Creek and Savage River remained relatively stable over time. Further research 
is needed to address inabilities to scale responses observed at the small watersheds. 
, 
A COMPARATIVE HYDROLOGIC ANALYSIS OF SURFACE MINED AND 
FORESTED WATERSHEDS IN WESTERN MARYLAND. 
by 
Timothy Lee Negley 
Thesis submitted to the Faculty of the Graduate School of the 
University of Maryland at College Park in partial fulfillment 
of the requirements for the degree of 
Master of Science 
2002 
Advisory Committee: 
Dr. Keith N. Eshleman, Chair 
Dr. William S. Currie 
Dr. Philip Townsend 
ACKNOWLEDGEMENTS 
I would like to acknowledge first of all my committee members Drs. Keith Eshleman, 
William Currie, and Phil Townsend for their guidance, direction, and insightful 
comments on this thesis. Field and lab supprnt for this project was provided by: Jim 
Kahl (Maryland Department of Environment) and Roger Thomas (Natural Resource 
Conservation Service) who assisted in site se lection; Keith Eshleman, Mark Castro, 
Matt Kline, Jamie Welch, and Derek Wiley w10 assisted in transporting flumes to these 
remote sites; Keith Eshleman and Randy Richardson who spent days working alongside 
me in the blazing summer heat installing the stream gages; and Randy Richardson who 
was instrumental in digitizing over a hundred aerial photos. This research made use of 
data and models provided by: Aaron Friend acd Ben Cooper (Allegany County Soil 
Conservation District) ; Mike Kozier (Garret County Soil Conservation District); Jeff 
Griffith, Charles Strain, and Robert James (USGS); Dr. Ian Littlewood (Center for 
Ecology and Hydrology); the National Climat ic Data Center; and the Maryland Bureau 
of Mines. This research would not have been possible without the permission and 
cooperation of landowners Mr. & Mrs. Paul Willison and the late Simon Moore. 
Funding for this research was provided by a grant from the A.W. Mellon Foundation to 
promote collaboration via a collaborative eco logical research project between the 
Appalachian Laboratory and the Appalachian College Association. 
Ill 
TABLE OF CONTENTS 
List of Tables ................................................................................................................. v 
List of Figures ............................................................................................................... vi 
Chapter I: INTRODUCTION ............................................................. ...................... 1 
A. Goals and Objectives .............................................................................. 6 
Chapter II: METHODS ............................................................................................. 8 
B. Study Sites .............................................................................................. 8 
C. Field Hydrologic Measurements ........................................................... 13 
D. Historical Hydrologic Measurements ................................................... 17 
E. Unit Hydrograph Deconvolution .......................................................... 19 
F. Historical Land Use/ Land Cover Derivation .. .................................... 20 
G. Data Analysis ........................................................................................ 22 
Chapter III: RESULTS ............................................................................................. 35 
Chapter IV: DISCUSSION ....................................................................................... 61 
Chapter V: CONCLUSIONS ................................................................................... 71 
Appendix I. Coordinates of Watersheds and Gages ................................................. 73 
Appendix II. LULC Classes, Codes, and Identification Key ..................................... 74 
Appendix III. PC-IHACRES Model Results ............................................................... 7 5 
Appendix IV. LULC Changes Over Time by Category (area in hectares) ................. 77 
REFERENCES ............................................................................................................ 78 
iv 
LIST OFTA BLES 
Table 1. Drainage area, slope, and elevation of watersheds used in comparative 
analysis ............................... ................................................... ... .................... 33 
Table 2. Stage discharge relationships for the stream gages installed in the small 
catchment study ............................................................................................ 34 
Table 3. Number of LULC patches, % of watershed, mean area, and total area in 
each LULC class for the Georges Creek watershed (1938-1997) ................ 60 
V 
LIST OF FIGURES 
Figure 1. Generalized model of abstraction rates: <I>-index and IHACRES .................. 23 
Figure 2. Schematic representation of IHACRES ................................ ........................ 24 
Figure 3. Site map and equipment locations for the small watersheds .. ..... ..... ............ .. 25 
Figure 4. Watershed boundaries and 1997 LULC for Georges Creek and Savage 
River watersheds ........................................................................................... 26 
Figure 5. Watershed slope: Georges Creek and Savage River ..................................... 27 
Figure 6. Equipment installed at MAT 1 ....................................................... ................ 28 
Figure 7. Plan and side views of flumes installed at MATl and NEFl ....................... 29 
Figure 8. Plan and side views of stilling stream gage installed at EBNR ...... .......... ..... 30 
Figure 9. Stage-discharge relationship for EBNR. ........................................................ 31 
Figure 10. Schematic of double-ring infiltrometer ................................... ..................... 3 2 
Figure 11. Average annual discharge and daily precipitation at MAT 1 and NEF 1. .... .43 
Figure 12. Long-term average cumulative precipitation at Savage River Dam ...... ..... .44 
Figure 13. Annual water balances for the small watersheds and larger basins ........... .45 
Figure 14. Mean watershed response characteristics for MATl and NEFl.. ............... .46 
Figure 15. Runoff coefficients and maximum rainfall intensities for MAT 1 and 
NEFl .............................................................................. ............................... 47 
Figure 16. Total event runoff and maximum rainfall intensities for MAT 1 and 
NEFl ............. .. ....... ..... ..... ............... ... ................ .. .................. .................. ..... 48 
Figure 17. Total rainfall vs total stormflow at MAT 1 and NEF 1.. ............................... .49 
Figure 18. Peak runoff and maximum rainfall intensities for MAT 1 and NEF 1. ......... 50 
Figure 19. Maximum rainfall intensity vs peak stormflow for MATl and NEFl. ....... 51 
Figure 20. Unit hydrographs deconvolved for MATl and NEFl. ..... ..... ...................... 52 
Figure 21. Cumulative depth of water infiltrated at MAT 1 and NEF 1. ............... ......... 53 
vi 
Figure 22. Maximum hourly rainfall intensity vs soil infiltration capacities at 
MATl and NEFl .......................................................................................... 54 
Figure 23. LULC change over time in the Georges Creek watershed (1938-1997) ...... 55 
Figure 24. Primary LULC changes by percent in the Georges Creek Basin ................. 56 
Figure 25. LULC changes by class in Georges Creek basin (1938-1997) .................... 57 
Figure 26. Centroid lag and trend lines for the Georges Creek and Savage River 
watersheds (1952-2000) ............................................................................... 58 
Figure 27. Runoff Ratios: George Creek vs Savage River .......................................... 59 
vii 
Chapter I: INTRODUCTION 
The relationship between land use and land cover (LULC) and the hydrologic response 
of watersheds is becoming highly scrutinized in science, management, and public 
policy. In the majority of studies, the effects of deforestation, agriculture, urbanization, 
and wetland drainage have been examined. The watershed is frequently chosen as the 
basic unit of study for examining the effects of LULC change because watersheds have 
a definable hydrologic boundary; the area within the watershed boundary can be 
thought of as a "black box" where the difference between inputs and outputs are stored 
within the system. However, disturbance occurring within the system has the potential 
to alter the hydrologic responsiveness of watersheds (Freeze 1974). In the majority of 
studies of LULC change, hydrologic response characteristics such as changes in peak 
stormflow, flood frequency, and rainfall/runoff ratios have been examined. 
A number of studies have investigated the hydrological effects of timber harvesting 
(Hornbeck et al. 1970, Swift et al. 1975, Burt and Swank 1992, Jones and Grant 1996, 
Burton 1997, Kochenderfer et al. 1997, Thomas and Megahan 1998), urbanization and 
suburbanization ( Burges et al. 1998, Rose and Peters in press), and changes in 
agricultural areas and practices (Gebert and Krug 1996, Kuhnle et al. 1996, Allan et al. 
1997, Mwendera and Mohamed-Saleem 1997). Changes in LULC often involve 
altering the land cover through intensive vegetation removal. Hornbeck et al. ( 1970), 
Burton ( 1997), and others have observed that intensive removal of vegetation can 
significantly increase runoff and flooding hazards. Most often, streamflow changes 
have been attributed to changes in evapotranspiration rates (Swift et al. 1975, Gifford et. 
al. 1984, Swanson 1984, Troendle and King 1987). Implementation of best 
management practices such as proper road design on logged lands can reduce the effects 
of timber harvesting (Kochenderfer et al. 1997, Thomas and Megahan 1998). In 
addition, Burt and Swank (1992) provide some evidence that as a forest regenerates it 
can exhibit evapotranspiration rates as high as dense grass. All of these studies suggest 
a strong relationship between land use change and watershed response, however. 
Another specific type of land use change that disturbs many of the physical properties 
of a watershed is the extraction of bituminous coal via surface mining. Surface mining 
and subsequent land reclamation has become widespread in the Appalachian region of 
the United States since the early 1950's (J. Carey, Maryland Bureau of Mines, personal 
communication) with the advent of large earthmovers. Under the Surface Mine Control 
and Reclamation Act of 1977 (SMCRA, PL 95-87) mine operators are obligated to 
reclaim surface mined lands to the approximate original contours and to acceptable 
LULC. The overall process involves extracting the material, or overburden, that 
overlies the coal seam. The topsoil is retained in a separate pile. Following coal 
extraction, the overburden is replaced, graded to the approximate original contour using 
large earthmovers, and typically seeded with grasses. A common result of reclamation 
is minesoils that are highly compacted (Bussler et al. 1984, Mcsweeney and Jansen 
1984, Bell et al. 1994, Chong and Cowsert 1997). 
The Georges Creek watershed in western Maryland is an example of a watershed that 
has undergone intensive surface mining. Catastrophic flooding in the Georges Creek 
watershed in June of 1995 and January of 1996 led to speculation that surface mining 
and reclamation has altered the hydrologic response of the watershed and increased the 
2 
potential for damaging floods and associated economic losses. In addition to the 
economic losses caused by flooding, increased flooding frequency can have deleterious 
effects on stream biota. However, in the Georges Creek watershed as elsewhere, the 
effects of this LULC change on watershed stormflow response are poorly understood 
and empirical data on this phenomenon are essentially non-existent. 
A limited number of studies have investigated the hydrological effects of surface 
mining and reclamation on watershed stormflow response, but essentially no research 
has focused on the long-term cumulative impacts of mine reclamation distributed 
throughout a watershed. In theory, watersheds subjected to mine reclamation may 
respond similar to those having undergone urbanization/suburbanization, as both 
activities act to decrease the perviousness of the landscape. Most imperviousness on 
reclaimed surface mines is the result of massive compaction (Bussler et al. 1984, 
McSweeney and Jansen 1984, Bell et al. 1994, Chong and Cowsert 1997). Compaction 
has been shown to substantially reduce infiltration rates (Barnhisel and Hower 1997) 
and essentially eliminate the macropore networks (Dunker et al. 1995) that increase 
infiltration capacities (Beven and Germann 1982). Mine reclamation can also disturb 
water table elevations and subsurface flow paths (Bonta et al. 1992). Ritter and Gardner 
(1993) observed that on newly reclaimed mine lands in Pennsylvania, infiltration-excess 
overland flow is the dominant runoff process. Likewise Bonta et al. ( 1997) observed 
increased peak streamflow rates in response to rnjne reclamation. It could be argued 
that the limited data available for the pre-mining period, however, were insufficient to 
compare pre- and post-mining impacts. 
3 
The unit hydrograph technique is one method that could help quantify the effects of 
strip mine reclamation. The method was first outlined by Sherman (1932) and is still 
widely used in hydrological studies (Chapman 1996a, b, Dietrich 1996, Sefton and 
Boorman 1997), particularly in urban planning. The unit hydrograph of a watershed is 
defined as the hydrograph of one unit (inch or cm) of storm runoff generated by a 
rainstorm of uniform intensity and distribution occurring within a specific period of 
time (Dunne and Leopold 1978). Unit hydrographs are conducive to investigating 
hydrological effects of LULC change because in theory they are affected by a) rainfall 
characteristics and b) watershed characteristics. For small watersheds on the order of 1 
km2 or less, many of the watershed characteristics are fixed from storm to storm (e.g., 
watershed area, topography, channel morphology, LULC, and soil properties). 
Therefore, one might expect that storms with similar rainfall characteristics (similar 
depth and intensity) would produce similar unit hydrographs. For larger watersheds on 
the order of 100 km2, however, the necessary assumption that a rain event is uniformly 
distributed over the watershed is usually difficult to achieve. Physiographic features, 
such as LULC, can change appreciably over relatively short time periods ( ~50 years). 
Based on unit hydrograph theory, if variations in rainfall characteristics can be 
minimized between watersheds and among the set of storms being analyzed, then 
differences in unit hydrograph shape could only be attributed to changes in physical 
watershed characteristics (e.g., LULC). 
Numerous methods exist for calculating unit hydrographs by deconvolution (Snyder 
1938, Langbein 1940, Rantz 1971, US Soil Conservation Service 1972), one of which is 
the relatively simple <D-index technique. The <D-index method produces an estimate of 
4 
the excess rainfall hyetograph by assuming a constant rate of rainfall "abstraction" (that 
fraction of the rainfall that does not contribute to stormflow) represented by <D (Figure 
IA). By definition, the excess rainfall is that portion of total rainfall that produces 
direct runoff. A more sophisticated approach is used by PC-IHACRES (identification 
of unit _hydrographs .wid fomponent flows from rainfall, ~vaporation and ~treamflow 
data) developed by Littlewood et al. (1997). IHACRES is a rainfall-runoff model that 
uses a) a non-linear loss module to determine effective rainfall and b) a linear loss 
module to develop a unit hydrograph used for estimating streamflow (Figure 1B  ). The 
main advantage of using IHACRES is its spatially 'lumped' approach (Figure 2); the 
sole data requirements are continuous time series of rainfall, streamflow, and air 
temperature (although the model can be calibrated without air temperature). A large 
number of studies published in the literature have applied and successfully calibrated 
IHACRES over a wide range of spatio-temporal conditions (Chiew et al. 1993, Hansen 
et al. 1996, Schreider et al. 1996, Andreassian et al. 2001, Kokkonen et al. 2001, 
Letcher et al. 2001, Schreider et al. 2001). The model produces reasonable estimates of 
unit hydrographs over watersheds ranging from 490 km2 to 10,000 km2 in size 
(Littlewood et al. 1997) and also for ephemeral watersheds in temperate regions (Ye et 
al. 1997, 1998). Despite the differences in sophistication between the <D-index method 
and IHACRES, both use the same basic unit-hydrograph theory and have been 
employed successfully in deconvolving unit hydrographs based on effective rainfall and 
direct runoff observations. 
5 
A. Goals and Objectives 
The primary goal of this study was to determine whether small watersheds subjected to 
mine reclamation practices display a stormflow response to rain events that is different 
from those displayed by similar watersheds that are covered by typical second-growth 
forests. A secondary goal was to investigate whether intensive surface mining in the 
Georges Creek watershed of western, Maryland has appreciably altered the stormflow 
response at the larger river basin scale. Specifically, the following hypothesis was 
tested. 
H : The mean difference between the stormflow response of a surface mined/reclaimed 
0
watershed and a reference watershed is not significantly different from zero. 
Ha: The mean difference between the stormflow response of a surface mined/reclaimed 
watershed and a reference watershed is significantly different from zero. 
Several objectives were developed to achieve these goals. The first objective was to 
select a pair of small (<I km2) watersheds that could be used to conduct a comparison 
of stormflow characteristics and soil hydraulic properties. The comparison would 
include measurements of soil infiltration capacities, stream discharge, rainfall runoff 
ratios, and response lag times. It was desired that watersheds have similar area, slopes, 
aspects, etc., but differ only in their present LULC. The second objective was to 
statistically compare the hydrologic respon~e characteristics of each small watershed to 
a set of storms. The third objective was to investigate the long-term hydrological 
responses of the larger Georges Creek ( 186 km2) watershed (extensively surface mined 
6 
and reclaimed) to the Savage River (127 km2) watershed, which is primarily 
undeveloped forest. A fourth objective was to test for statistically significant 
differences in watershed response characteristics between each of the river basins as 
well as within each river basin over time. This final object was to address the question 
of whether the hydrological response to rain events within the Georges Creek basin has 
resulted from a concomitant increase in surface mining and reclamation throughout the 
watershed. 
7 
Chapter II: METHODS 
Five watersheds were used in this study to investigate the hydrological effects of LULC 
disturbances. The first component of the study involved a characterization and 
comparison of the hydrologic responses of two small watersheds subjected to different 
LULC disturbances. The two watersheds would have approximately identical physical 
features, with the exception that one watershed was surface mined and reclaimed. 
However, a number of other factors also required consideration in selecting the pair of 
watersheds including a) proximity to the laboratory and to each other, b) ability to gain 
landowner permissions and state environmental permits, and c) suitability of sites for 
stream gaging. A third small watershed was later added for comparisons of annual 
water balances. The second component of the study involved an historical 
characterization of hydrologic responses of two larger river basins subjected to different 
L ULC disturbances. The basins were selected on the basis of anecdotal flooding 
history, proximity to each other, degree of surface mining and reclamation, and 
availability of historical aerial photos, long-term historical rainfall and streamflow data. 
After preliminary land use history data were obtained, the larger watersheds were 
investigated within the following time periods: a) 1950-1966, which represented a 
period before intensive surface mining and reclamation; b) 1967-1984, which 
represented a period of early surface mine reclamation; and c) 1985-2000, which 
represented a period of widespread surface mine reclamation. 
B. Study Sites 
In this study I used a Geographic Information Systems (GIS) database to identify 
watershed physical features (Table 1) for 5 watersheds located in Allegany and Garrett 
8 
Counties of western Maryland. For the two small watersheds (MAT l and NEF 1), I 
delineated watershed boundaries using a Trimble Pro XR Global Positioning System 
(OPS). The basic approach involved walking the perimeter of each watershed while 
recording position coordinates at every major change in direction. After collecting 
coordinates in the field, I differentially corrected the positions using Pathfinder office (v 
2.51) and base station data from Morgantown, West Virginia to obtain the highest 
possible accuracy of watershed area. For two larger basins (Georges Creek and Savage 
River) I delineated watershed boundaries using topographic maps obtained from the 
USGS. To calculate watershed slopes, I used a digital elevation model with 30 m 
resolution generated by the USGS to calculate average slope in ArcView? 3.1 using 
the Spatial Analyst? extension. The software derives the slope of a DEM grid cell 
using the distance to and elevation of its nearest neighbors. In addition to slope, I 
calculated drainage densities for the Georges Creek and Savage River watersheds in 
Arc View? 3.1 by summing the lengths of all stream segments in the basin and dividing 
by the total drainage area. 
The first watershed (hereafter referred to as MAT l) is a small watershed that has 
undergone significant surface mining and reclamation (39? 35' 39" N; 78? 53' 29") 
located in the larger Matthew Run watershed. MAT 1 is drained by an ephemeral 
diversion ditch that is a tributary to Matthew Run (Figure 3). The watershed has a 
drainage area of 27. I-ha, of which 12.4-ha (4 6%) has been mined and reclaimed. 
Mining at MAT 1 began in 1982 and reclamation was finished in 1984; reclamation 
involved returning the land to the approximate original contour and seeding the site 
with a mix of grasses for hay/pasture (Mining Permit #371 , Maryland Bureau of 
9 
Mines). Nearly 20 years later the site remains primarily herbaceous vegetation, with 
several black locust trees (Robinia pseudoacacia). Woody vegetation in the forested 
area of MAT 1 watershed has been inventoried by K. Kuers (unpublished data, 
Department of Forestry and Geology, The University of the South) and summarized by 
importance value (a combined value for frequency, density, and basal area). The most 
important species in the forested area at MATl are black birch (Betula lenta), red maple 
(Acer rubra), and chestnut oak (Quercus prinus). Prior to mining, soils were mapped as 
a combination of Cookport silt loam and a Cookport very stony silt loam (US Soil 
Conservation Service 1974a). After reclamation, slopes at MATl are northwest facing 
and average 4.5 degrees, with swales and depressions located over much of the area. 
A second small watershed (hereafter referred to as NEF 1) was selected as a reference 
site that is characterized as roughly 3.0-ha of contiguous forest (39? 35' 47" N; 78? 54' 
29" W) that has never been surface mined, but was selectively timbered nearly 20 years 
ago (Kuers, unpublished data). NEFl is located approximately 1.5-km to the west of 
MAT 1. The ephemeral stream draining NEF 1 functions as a tributary to Neff Run. 
Forest cover on NEF 1 is generally an unevenly aged deciduous stand consisting 
primarily of black cherry (Prunus serotina), black birch (Betula lenta), and sugar 
maple (Acer saccharum). Soils are mapped as Cookport very stony silt loam (US Soil 
Conservation Service (1974a). Similar to MATl, slopes at NEFl are also northwest-
facing and average 9.9 degrees. 
A third small watershed referred to as the East Branch of Neff Run (EBNR) was added 
for comparisons of hydrologic budgets. EBNR has a drainage area of l 04.2 ha and 
10 
contains the entire NEF l site. In contrast to the other two small watersheds (MAT l and 
NEFl), EBNR is drained by a perennial stream. Approximately 6 ha (6%) of the 
watershed has been surface mined and reclaimed, with the remaining 94% forested. 
Dominant species include black cherry (Prunus serotina), red maple (Acer rubrum), 
black birch (Betula lenta), and northern red oak (Quercus rubra) (Kuers, unpublished 
data). The average slope of EBNR is 8.0 degrees. Soils are classified as a combination 
of Cookport very stony silt loam and Buchanan very stony silt loam (US Soil 
Conservation Service ( 1974a). 
The Georges Creek basin (39? 35' N; 79? 00' W) located in western Maryland (Figure 4) 
was selected as the river basin that has undergone intensive surface mining and 
reclamation. Stream discharge for the 187.5 km2 (72.4 mi2 ) basin has been 
continuously monitored by a gage at Franklin, Maryland, operated by the United States 
Geological Survey (USGS) since 1929 (station# 01599000). The soils have been 
mapped by the US Soil Conservation Service (1974a) as belonging to the Gilpin-
Dekalb-Cookport Association. This association is gently sloping to very steep, well 
drained and moderately well drained. The soils are mostly very stony and are 
moderately deep over sandstone and shale. The average slope of the watershed is 9.5 
degrees (Figure 5). 
The Savage River watershed (39? 35' N; 79? 05' W) was selected as a reference site for 
the larger basin comparison and is located immediately west of Georges Creek 
watershed (Figure 4 ). The watershed has a drainage area of 127 km2 ( 49 .1 mi2 ), 
slightly smaller than the area gaged in the Georges Creek watershed. The USGS has 
11 
also been continuously recording stream discharge on Savage River, above the Savage 
River Dam since 1948 (station # 01596500). In contrast to Georges Creek watershed, 
1997 data from the Maryland Office of Planning indicates the watershed is 
predominantly forested (83%) with some agriculture (15%), development (<2%), and 
wetlands ( <l % ) (Hypio 2000). Soils within the Savage River watershed have been 
mapped as belonging to the Dekalb-Calvin-Gilpin and Calvin-Gilpin Associations (US 
Soil Conservation Service 1974b ). These soils are gently sloping to steep, moderately 
deep, well-drained and moderately well drained soils. The watershed is slightly steeper 
than the Georges Creek watershed with an average slope of 12.0 degrees (Figure 5). 
LULC data for the Savage River watershed are less detailed than those generated in this 
study for the Georges Creek watershed. Similar to Georges Creek, agricultural land in 
Garrett County (which completely contains the Savage River watershed) continues to 
decrease. From 1987 to 1997 alone, the total farmland in Garrett County decreased 
11 % from over 49,181 ha to 43,582 ha (USDA 2001). Based on data at the county 
level, LULC in the Savage River watershed has likely remained predominantly forest 
over the last 50 years with some decrease in agricultural land. Population in Garrett 
County has exhibited a 28 % net increase since 1940, increasing from 21,981 in 1940 to 
28,138 in 1990 (Forstall 1995). However, it is unlikely that the Savage River watershed 
has experienced this rate of growth since there are few population centers in the 
watershed. As such, development in the watershed over that last 50 years has likely 
remained at or below 2% of the total watershed area. 
12 
C. Field Hydrologic Measurements 
Characterizing the hydrological responses of the three small watersheds required a 
number of primary field hydrologic measurements, while the study of the larger 
watersheds required acquisition of historical hydrologic data. Hydrologic 
measurements made at the small watersheds included a) continuous time series of 
watershed discharge, b) continuous time series of precipitation depths over each 
watershed, and c) measurements of soil infiltration capacities. Historical hydrologic 
data for the larger Georges Creek and Savage River watersheds included hourly 
watershed discharge and rainfall depths measured at two locations within the 
watersheds. 
I installed stream gages in the three small watersheds to provide a continuous record of 
watershed discharge (Figure 6). Because all work dealing with stream channels in 
Maryland requires appropriate permits, the Maryland Department of the Environment 
provided me with permits for the installation of temporary monitoring and research 
devices (permit #'s 1999965783/99-NT-3220 and 200065922/00-NT-3206). Gages at 
NEF 1 and MAT 1 were operational prior to the commencement of the water year on 1 
October 1999. Because both of these watersheds lacked natural bedrock controls, I 
obtained and installed pre-fabricated, pre-calibrated "Montana" flumes (Figure 7) at 
each site. Each flume (manufactured by Free Flow Inc, Omaha, Nebraska) was 
constructed of ruggedized 9 mm (3/8 in) thick fiberglass (Figure 7), shipped to 
Appalachian Laboratory, and transported by truck to each site for installation. Flumes 
were anchored to 15 cm x 15 cm timbers and wingwalls buried in the streambed and 
bank. Each flume was equipped with stilling wells that were prefabricated directly onto 
13 
the flume at the factory. Each gage included a Stevens Type A instantaneous stage 
recorder equipped with a 15 cm (6 in) diameter painted copper float. Stage recorders 
were sensitive to water level fluctuations greater than 3 mm (0.12 in). 
I chose the appropriate flume dimensions based on the rational runoff method to 
accommodate runoff produced from rain events with a l O yr return period. At MAT 1, 
the gage is located in an armored diversion ditch that intercepts surface runoff from the 
watershed (Figure 3 ). The flume has a 91 cm (36 in) throat width with a peak capacity 
of 1.4 m3/ sec (Figure 7). At NEF 1, the flume has a 30.5 cm ( 12 in) throat width with a 
maximum capacity of 0.45 m3/sec (Figure 7). Throughout the study I checked stage 
records on a bi-monthly basis and returned charts to the lab quarterly. At the lab I 
digitized the charts showing instantaneous stage heights (obtained from Stevens Type A 
recorders) and subsequently converted stage heights to hourly discharge and average 
daily discharge based on the rating curves obtained from Free Flow, Inc. (Table 2). 
In all cases I checked stage records for errors and adjusted if necessary before 
generating discharge measurements. For example, winter months posed a particular 
problem for data collection since gages were located in remote terrain and could not be 
heated. At times, stilling wells were subjected to freezing temperatures, which 
prevented the float from responding to changes in streamflow. Between rain and melt 
events, the ephemeral streams were dry. On occasions when a rain or melt event was 
expected, however, stilling wells were thawed using heated water from a campstove to 
avoid missing a response. Streamflow data were missed at MAT l for a storm event that 
occurred on 3 and 4 August 2001 when the pulley on the stage recorder jammed after 
14 
the cover had been replaced. In addition, data were missed at NEF 1 between 1 August 
200 l and 30 September 2001 when a rodent derailed the float chain from the stage 
recorder. In these cases where data missed, I estimated streamflow using the IHACRES 
software. I only used the modeled data for computation of annual water balances and 
not in statistical analyses, however. 
At EBNR I used a continuously recording stage recorder and a natural channel control 
(Figure 8) to estimate discharge. The gage was operational on 25 July 2000. The gage 
consisted of a stilling well set into the stream bank to a depth of 1.5 m. A local metal 
shop fabricated a stilling well from 24" (ID) steel culvert pipe and a piece of sheet metal 
that formed the well bottom. The bottom of the well provided for a 15 cm sump below 
a PVC connection pipe that allowed flow from the stream channel to the well. The pipe 
was capped and perforated to allow streamwater to enter the well and 'still ' to the same 
elevation as the surface of the stream. A painted copper float with corresponding 
counterweight was attached to a Stevens Type F chart recorder, which continuously 
tracked the level of the water in the stilling well. A staff gage mounted to a galvanized 
metal rod was installed in the stream channel to provide a field check for the chart 
recorder and serve as a point of reference for stage-discharge calculations. A precision 
digital water level recorder (model 6541-4) manufactured by Unidata Australia was 
installed along side the Type F recorder on 28 May 2001 to reduce data post-processing 
time in the lab. 
A stage-discharge relationship was developed for the gage at EBNR. Flow 
measurements were made 11 times throughout the year at stage heights ranging from 
15 
9.2 cm to 22 cm. According to standard procedures, the stream was divided cross-
sectionally into cells where cell depth and width were recorded. Mean water velocity of 
each cell was measured at 0.6 times the water depth using a Marsh-McBimey 'FLOW-
MATE' model 2000 portable flowmeter. Instantaneous discharge was then calculated 
as Q = A * V, where Q is discharge in m3/sec, A is the cross-sectional cell area in 
meters, and V is the average velocity in the cell in m/sec. The rating curve that was 
used for EBNR between 25 July 2000 and 1 October 200l(r2=0.884; n=l l, Figure 9) is 
given in Table 2. 
I measured hourly precipitation using two Belfort universal weighing type precipitation 
gages (model 5-780-300) manufactured by Belfort Instrument Company. One gage was 
located in a clearing immediately adjacent to the flume at MATl (29? 35' 31.5" N, 78? 
53' 48.9" W) and a second identical gage was located in a clearing near the eastern end 
of MAT 1 (Figure 6). Gages were anchored to wooden platforms located approximately 
1 m above the ground surface. Rain gages operated on 8 d hourly charts and were 
sensitive to precipitation depths greater than l mm. During winter months, antifreeze 
added to the gages prevented freezing and allowed for recording liquid water 
equivalents (LWE) of snowfall. Unfortunately due to equipment failure, rainfall data 
collection did not start until 18 December 1999. However, little precipitation occurred 
during this period. For daily precipitation data lost during the 2 month period I obtained 
records from a National Weather Service cooperative observing station in Frostburg, 
MD (located approximately 5 km to the north of the site). 
16 
Measurements of soil infiltration capacity were made on each of three randomly located 
plots on each watershed. The measurements were made by temporarily installing 
double-ring cylinder infiltrometers (Figure 10) on 12 July 2001 within each of 3 
permanent 20 m x 20 m plots established on each watershed (Figure 3). Infiltrometers 
could not be installed in a strictly random manner on the plots, since the instruments 
required sites that were relatively level and stone free. Infiltrometers were constructed 
of 16 ga. sheet metal formed into an outer and inner ring. The water supply reservoir 
was constructed from a 45 cm length of cylindrical PVC pipe (30 cm I.D.). Each end 
was capped using 0.03 cm thick plexi-glass. The bottom end of the reservoir contained 
two short sections of 1.3 cm PVC pipe. One pipe extended out of the reservoir 2 cm 
more than the other allowing for water to fill the infiltrometer. The shorter pipe allowed 
air into the reservoir and kept the water at a constant depth in the infiltrometer. The 
reservoir was also equipped with a graduated tube (modified from a laboratory burette) 
to monitor declines in water level in the reservoir and provide a direct measurement of 
water entry to the soil over time. Water in the outer ring was maintained at the same 
level as the inner ring to avoid effects of differential head pressures. Infiltration 
capacity (mm/min) was determined as the rate at which water was added to maintain a 
constant water level in the center ring of the infiltrometer. For the purpose of this study 
and based on the relatively homogeneous soils, it was assumed that the sampling design 
yielded estimates that are representative of the entire watershed. 
D. Historical Hydrologic Measurements 
Hydrological characterization of the George's Creek and Savage River watersheds over 
the long-term was accomplished utilizing various data obtained from the USGS and the 
17 
National Climatic Data Center (NCDC). Data needs includes the following long-term 
records: a) daily precipitation depths at Frostburg and Savage River Dam from 1948 to 
present; b) hourly precipitation records for Savage River Dam 1948 to present; and c) 
hourly streamflow at Georges Creek (1929 to present) and at Savage River Dam (1948 
to present). Each of these data sets, with the exception of the mining history, has been 
quality checked and is available from the USGS and NCDC. 
Rain events were selected for intensive characterization of watershed responses. The 
fifteen most intense storms ( daily resolution) were included if they met the following 
criteria: a) events occurred between May and October; b) the events were of relatively 
uniform intensity over the entire watershed; c) and the events were relatively isolated 
from other storms. Storms were required to be of uniform intensity over the 
watersheds for unit hydrograph analysis. Events were considered uniform over both 
watersheds if daily precipitation measured by the stations at the northern and southern 
ends of the watershed differed by less than 20%. At the northern end of the watershed, 
a National Weather Service cooperative observing station at Frostburg has recorded 
historical daily precipitation for more than 50 years. At the southern end, the USGS 
station at Savage River Dam has recorded hourly precipitation for 54 years. Storms 
were also required to be isolated to eliminate multiple hydrograph peaks and dampen 
the effects of antecedent moisture conditions that can affect runoff generation from 
storm to storm. Rain events were considered isolated if no precipitation occurred within 
an arbitrarily chosen three days before or after a specific event. The hourly data from 
the USGS station at the Savage River Dam were ultimately used to estimate areal 
rainfall over each of the two larger basins. 
18 
Historical stream discharge records for Georges Creek and Savage River were obtained 
from the USGS for each of the selected rainstorms. It was necessary to reconstruct 
hourly discharge values from historical stage "strip" charts and stage-discharge 
relationships, since hourly stream discharge data were not archived digitally. Stage and 
stage-discharge data were obtained from archives at the USGS field office in La Vale, 
Maryland, with the exception of records dated prior to 1965 that were requested from 
the national archives located in Washington, DC. Records were then delivered to the 
LaV ale USGS field office where the stage records for selected storms were 
photocopied. Stage records were then digitized and converted to hourly stage values at 
the Appalachian Laboratory using "MDFLOW", a software program developed by K.N. 
Eshleman (personal communication). Based on historical rating curves reconstructed 
from rating tables, hourly stage values were then converted to instantaneous discharge. 
E. Unit Hydrograph Deconvolution 
Several unit hydrographs for the small watersheds were deconvolved using both the <!>-
index method (Chow, et al. 1988) and IHACRES. Each of the models was based on 
basic unit hydro graph theory, although IHACRES uses a more sophisticated 
mathematical approach. However, satisfactory model fits could not be obtained when 
IHACRES was used to generate unit hydrographs for historical rainfall-runoff data from 
the larger Georges Creek and Savage River watersheds. 
The <!>-index method was used for deconvolving unit hydrographs for the small 
watersheds for a thunderstorm that occurred on 6 August 2000 (this storm caused low-
lying areas in the Georges Creek to be flooded). Based on the <!>-index approach, unit 
19 
hydrographs were deconvolved assuming a constant rate of rainfall abstraction. Values 
for <D were calculated via trial and error by subtracting a value (<D) from the rainfall 
pulses that were believed to have contributed to direct runoff until a value for <D yielded 
an effective rainfall hyetograph that, when integrated, was equal to the area under the 
direct runoff hydrograph. 
Because of the sophistication of IHACRES, the software was used to unit hydrographs 
from June to October for water years ending 2000 (year 1) and 200l(year 2). These 
months were chosen based on the constraints of the model that allowed for best model 
fi t. The primary constraints of the model required that a) the subperiod of record started 
and ended at times when flow was at or near zero, and b) the subperiod of record 
contained no snowfall events (typically November to April in western MD). The basic 
modeling approach involved a non-linear component in the model that used a watershed 
wetness index that varied from zero to unity depending on the time since last rain. 
Effective rainfall (or the rainfall that was exported from the watershed as streamflow) 
was then modeled over a defined time step as a percentage (ranging from 0-100%) of 
the watershed wetness index. The second component assumed a linear relationship 
between effective rainfall and flow (streamflow subsequently decays exponentially 
following a unit impulse of effective rainfall). Unit hydrograph theory is then used to 
estimate total streamflow over the user-defined sub-period. 
F. Historical Land Use/ Land Cover Derivation 
Historical LULC for the Georges Creek basin was derived for several time periods 
based on historical aerial photographs. Aerial photographs beginning with 1938 were 
20 
obtained from the United States Department of Agriculture (Ben Cooper, Allegany 
County Soil Conservation District, Cumberland, MD). Certain years of photographs 
were used to provide detailed resolution of LULC change over the time periods used in 
this study: a) 1938, pre-surface mining; b) 1962 and 1982, pre to early surface mine 
reclamation; and c) 1997, widespread surface mine reclamation. Due to time limitations 
and labor intensity, LULC data for Savage River basin were obtained from the 
Maryland Department of Planning (http://www.mdp.state.md.us) other historical data 
were obtained from the U.S. Census Bureau (http://www.census.gov) and the U.S. 
Department of Agriculture (USDA 2001) . 
Aerial photographs for Georges Creek required georeferencing before LULC classes 
could be digitized from the photos. Approximately 120 photographs were scanned with 
a UMAX Mirage II scanner at 600 dpi and were archived on CD-ROMs. Images were 
georeferenced using the Imagewarp extension in Arcview? GIS v.3.1. In general, 
Imagewarp references photos using ground control points (a minimum of 10 in this 
study) and warps images using cubic convolution on a 4th order polynomial. 
Photographs were georeferenced to control points selected on USGS orthorectified 
digital topographic quadrangles. 
LULC was classified into nine dominant classes (Appendix II). These classes were 
based on the Multi-Resolution Land Characteristics Consortium (MRLC) classification 
scheme used by the U.S. Environmental Protection Agency (USEPA) 
(http://www.epa.gov/mrlc/). Appendix II lists each of the classes and criteria used 
during on-screen digitizing in ArcviewTM GIS v3. l. 
21 
G. Data Analysis 
Annual water balances for each watershed were based on a mass balance approach. 
Annual evapotranspiration was calculated as the residual of areal precipitation inputs 
and annual watershed runoff. Calculation of evapotranspiration by difference is a 
standard approach in hydrological studies where actual evapotranspiration cannot be 
readily measured (Dunne and Leopold 1978). The effects of changes in soil storage 
were assumed to be negligible based on the approach used by the USGS (Dunne and 
Leopold 1978). Annual water balances were calculated on a water year basis 
beginning on 1 October and ending on 30 September of the following year. 
Watershed response characteristics for each watershed were analyzed for differences 
both between watersheds on a storm by storm basis, as well as for changes over time 
from the onset of surface mine reclamation. Storm hydrographs were separated into 
baseflow and stormflow based on the separation method detailed in Dunne and Leopold 
(1978) (Figure 10-4 [b]). A paired t-test was used to test the hypothesis that the mean 
difference in watershed responses between the two watersheds is not significantly 
different from zero. Changes in hydrological response characteristics for each basin 
over time were tested using simple linear regression with storm date as the independent 
variable (1950 to 2000). 
22 
(A) 
Effective 
rainfall 
Abstracted 
rainfall 
1.0 
(B) 
0.8 
0.6 Unit effective rainfall 
0.4 
0.2 
0 1 2 3 4 5 6 7 8 9 10 11 12 13 1415 
Time step 
0.2 ----b ~ 
~ 
0.1 
0 1 2 3 4 5 6 7 8 9 10 11 1213 1415 
Time step 
Figure 1. Generalized model of constant rate of abstraction used in the <I>-index method 
(A) and non-linear decay used in IHACRES (B, from Littlewood et al. 1997). 
23 
.. 
Rainfall Effective rainfall Streamflow 
(mm/day) {mm/day) {mm/day) 
10s ._,-_ -------~ . ,_ _ Ct,Nr.w.,,d  
~lit 
u--:::1:~a~--------- ------l~n~:r~~:~~----t----__-  __, 
: loss module .,____.__ function (Unit 
1 ( ) Hydrograph) module 
~ 3 parameters . 
I :_ ___________________ J----~~~~~~~:~ ______ 1 J 
400 ,--------~ 
--To~lll --- TOIi! ...... ao,, - s. ...  
!S 10 1$ 
Air temperature Unit hydrographs Hydrograph separation 
(DC) (cumecs) {cumecs) 
Figure 2. Schematic representation of IHACRES used to deconvolve unit hydrographs 
at MAT l and NEF l adapted from Littlewood et al. ( 1997). 
24 
Figure 3. Locations of stream gages, watershed boundaries, permanent plots, and 
surface mined area (shaded) for the small watershed study located on Dans Mountain, 
Allegany County, Maryland. 
25 
N 
"'tE Savage River 
Georges Creek 
D Watershed Boundary 
1997 Land Use Class 1 -o 1- 2 3- 4 6 Kilometers - Low intensity residential 
- High intensity residential -- - -
Agriculture (hay1 1, asture/crop) 
- Forest (evergreen/deciduous/mixed) 
,li''ii,,~ Active mine 
.Jili.=!~J 
- Reclaimed mine 
- Abandoned mine 
-a Spoil pile Landfill 
Quary 
- Water/wetland 
Figure 4. Locations, watershed boundaries, and 1997 LULC for the Georges Creek and 
Savage River watersheds, Allegany and Garrett Counties of western Maryland. 
26 
Georges Creek 
CJ Watershed Boundary 
Slope (degrees) 
CJ0 - 5 
D 5 - 10 
D 10 - 15 
CJ 15 - 20 
19 ? 20 1 0 1 2 3 4 5 Kilometers 
- 20 - 25 
-25 - 30 
- 30 - 35 
- 35 - 40 
- 40 - 45 
Figure 5. Slope map of Georges Creek and Savage River watersheds, Allegany 
and Garrett Counties. 
27 
--
Figure 6. Equipment installed at MAT 1 for gaging watershed hydrologic inputs and 
outputs. Stream gage (Montana flume) is similar to that installed at NEFl. 
28 
-- . ???? . 
Throat width A B C D E l - ?---F 
36 88-3/4 36 61-7/8 27 18 I 17 
12 76-7/8 36 33-1/4 27 18 I 17 
(A) 
A ? 
Channel bank 
Wing wall 
C Rip-rap area 
-, 
1 rilling , 
\ ,w_ell  ~ Channel bank 
~ ..... 
D ~~ ... "" f~:?] ..\? :  6" X 6" timbers 
Backfill areas buried in channel 
bank 
--------------------------------------~---------------------------------------
(B) t Instrument F shelter L-.---.----J 
1.? 
- = Flow ! . ' ! i : ..... ? ............. j . ~ 
i 
= 
Figure 7. Plan view (A) and side view (B) of Montana flumes installed at the MAT l 
and NEFl watersheds (dimensions in inches shown in accompanying table). 
29 
(A) Stream Channel 
2" PVC Pipe 
18" Galvanized Culvert 
(Stilling Well) 
(B) r-- Instrument Shelter 
.-> &Recorder 
Figure 8. Plan view (A) and side views (B) of stilling stream gage installed on the East 
Branch of Neff Run (EBNR) located on Dans Mountain, Allegany County, 
Maryland. 
30 
Arithmetic Axes 
0.07 
0.06 ? 
0.05 
...,. -__ 
' rJ:J 0.04 
~6 
0 0.03 
0.02 ? 
- __ y "':' ~l.454x4.3167 
0.01 2 
R =0.8838 
0 
0 0.05 0.1 0.15 0.2 0.25 
Stage (m) 
Figure 9. Stage-discharge relationship for the gage installed on the East Branch of Neff 
Run (EBNR) during the period from 25 July 2000 to 1 October 2001. ' 
31 
Transparent Graduated 
Tube 
30 cm 
diameter 
PVC Pipe 
35 cmx 32 cmx 0.03 cm 
() 
Base Rests Plexi-glas Base (Top and Bottom) 
Atop Inner 
Ring of the 
' 
Double-Ring ' 
lnfiltrometer 
Reservoir for double-ring infiltrometer 
RESERVOIR 
1\ir 
Bubbles 0 
0 Filler 
- - - - -
Flow regulator for double-ring inJUtrometer 
Figure 10. Schematic of double-ring infiltrometer and water reservoir used to measure 
infiltration capacity (adapted from Eshleman 1985) 
32 
Table 1. Drainage area, slope, and elevation of watersheds used in comparative 
analysis. 
Area (km2) Sloee (degrees) Elevation (meters AMS) 
Min Max Mean Min Max Mean 
MATl 0.27 0 15 5 783 851 825 
NEFl 0.03 6 13 10 689 778 727 
EBNR 1.0 0 17 8 679 849 771 
GCRK 186 0 36 10 302 911 658 
SRIV 127 0 43 12 459 920 746 
33 
Table 2. Stage discharge relationships for the stream gages installed in the small 
catchment study. 
# Watershed Throat Width (in) Rating Curve 
1. MATl 36 Q (cfs)= 12 H 
1.5661 
a 
2. NEFI 12 Q (cfs)_- 4W H a c1.522 w 
II o.026J 
3. EBNR NIA Qccms)=3 l .45 [H]4 3167 ?
Q = discharge, Ha= head (depth) in feet, H = stage (height) in meters, W = flume throat 
width in feet. 
34 
Chapter III: RESULTS 
This study found a number of significant differences in the hydrological responses of 
watersheds subjected to LULC, specifically when comparing the surface mined and 
reclaimed watershed to one that was entirely forested. The small watersheds (MAT 1 
and NEF 1) responded similarly on a water year basis, but varied in their response to 
individual rain events. Storms at MATl produced significantly greater runoff ratios, 
total runoff, and peak runoff than at NEF 1. Lag times for the two small watersheds 
were similar for the events analyzed in this study. However, at the river basin scale 
(Georges Creek and Savage River), watersheds varied from each other primarily in the 
timing of response to rainfall events. Despite widespread LULC change, other 
hydrological response characteristics (runoff ratios, peak runoff, total runoff) exhibited 
little difference at the river basin scale when compared between the two basins or within 
the basins over time. 
Annual hydrographs for water years 1 (2000) and 2 (2001) at MATl and NEFl can be 
found in Figure 11. Hydro graphs for the two water years of streamflow indicate that 
MAT 1 tends to produce higher, narrower peaks than NEF 1. For the majority of the 
water year, both watersheds responded primarily to major rain events and snow melts 
and produced little to no baseflow between rain events. However, during wetter months 
(e.g., April and May) NEFl produced some sustained baseflow between storms. 
Annual water balances ( l October to 30 September) for all sites as well as the normal 
year for Georges Creek and Savage River are shown in Figure 13. On an annual basis, 
each of the watersheds produced similar runoff yields, although MAT 1 tended to 
35 
produce more total runoff than NEF 1. Over the two years, roughly 26% of the rainfall 
input to the MAT 1 watershed leaves as surface runoff, compared to 25% at the NEF 1 
watershed. Runoff yields varied slightly between years 1 and 2. MAT 1 produced 
similar annual runoff in both years, while NEFI decreased by 15%. Interestingly, this 
decrease at NEF 1 occurred despite an increase of 139 mm of precipitation in year 2 
making it a wetter than normal year (Figure 12). Compared to long-term records for 
Georges Creek and Savage River, these estimates of annual yield are close to normal 
values (Figure 13) although the 2001 water year tended to be somewhat wetter than 
normal. ET yields were slightly higher at the small watersheds for both of the water 
years and also higher than the long-term averages. Long-term annual runoff yields for 
Georges Creek tend to be 100 mm less than the long-term average for Savage River 
Watershed. 
Statistically significant differences were observed between runoff coefficients and total 
event runoff produced at MAT 1 and NEF 1. Mean runoff coefficients for were 
calculated for the eight largest storms for which data exist at both gages (Figure 14). 
Runoff coefficients were significantly higher at MATI than at NEFI (p ~ 0.03), on 
average by as much as 2.5 times (Figure 15). Runoff coefficients at MAT 1 averaged 
0.11 ranging from less than 0.01 to 0.26 (s.e. = 0.013) compared to NEFI where runoff 
coefficients averaged 0.04 ranging from no response to 0.13 (s.e. = 0.007). Runoff 
coefficients did not correlate with maximum rainfall intensity or total event rainfall (p ~ 
0.05). The mean difference in total runoff (MATI - NEFI) for the eight largest storms 
where data exist for both gages was significantly greater than zero based on a one-tailed 
t-test. (p ~ 0.05). MAT 1 yielded roughly three times more total event flow than NEF 1 
36 
(Figure 16). Total event runoff at MATl averaged 5.0 mm per event and ranged from 
no response to 14 mm. Response at NEFl was significantly lower averaging 1.7 mm 
per event ranging from no response to 5.1 mm. This trend of greater total runoff at 
MAT 1 was observed in all but one of the storms, which occurred on 31 July 2000 
(Figure 17). Total runoff at MAT 1 was significantly correlated to total event rainfall 
(r= 0. 794; p ~ 0.01; s.e. = 3.1; n= 10). Total rainfall explained 63% of the variation in 
runoff at MAT 1. At NEF 1 however, total rainfall explained less than 21 % of the 
variation in total runoff and a statistically significant relationship (r = 0.455; p ~ 0.05; 
s.e. = 1.8; n = 8) was not observed. 
Peak runoff rates for the eight storms investigated at MAT 1 were on average twice as 
large as those found for NEF 1, although the mean difference between the two 
watersheds was not significantly different from zero at the 0.05 level. Peak runoff rates 
at MAT 1 were consistently higher than at NEF 1, with the exception of one storm on 31 
July 2001 (Figure 18). MATl averaged 1.0 mm/h (S.E. = 0.18) ranging from less than 
0.1 mm/h response to 3.6 mm/h. On one occasion, peak runoff rates at MAT 1 reached 
5.9 mm/h. Peak runoff rates at NEFl were lower, averaging 0.5 mmJh (s.e.= 0.09) and 
ranging from no response to 1.6 mm/h. Peak runoff rates at MAT 1 were significantly 
correlated with maximum rainfall intensity (r = 0.670; p ~ 0.05; S.E.= 1.5; n= 10), and a 
linear regression model explained 45% of the variation in peak stormflow (Figure 19). 
In comparison, peak runoff rates at NEFl were not significantly correlated with 
maximum rainfall intensities (r = 0.543; p ~ 0.05; s.e:.= 0.6; n=8), and a regression 
model only 30% of the total variation in peak stormflow. 
37 
Timing of runoff response at NEF 1 and MAT 1 was also calculated to compare the 
responsiveness of the watersheds to the accumulation of the rainfall pulses. For the 
eight storms compared in this study, the mean centroid lags (center of rainfall mass to 
center of runoff) based on hourly rainfall and runoff data for each watershed averaged 3 
h, indicating no significant difference in the timing of response to rainfall. 
Two-hour unit hydrographs were deconvolved from rainfall and runoff observations for 
a thunderstorm occurring on 6 August 2000 using the <!>-index method (Figure 20A). 
IHACRES was also used to deconvolve 1-hr unit hydrographs for the growing season 
(June to October) for both years 1 and 2 (Figure 20B). For the thunderstorm event, unit 
hydrograph shapes for each watershed were strikingly similar. MATl peaked slightly 
higher and receded more steeply than NEF 1. Similar results were obtained using 
IHACRES (see Appendix III for model fit parameters) when unit hydrographs were 
developed over the entire growing season. Similar to the 2-hr unitgraph for 6 August 
2000, hydro graphs peaked slightly higher at MAT 1 than at NEF 1. Little change in unit 
hydrograph response was observed between year 1 and year 2. Unit hydrographs for 
both watersheds peaked approximately 0.1 mm higher in year 1 than in year 2. 
Soil infiltration capacity (or maximum infiltration rate assuming ponded water 
conditions) was also measured as an important variable influencing watershed 
stormflow response. As expected, steady-state infiltration capacities were lower at the 
reclaimed surface mine plots than on the forested reference watershed plots. This can be 
readily seen when examining the cumulative depth of water infiltrated at each plot 
(Figure 21). The point at which the curves become nearly linear provides an estimate of 
38 
the steady-state infiltration capacity of the soil. The reclaimed area at MAT 1 exhibited 
steady-state infiltration rates less than 1 cm/hr (n=3), or rates below the detection level 
of the infiltrometer. In contrast, soil infiltration experiments at NEF 1 yielded 
infiltration capacities that averaged nearly 30 cm/hr (n=2). For the 10 most intense 
storms at MAT 1 and NEF 1, soil infiltration capacities were exceeded in every case at 
MATl (Figure 21). However, at NEFl soil infiltration capacity was never exceeded by 
maximum hourly rainfall intensities. 
In addition to intensive measurements made for the two small watersheds, the physical 
and hydrological characteristics of Georges Creek and Savage River basins were also 
determined. During the last 60 years, the Georges Creek basin has undergone a wide 
range of LULC changes. Most change has occurred primarily within four LULC 
classes: surface mining, agriculture, forest, and development (Figure 23). Although 
some of these categories ( e.g., forests) have experienced minimal net change, others like 
agriculture and mining have undergone significant changes (Figure 24). Of most 
importance to this study are those LULC changes that have a direct effect on stream 
hydrology, such as those that alter evapotranspiration and imperviousness. LULC 
change statistics in the Georges Creek watershed for each time period are given in Table 
3. 
Surface mining and reclamation in the Georges Creek watershed was one of the most 
visible and potentially the most influential factor on hydrology in the watershed (Figure 
25A). Before the enactment of SMCRA in 1977, a limited number of surface mined 
lands underwent reclamation with a peak in surface mining and reclamation occurring 
39 
in the I 980's (John Camey, MD Bureau of Mines, personal communication). 
According to results obtained in this study, by 1982, 8.8% (1,649 ha) of the watershed 
had been surface mined and reclaimed. Total reclaimed area in the watershed has 
continued to increase such that by 1997 13% (2400 ha) of the watershed is reclaimed 
minesoils. Most of this land conversion came from two sources: agricultural lands and 
forestlands that were mined and reclaimed. In 1997, nearly 26 of the 105 mines in the 
watershed were within 50 m from surface waters, making stream channels highly 
susceptible to runoff produced from these adjacent mine lands. 
Active mining operations began in the late 1940's and peaked by the 1980's (Figure 
25B). In 1962, active mining represented less than 1%  of LULC in the watershed. 
Only 20 years later, active mining operations had increased by 700%. Available aerial 
photos show that surface mining peaked in 1982 with approximately 53 mines in active 
operation, representing nearly 3.5% of the total watershed LULC. Between 1938 and 
1982, active surface mines were primarily located on lands that were previously 
agricultural land or had been forested. In 1997, most active mines were either 
previously forested, or were mines that were still in operation from 1962 or had been 
abandoned and re-mined (Figure 25C). 
Agricultural lands were one of several land uses that declined over the 60 yr period 
(Figure 25D). In 1938, 32% of the watershed was pastured, cropped, or planted as 
hayland. Over the next several decades, agricultural lands steadily declined. By 1997, 
less than 10% of the watershed was in active agriculture. Agricultural lands were 
primarily lost to surface mining or had been abandoned to re-generate as forestland. 
40 
Overall change in forested areas remained relatively low over the study period (Figure 
25E). Nearly 65% of the Georges Creek watershed was covered in forests in 1938. 
Some forest regrowth occurred between 1938 and 1962 (74%), but was followed by a 
slight decrease through 1997 (70% ). Most forest regrowth occurred on abandoned 
agricultural lands. However, losses to surface mining and development offset any 
increases resulting in a net decrease in forest area. In 1997, a small area of strip mined 
land (4% of all forest lands) had returned to forest after mining. 
Overall development in Georges Creek on a per area basis, both commercial and 
residential, changed slightly from 1938 - 1997 (Figure 25F). The basin underwent a 
net increase in development of 2.2% (2.4 - 4.6). However, this represented 
approximately a doubling of developed area. Most of the change occurred in the low 
intensity residential category, which rose from 0.6 to 2.5%. 
The area of wetlands in the Georges Creek basin was found to be negligible. According 
to the National Wetlands Inventory (NWI), wetlands cover less than 1%  of the basin. 
Little difference was observed in response characteristics on a storm-by-storm basis for 
15 storms analyzed the Georges Creek and Savage River basin study. The primary 
diffei?ence in watershed response between the two basins was in the mean centroid lag. 
Georges Creek tended to respond on average three hours more quickly than Savage 
River (Figure 26). Mean runoff ratios for Georges Creek were significantly correlated 
to runoff ratios calculated for Savage River (Figure 27) (p ~ 0.001, r2 =0 .99, s.e. = 
0.0139, n = 15). In addition the slope of the regression line was not significantly 
41 
different from 1 nor was the intercept significantly different from zero (p <0.05). No 
significant difference (p ~ 0.05) was observed in the mean peak runoff or runoff ratios 
for the Georges Creek and Savage River basins, even though Georges Creek often (73% 
of the time) produced higher peak runoff than the Savage River. Mean runoff ratios 
were essentially identical (GC = 0.068; SR= 0.075). In addition, there was no 
significant trend in runoff ratio, peak runoff, total runoff, or centroid lag when 
examined for each watershed over time (p ~ 0.05, n = 15). 
42 
,,......, 
"7>, 
<SI 
"O 
s ~t~~ rr?~rr1;~~,  
s 
'--' 
<1.) 
~ 
C<:I I 
..c: 10 T1 -Ncfl l 1 .ca,:s - 80 .-:::  
<.) 
Q"'  : t , . I J Mm DNh I 11 I I , .. ' ' /\ ,1.1\ J' '  t I :: l 
~cf, ~ ...._,#~ ~~~ ....,~,SJ ~,SJ oc; .....,#~ ' ~~' .....,~~' ~~' '?""~ O'"' '?""~ O'"' 
+'-
l;) 
30 n 1. ,. II I HI 111111 ..,  I o 1 20 ,-.._ "7>, ,-.._ 25 <SI 
--;'>, "O 
C<:I 
"O 20 ~- - 40 s 
s s '--' 
s 15 
'--' - Precipitation 60 ? ?.:::J 
<1.) 
--Matl ?sC<.:I. ~ C<:I 10 80  
..c: ?c:; 
.<,..,)  <1.) 
Q 5 100 cl: 
o Il  . t l :J ,J ?!~ JAt : .J ' l, , , n, : uv\/i, ~" . /lJL1 :. J :1 120 
-.:?'?' -~~ ~P~ \.&:~ '<.t~ ov ov \#~"\, 
~~"\, \.&:~"\, ..., P' 
\'?>-v "?--~ "?--~ ov 
Figure 11. Average annual discharge (normalized by area) and daily precipitation at MATI and NEFI from 1 October to 30 
September 2001. 
1200 ??r------- - ------- - - --------- -----~ 
c:::::::J Water Year 2000 
c:::::::J Water Year 2001 
1000 -+- Long-Term Average -????-?--?- ?? ? 
800 ? 
400 
200 
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep 
Figure 12. Long-term average (1961-1990) cumulative precipitation observed at 
Savage River Dam station, Garret County, Maryland by water year. 
44 
Water Year Ending 2000 ? ET 
1200 D Runoff 
1000 - - - - - -
... 800 --"' :l> 425 :l> 564 
.? 600 - 674 630 639 534 
a ....___ 
~ 400 '--- ,__ 
'-- .___ 
~ 530 
200 
303 376 391 259 271 
0 
MAT! NEFI EBNR Georges Savage Georges Savage LT 
Creek River LT 
Watershed 
Water Year Ending 2001 
1200 
- - -
1000 -
- - - -
... 800 
-
-"' 425 :l> 802 768 443 564 :l> 
.? 600 
884 588 
:::i '---.___ 
~ 400 -
-
I--- -
200 - .___ 463 
530 
391 
270 304 318 188 
0 -
MATl NEFI EBNR Georges Savage Georges Savage LT 
Creek River LT 
Watershed 
Figure 13. Annual water balances for the small watersheds and larger basins for water 
years ending 2000 and 2001 and long-term (LT) averages. 
45 
1.6 -.-----------...... 
1.4 
R~  1.2 
a 1.0 
'-' 
~ 0.8 - .. 
0 
i:::: 
~ 0.6 
~ 0.4 
Q) 
o... 0.2 
0.0 -+--__.__-~- --.--'----'-~ 
TRIBMATI TRIB NEFI 
6.0 
,,...__ 5.0 T 
..~.... 
1 
, 4.0 -
~ 
g 3.0 
~ 
ca 2.0 T 
.L 
~ 1.0 
0.0 
TRIBMATl TRIB NEFI 
0.14 
.... 0.12 T 
?u~  0.10 1 
s 
Q) 0.08 
8 0.06 
~ 
0 T 
i:::: 0.04 J.. 
~ 
0.02 
0.00 
TRIBMATl TRIB NEFl 
Figure 14. Mean watershed response characteristics(? s.e.) for the eight most intense 
storms at the MAT l and NEFl watersheds between May and October 1999 to 
2001. 
46 
0.40 0 -~ 0 
0 0 0 0 --?[]--?-~-
0 0 0 0 
.(."."..j  .N... . ("]
er, ..... 
 - .~.... ~~.....  N -gc...-.,.  f.c..-.,.   
0.35 - ..... 
..... ~ ~ 
."..'"..  .'.C..i.  N N ..--< ..--< in N00 00 00 ... .. .N... . ....-.-.<.  .c.<...,  .c..-.,.  in s ..--<
in '-0 .....  ..... - 20 ?' t- ?' '-0 t-
0.30 -- 40 
....... 
C 
~ 
?u 0.25 DMATl 
~ 
~ ? NEFl - 60 i 
0 
U 0.20 - D Max Hourly Rain 
lt:: - - 80 ~ 
0 61 Total Rain 
C ?? 
~ 0.15 - ~ 
100 
0.10 -
- 120 
0.05 -
~ - 140 
0.00 -- ~Z,--L-'-=l,._L_ 
1 2 3 4 5 6 7 8 9 10 
StormRank 
Figure 15. Runoff coefficients and maximum rainfall intensities for the 10 most intense 
storms at the MAT 1 and NEF 1 watersheds between May and October 1999 to 
2001. 
47 
- 0 0 0 0- 0 
0 0 0 
N N N 
-N- - N-- -
N N ---- <") 
-I/-"-) --- --.. ..... \0 ?' t- -\0 ---- - 20 tw 
40 
,,----.. 
? DMATl ,,----.. 
___, l5 - ? NEFl - 60 
-' 1-, 
..c:: 
ti:: 
0 
i:::: ? Maxlntens ~ 
;:::::l '--' 
0::: ? Total Rainfall 80 :::::::I 
?fl lO - <S 
0 .s 
E-< cd 10( 0::: 
5 - - 12( 
- 14( 
0 -
l 2 3 4 5 6 7 8 9 10 
Storm Rank 
Figure 16. Total event runoff and maximum rainfall intensities for the 10 most intense 
storms at the MAT 1 and NEF 1 watersheds between May and October 1999 to 
2001. Excludes 2 storms at NEFl when animals disrupted the gage. 
48 
16 
14 -
D 
12 
D 
s 
? 10 y = 0.1728x- 2.0977 
~ R2 = 0.6319 
0 
C 
? 8 
B 
Cl) D 
s 6-
0 
f-, 
? 
4 
DD D ------------?-
2 --- -- .. - ? - ?-?- .. - -- .. - .. - -;?;0 .0364x+ 0.3348 
R2? =0.2074 ? 
0 
0 10 20 30 40 50 60 70 80 
Total Rainfall (mm) 
o TRIB MATl ? TRIB NEFl 
--Linear(TRIB MATl) -----Linear(fRIB NEFl) 
Figure 17. Relationship between total rainfall and total stormflow for the 10 most 
intense storms (May thru October) from 1999 to October 2001 at MATl (?) 
and NEFl (?) 
49 
10 -- 0 0 0 ;; 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
~ N N ~ 
9 - ~ ~ 
ON u!N  
~ ~---
~ 
io N N  ~ V) 10 
00 00 N 00 ;;:; ~ -0\ ? N -'? r'.-:! 0\ ~ -r:::: 
8 - 20 
,-... 
7 30 '-;E'  
OMATI 
,-... 6 - ? NEFI 40 ! 
0 
'E D Rainfall ?;;:1 i= 
! 5 50 ~ .El 
~ ;? 
0 4 60 
i= 
:::i -~ 
i:.::: i:.::: 
.;,: 3 - 70 s ~ 
11) :::i 
0... .? 
2 - 80 >< ~ 
::E 
90 
z<C  <C 
0 z - 100 
2 3 4 5 6 7 8 9 10 
StonnRank 
Figure 18. Peak runoff and maximum rainfall intensities for the 10 most intense storms 
(May thru October) from October 1999 to October 2001. Excludes two storms 
in August 2001 due to equipment problems at NEF 1. 
50 
7.0 
6.0 D 
y = 0.0942x - 0.2026 
-;:;- 5.0 R2 = 0.4483 
1 
~ 4.0 -
0 
? D 
0 
<Zi 3.0 ___ . -?l'.l 
--"" 
"' _. - ?- ? -y?  = 0.0649x - 0.4604 
0"..'. _ __ .. _ .. .,... R2 =0.2958 
2.0 _.. -~-? 
__? . ---,,,.-- .. -?-
1.0 
_g-- ?? -?-
0.0 -- - -- --? 
0 JO 20 30 40 50 60 
Maximum Rainfall Intensity (mm/hr) 
D TRJB MAT! ? TRIB NEFI 
--Linear (TRIB MAT!) - ? - ??Linear (TRIB NEFI) 
Figure 19. Relationship between maximum hourly rainfall intensity and peak 
stormflow for the 10 most intense storms (May thru October) from 1999 to 
October 2001 at MATl (?) and NEFl (? ). 
51 
0.3 5 
? 
\ 
,..-.....  ..... .- ~ 4 ,.-.....Effective Rainfall . . ::l ::l 
0.. \ 
.s 0.2 :. (A) 
0.. 
?? MAT! 6-Aug 2000 .i. :.:.:.  
E 3 E 
E....  ? 
~ ?? E ? NEFI 6-Aug 2000 .... ~ 0.. 2 0.. 
..~__ 0.1 , ? .?.  ? ..__, ~ ~ 0.....0  I 
..cc<::l  
?? - ..rc:::  
(.) (.) 
. :!: --.----- -I ?------ - . :!: 
0 O ? - I- - - I -,- 1-- I - -I t t - -? - - 0 0 
0 2 3 4 5 6 7 8 9 IO 11 12 
Hours 
0.5 - - - ?----~ - - --~- -- -
,.-......  
::l 
0.. ? MATI 2000 
.5 0.4 (B) ? NEFI 2000 
~ ?-? ? ? MAT! 2001 
t 0 .3 
0.. . . . a. ? - NEFI 2001 
I 0.2 
~ 
0.0 
~ 
..c:: i:! 
..~...  0.1 ' ' + 
0 ' 
0 ?? , , ~~ . . G ? ... -r.1 ? :...:.? -
0 2 3 4 5 6 7 8 9 10 11 12 
Hours 
Figure 20 . Two-hour unit hydrographs (A) for a thunderstorm (6 August 2000) and 
one-hour unit hydrographs from June to October 2000 and 2001. 
52 
NEF-1 (A) 
100 -
I 90 - ~ _- I 
"Cl 80 - .... .  ! 
~ 
crj 70 -
~ 
~ 
.5 60 
,B 50 
0.. 
Q) 
"Cl 40 
Q) 
-~~  30 -crj s 20 
;:j 10 
u 0 
0 2 4 6 8 10 12 
Tirre from start of infiltration (min) 
(B) 
MAT-1 
3.0 -,-----,-----,-----,-------~- - -~ 
. -e- Plot 1 . 
-G- Plot 2 
-1r-Plot3 
0 10 20 30 40 50 
Tirre from start of infiltration (min) 
Figure 21. Cumulative depth of water infiltrated at each of three plots at MAT l (A) and 
NEFl (B). 
53 
60 -i----------- - --------- ---~ 
~--------------------------------------------------
-
50 
D Maximum Rainfall Intensity 
40 ....... Soil Infiltration Capacity (MA Tl) 
-- Soil Infiltration Capacity (NEF I) 
-
30 
--; 
..... 
..c:: 
E 20 -
6 
-
10 
.......................... _.. .................... ~- .......... -..... . 
2 3 4 5 6 7 8 9 10 
StonnRank 
Figure 22. Maximum hourly rainfall intensity versus soil infiltration capacities at 
MATl and NEFl. 
54 
N 
+ 
1938 
/ / Strums 
Land Un CltHU 
- Low Intensity residential 
- Hgh Intensity residential 
D Agriculture (hay/paeturelcrop) 
- For11t (avergrten/dtclduoue/mlxed) 
D Active mine 
- Reclaimed min? 
- Abandoned mine 
- Spollpllt 
- Landllll 
1982 1997 
Figure 23. Results of LULC change over time in the Georges Creek watershed in 
western, Maryland from 1938 to 1997. 
55 
80 
60 ~Forest 
-tr-- Agriculture 
-a-Reclaimed Mines 
'O 50 -
Q,) -o-- Active Mining 
..c:: 
rJJ 
i.. - - ? -- Abandoned Mines 
~ 
~ - --? - - Development 
.-.
~...  40 -
0 
=Q,)  
(,J 
i.. 
Q,) 
~ 30 
20 -
10 
--- -- - ? -? ???????-0 
1938 1962 1982 1997 
Figure 24. Primary LULC changes for each of the major classes in the Georges Creek 
basin, 1938 to 1997. 
56 
35 Reclaimed Mines (A) 35 31.9 Agric ultural (D) 
30 - 30 -
25 - 25 
i= ~ i::: 
0) 20 0) 19.2 
.(....).  12.8 .
(....).  20 
0) 
0.. 15 
0) 
0.. 
8.8 15 12.2 
10 9.5 
5 - IO 
0 .0 0.0 
0 D D 5 -
1938 1962 1982 1997 1938 1962 1982 1997 
35 (B) - (E) Active Mines 85 Forest 
30 ? 80 -
73.9 
25 75 - - 71. l 70.1 
i= ~ - -20 ~ 70 -0) 
.(....).  ~ 65.3 
0) 
0.. 15 & 65 - -
10 - 60 -
5 3.5 2.1 55 -
0.0 0.5 
0 50 -
1938 1962 1982 1997 1938 1962 1982 1997 
35 - Abandoned Mines (C) 35 D evelopment (F) 
30 30 
25 - 25 
i= 20 - i= 0) 0) 20 
.(....).  ~ 
~ 15 0) 0.. 15 -
IO 10 
3.3 4.6 5 - 3.3 2.4 2.5 
0.0 0.9 0.6 
5 
0 0 - 7 
1938 1962 1982 1997 1938 1962 1982 1997 
Figure 25. LULC changes by class in the Georges Creek watershed (1938-1997). 
Values represent percentage of the entire watershed. 
57 
GC & SR Centroid Lag 
45 
40 -
? 
35 0 
30 -
? 
~ 25 - .. y = -0.000 Ix+ 24.253 ::I 2 0 R - 0.0048 ::r: 20 - ?-?-? :,.?- ~----- -- .4 . _______ ___ _ ..
- ?- ?-?-A-A .. . 
? A -?-?- ?-?~?- ?-?-?n-?-
15 - D ? ? D y = -0.0002x + 24.495 A 
A D R
2 =0.0186 ? ? 
10 A A 
? 
5 ? 
0 
1949 1954 1960 1965 1971 1976 1982 1987 1993 1998 
StonnDate 
D Georges Creek ? Savage River 
- ? - -- Llnear (Georges Creek) --Linear (Savage River) 
Figure 26. Centroid lag and trend lines for the 15 selected storms in the Georges Creek 
(?) and Savage River ( A) watersheds ( 1952-2000). 
58 
0.50 
0.45 
y = l.1046x - 0.0002 
0.40 
2 
R = 0.9863 1: 1 
I--
<l) 0.35 
> 
~ 0.30 
<l) 
W0 .25 
~ 
r/) 0.20 
0.15 
0.10 
0.05 
0.00 -
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 
Georges Creek 
Figure 27. Correlation between runoff ratios for Georges Creek and Savage River 
watersheds for 15 selected storms of relatively uniform intensity and distribution 
( 1952 - 2001) 
59 
Table 3. Number of LULC patches, % of watershed, mean area, and total area in each LULC class for the Georges Creek 
watershed, western Maryland (1938-1997). 
--
1938 1962 1982 1997 
Land Use Class Area (ha) Area (ha) Area (ha) Area (ha) 
(LU code) # % Mean Total # % Mean Total # % Mean Total # % Mean Total 
Development 
low intensity (2) 13 0.6 8.9 108 36 0.8 4.4 158 33 1.6 9.0 296 37 2.5 12.6 464 
High intensity (3 ) 8 1.8 43.0 343 8 1.6 38.0 304 8 1.8 41.5 331 8 2.2 50.8 406 
Agriculture (5) 152 31.9 39.4 5992 389 19.2 9.3 3608 111 12.2 20.7 2294 132 9.5 13.6 1792 
Forest (10) 74 65.3 165.9 12268 97 73.9 143.0 13880 24 71.1 556.6 13359 39 70.1 337.7 13164 
0\ 
0 Surface Mining 
Active (13) 0 0.0 0.0 0 31 0.5 5.9 97 53 3.5 12.3 651 47 2. 1 8.5 399 
Abandoned (14) 0 0.0 0.0 0 124 3.3 5.0 617 20 0.9 8.2 164 25 0.6 4.2 105 
Reclaimed ( 15) 0 0.0 0.0 0 0 0.0 0.0 0 72 8.8 22.9 1649 105 12.8 23.5 2400 
Spoil (16) 33 0.4 2.2 70 97 0.6 1.2 116 14 0.2 2.6 36 26 0.1 0.8 19 
Landfill (17) 0 0.0 0.0 0 0 0.0 0.0 0 0 0.0 0.0 0 1 0.2 31.2 31 
Total 100 18781 100 18781 100 18781 100 18781 
Chapter IV: DISCUSSION 
Results from this study indicate that surface mining and reclamation has the potential to 
impact the hydrological responses of watersheds, especially small watersheds where a 
relatively large percentage of the watershed area has been mined. At the small 
watershed scale, the watershed response to total rainfall was significantly different from 
the response of a forested reference watershed that was otherwise physically similar. 
However, when comparing only the unit response of each watershed to effective rainfall 
inputs, both catchments behaved very similarly. In general, the differences in 
hydrological response between the partially surface mined and reclaimed watershed and 
the entirely forested could be underestimated based on a number of factors. As 
watershed scale increases however, these differences in hydrologic response observed at 
a small scale are apparently much less detectable due to either natural variations, or 
attenuation as water is routed through larger channels downstream. 
On an annual basis, the surface mined and reclaimed watershed produced similar water 
balances to that of the forested watershed. This occurred regardless of higher total 
event runoff at MAT 1. NEF l however produced baseflow during the spring months 
when MAT I did not, making the overall annual water balances similar. Annual runoff 
yields and evapotranspiration rates are similar to the zero order watersheds in a study by 
Burges et al. (1998) where they compared a previously logged and reforested watershed 
to a suburban watershed. In that comparison the runoff yields were lower at their 
forested reference site (12 %) than the suburbanized site (44%). In comparison, in this 
study annual runoff yields at the reference site were approximately 25% compared to 
26% at the reclaimed mine. 
61 
While results of this study indicate that the watersheds produced similar responses when 
averaged over an entire water year, the responses to individual storms were significantly 
different, primarily in the amount of rainfall abstracted (i.e., water that never reaches 
the watershed outlet) during any given storm. The amount ofrainfall abstracted was 
greatly reduced at the surface mined and reclaimed watershed relative to the forested 
reference site. This decrease in rainfall abstraction at MATl has a number of 
consequences including significant differences in peak runoff, runoff ratios, and total 
runoff. Runoff ratios and total runoff for rain events occurring at the surface mined and 
reclaimed watershed were significantly higher that at the forested watershed. This 
finding is supported by the results of Schueler (1994) who observed that increasing the 
imperviousness of a watershed by 35%-50% can result in as much as a 20% increase in 
runoff. After the rate of rainfall abstraction has been exceeded and effective rainfall 
was produced, both watersheds responded in a surprisingly similar manner to a unit 
impulse of effective rainfall. Both watersheds produced unit hydrographs that were 
nearly identical in both the timing and geometric shape. As each watershed was 
subjected to a unit of effective rainfall, both watersheds exported that volume (or pulse) 
of rainfall at very similar rates. 
There are a number of possible explanations for the observed differences in rainfall 
abstraction at each of the small watersheds including a) altered soil characteristics and 
b) altered LULC (Hornbeck et al. 1970, Burton 1997). At MAT 1, it appears that 
surface mining and reclamation likely resulted in highly compacted soils with high bulk 
density and low infiltration rates. Furthermore, clays (at least at MATl) brought to the 
surface during reclamation have the potential to inwash, or clog surface pores in the 
62 
soils, further reducing infiltration rates. In addition to altering soil hydrologic 
properties, intense vegetation removal nearly 15 years ago has resulted in a cleared area 
that is now covered with dense grasses capable of influencing evapotranspiration and 
throughfall rates. 
One of most obvious explanations for decreased rates of abstraction at MAT 1 is the 
physical soil properties. Extreme soil compaction has greatly suppressed infiltration 
capacities at MAT 1 to less than 1 cm/hr, compared to nearly 55 cm/hr at the reference 
watershed. These low infiltration capacities at the MAT 1 plots are typical of results 
from several other studies on reclaimed surface mines that have found infiltration rates 
near or less than 1 cm/hr (Chong and Cowsert 1997; Guebert and Gardner 2001). In 
addition to low infiltration capacities, soil bulk density was also greater as a result of 
compaction. At MAT 1 soil bulk density was 1.43 g/cm3 compared to 0.98 g/cm3 at 
NEF 1 (Currie et al. unpublished data). The low infiltration capacities at MAT 1 and the 
fact that in all the 10 storms analyzed maximum rainfall intensities at MAT 1 
consistently exceeded the maximum rate at which rainfall could infiltrate the soil 
surface suggest that the dominant rnnoff mechanisms at MATl are likely a) Hortonian 
overland flow and/orb) saturation overland flow. Saturation overland was more likely 
to be an important factor in rnnoff generation for storms of longer duration or storms 
where antecedent moisture conditions were high. The small amount of rainfall that was 
abstracted during these storms was most likely due to storage in the various 
depressional ponds that developed as the surface mine subsided. 
63 
In contrast, soil physical properties were very different at the NEF 1 watershed, which 
abstracted considerably more rainfall per storm than the MAT 1 watershed. Even 
though NEFl yielded a unit hydrograph very similar to MATl, effective rainfall could 
not have been exported as surface runoff via the same mechanism. Recall that soils at 
NEF 1 are deep, well-drained, and capable of quickly abstracting rainfall. In fact, 
according to the infiltration capacity curves at NEF 1, a rainstorm of nearly 55 crn/h 
would be required to exceed the soil infiltration capacity (Figure 22)! During this 
study, the soil infiltration capacity at NEFl was never exceeded suggesting that after the 
soil was wetted soil water was quickly routed to the stream channel via a subsurface 
stormflow mechanism. This mechanism is well documented in the literature (Whipkey 
et al. 1965, Hewlett and Hibbert 1967, Freeze 1974, Beven and Germann 1982). In 
fact , in order for the unit hydrographs to be as similar as they are, subsurface stormflow 
at NEF 1 must be routed as quickly to the stream channel as it is via overland flow at 
MATl. 
One unusual observation in the relationship between infiltration capacity and runoff 
generation in this study requires some discussion. In particular, the cumulative depth of 
water infiltrated into the soil typically increases at a decreasing rate (i.e. the infiltration 
capacity is very high at the beginning of the curve and decreases over time) due to the 
initial negative pressure head caused by capillary pressure. At NEFl however, 
infiltration capacity curves tended to remain linear throughout the experiment, with no 
decrease in infiltration rate. One likely explanation for this anomaly is that 
measurements of infiltration capacity were made in a year that was relatively wet. At 
the time of measurement total precipitation for the year (January to June) was nearly 74 
64 
mm (3 in) above average. In addition, soils had been wetted by an intense thunderstorm 
2 days earlier that had added 20 mm (0.8 in) of rainfall. This thunderstorm may have 
been sufficient enough to wet the soils and reduce the capillary pressure head. This 
anomaly, however, does not affect the overall interpretation of the large difference in 
infiltration capacities observed at the two sites. 
In addition to the effects of soil properties on runoff generation, the difference in LULC 
is the second most obvious difference between the two watersheds that is responsible 
for observed differences in the hydrologic responses of MAT 1 and NEF 1. Roughly 
46% of MAT 1 has undergone extensive vegetation removal and is currently covered by 
tall grasses or patchy areas of bare soil. On one hand, woody vegetation removal 
normally decreases evapotranspiration back since roots are not longer transpiring, 
causing more water to leave the watershed as surface runoff. On the other hand, it 
should be noted that the lack of woody vegetation at MAT 1 increases exposure to wind 
and solar radiation, which is undoubtedly increasing the evaporative demand at MAT 1 
(Swift et al. 1975). Therefore, although transpiration rates may be decreased by 
vegetation removal, evaporation rates may make up the difference in loss to the 
atmosphere. This would explain the similarities in evapotranspiration rates in the 
annual water budgets for the two watersheds. 
Observed differences in watershed response may be even greater when considering a 
number of factors that make estimates of the hydrologic response at MAT 1 
conservative. It is suggested that these estimates when compared to the reference 
watershed are conservative based on the fact that a) MAT 1 is not entirely surface mined 
65 
and reclaimed; b) the slope of MAT 1 is not as steep as NEF 1 and; c) MAT 1 is larger 
than NEF 1. It is expected that the runoff differences would be even greater if MAT 1 
was as steep as NEF 1. Furthermore, estimates of runoff from the reclaimed area of 
MAT 1 are probably low, since watershed outflow is actually an average of runoff 
produced on the two different types of landcover in the watershed (4 5% reclaimed and 
55% percent forested). Although one would expect the reclaimed area to produce more 
runoff than the forested area, future research should be aimed at resolving uncertainties 
in these two contributing areas. 
It should be noted that although no statistically significant (p ~ 0.05) difference was 
detected in the timing of response (c entroid lag) between the two small watersheds it is 
likely an mtifact of the stream gage resolution. Stream gages used in this study were 
only accurate to? 1 hr, which is likely too coarse resolution for these watersheds. 
Considering that these watersheds respond on average within 3 hr of the center of 
rainfall mass, differences in response times may only be detectable with gages that can 
monitor flow changes on the order of minutes. 
It was expected that Georges Creek would be more responsive to rainfall events than 
Savage River based on the differences in LULC and amounts of impervious area. 
However, characteristics such as peak runoff, total runoff, and runoff ratios were not 
significantly different between the two watersheds. Results of the larger basin (Georges 
Creek and Savage River) study were different than originally hypothesized, regardless 
of a 13% increase in surface mine reclamation in the Georges Creek watershed since 
1977. Based on an analysis of 15 storms, Georges Creek was found to be substantially 
66 
flashier than Savage River, but no statistically significant difference in the hydrological 
response characteristics of the two basins could be detected. In addition, no significant 
trends could be found over time. 
On average, the Georges Creek watershed responded 3 hours more quickly (center of 
mass to center of runoff) than the Savage River watershed. This is a particularly 
interesting considering it is the only watershed response characteristic that significantly 
differs between the two watersheds. In fact, this difference may be even greater and 
more significant when considering that observations on the timing of rainfall were made 
in the Savage River watershed. Since the majority of storms approach from the west 
(prevailing wind) and cross over the Savage River watershed before reaching Georges 
Creek, the Savage River should respond sooner than the Georges Creek watershed (all 
else being equal). Storms could conceivably ruTive in the Georges Creek and hour or 
more after occurring in the Savage River watershed. This estimate could mean that 
Georges Creek may actually respond nearly 4 hours more quickly than the Savage River 
watershed. 
There are several possible explanations why the hydrological effects of surface mining 
and reclamation observed at the small watershed scale were not observed at the larger 
basin scale. Two possible explanations deal with a) data quality and availability and; b) 
LULC heterogeneity and other physical watershed properties. Based on the criteria 
used in this study to select representative storms, the availability of storms was 
restricted to 15 events that occurred over each basin. Additional storms would improve 
statistical power. In addition, this study would benefit from increased spatial resolution 
67 
for estimates of areal precipitation. The main rain gage used in this analysis was 
located in the Savage River watershed. These watersheds are located in mountainous 
terrain subject to orographic effects, however. In addition, Georges Creek basin is 
separated by the gage at Savage River Darn by Big Savage Mountain. Since many of 
the storms in this area approach from the west, Georges Creek may experience a 
rainshadow effect from Big Savage Mountain, causing actual precipitation depths to be 
lower than those measured in Savage River. A comparison of the long-term water 
balances for the Georges Creek and Savage River basins supports this observation as 
well. In general, Georges Creek tends to yield over 100 mm less runoff than Savage 
River, probably a result of less rainfall occurring within the watershed or from loss to 
the Hoffman Drainage Tunnel. 
The heterogeneity in LULC may also have confounded analyses aimed at correlating 
LULC change with hydrological responses. This study is one of few that investigate the 
long-term changes in streamflow trends with changes in watershed LULC. Gebert and 
Krug (1996) performed a similar analysis in Wisconsin's "Driftless Area" (non-
glaciated) where they investigated trends in streamflow characteristics for watersheds 
with various LULC histories. The study found that in forested areas no trends were 
observed in streamflow characteristics. However, the study also found that in 
predominantly agricultural areas, annual flood peaks increased while annual seven-day 
low flows decreased. The authors attributed this change to improved agricultural 
practices that decreased compaction and runoff from agricultural lands. Similar 
responses may be occurring in the Georges Creek watershed that effectively 
counterbalance any increases in runoff observed at the small watershed sites. At the 
68 
same time forest regeneration has occurred, agricultural land has decreased, and 
stormwater management is improving in the basin, all of which are likely reducing 
runoff. 
Another possible explanation for the lack of difference in watershed responses despite 
very different LULC may be the differences in watershed physical properties. In 
general, Savage River is a steeper watershed (12 degrees compared to 9.5 degrees in 
Georges Creek). Infiltration tends to decrease and overland flow tends to increase with 
increasing slope (Dingman 1994 ). In addition, Savage River may be responding more 
readily to rainfall than Georges Creek due to its slightly higher drainage density (0.77 
and 0.69, respectively) which is a measure of how efficiently the watershed is drained 
by streams. The Hoffman Drainage Tunnel, constructed in 1907 drains groundwater 
from the Georges Creek watershed and discharges into the Braddock Run watershed (an 
adjacent basin). The tunnel is approximately two miles long and drains approximately 
36 km2 of the Georges Creek watershed (Maryland Department of Natural Resources 
200 l ). As a result, estimates of annual runoff as well as event runoff for Georges Creek 
are lower due to loss to the Tunnel. Unfortunately, useful estimates of how much water 
that is diverted are lacking. 
One way to address the non-uniform spatial distribution of rainfall would be to 
incorporate NEXRAD (NEXt Generation RADar) data, formerly known as WSR-88D 
(Weather Surveillance Radar-1988 Doppler). NEXRAD utilizes a suite of algorithms to 
generate real-time precipitation depths over an area with spatial resolutions from 4 to 8 
km (French and Krajewski 1994). Cun-ent NEXRAD technology is capable of sensing 
69 
rainfall at resolutions as fine as 1 km2? This data could then be coupled with a DEM in 
GIS where a cell-by-cell comparison could be conducted between elevation and 
precipitation. NEXRAD has a number of limitations, however. Some research has 
indicated that radar underestimates precipitation when compared to traditional rain gage 
estimates (Smith et al. 1996). In addition, the ability of NEXRAD to provide 
information about historical rainfall distribution is limited. Smith and Krajewski (1991) 
argue that only rain gage and radar data from the same time period be used for this 
relationship. Even though the actual quantities may not be precise, NEXRAD would 
provide useful information on the spatial patterns of precipitation within the watersheds. 
Improved spatial resolution of rainfall data could conceivably increase the ability of 
IHACRES to model unit hydrographs for each of the watersheds as well. In fact, 
IHACRES may be the best justification of the need for more spatially explicit data, 
specifically areal rainfall estimates. Initially, it was proposed that IHACRES would be 
used to deconvolve unit hydrographs to examine differences in watershed response due 
primarily to LULC change. However, IHACRES generated fatal errors while trying to 
model streamflow, suggesting that rainfall estimates for the watershed may be 
inadequate. This has been observed in other studies that have found IHACRES to be 
highly sensitive to the density ofrain gages in a watershed (Hansen et al. 1996, 
Schreider et al. 1996, Andreassian et al. 200 l ). Incorporating NEXRAD into a model 
that generates spatially explicit estimates of rainfall depths may increase the ability to 
scale hydrologic responses observed at the small catchments to the larger basins. 
70 
Chapter V: CONCLUSIONS 
The primary goal of this study was to determine whether small watersheds subjected to 
mine reclamation practices display a stormflow response to rain events that is different 
from those displayed by similar watersheds that are covered by typical second-growth 
forests . A secondary goal was to investigate whether intensive surface mining in the 
Georges Creek watershed of western Maryland has appreciably altered the stormflow 
response at the larger river basin scale. Based on intensive field_hydrological 
measurements at the small watersheds, LULC data obtained from digitized aerial 
photographs from 1938 to 1997, and historical precipitation and streamflow data, results 
from this study indicate that surface mining and reclamation can impact the 
hydrological responses of watersheds, especially small watersheds where a relatively 
large percentage of the watershed area has been mined. 
At a river basin scale however, regardless of a 13% increase in surface mine 
reclamation in the Georges Creek basin since 1977 very little difference in stormflow 
response characteristics was observed. Georges Creek was found to be substantially 
flashier than Savage River, but no statistically significant difference in the hydrological 
response characteristics of the two basins could be detected. In addition, no significant 
trends could be found over time. The lack of response was different than hypothesized 
and may be the result of a number of factors that hinder scaling the runoff responses 
observed at the small watershed scale to the larger river basin scale. Finally, I believe 
there is a need to conduct watershed studies of runoff generation on a variety of 
reclaimed mines that are representative of the diversity of reclamation practices that 
have actually been employed in western Maryland and at other locations where flooding 
71 
may be a major concern. Based on the findings in this study, it is critical that future 
research, land management, and watershed planning decisions consider the relationship 
between surface mining and hydrological response in the Georges Creek watershed as 
well as other similar watersheds. 
72 
Appendix I. Coordinates of watersheds and gages 
Latitude Longitude 
Watersheds 
MAT ! 39? 35' 39" N 78? 53' 29" W 27 ha 
NEFI 39? 35' 47" N 78? 54' 29" w 3 ha 
EBNR 39? 36' OI" N 78? 54' 06" W 104 ha 
Georges Creek (as gaged at Franklin) 39? 35' 00" N 79? 00' 00" w 187.6 km2 
Savage Ri ver (as gaged near Barton) 39? 35' 00" N 79? 05' 00" w 127.2 km2 
Gage Locations Used 
MATI 39? 35' 32" N 78? 53' 49" w 
NEFI 39? 35' 54" N 78? 54' 13" W 
EBNR 39? 35' 52" N 78? 54' 38" w 
Geo rges Creek at Franklin 39? 29' 38" N 79? 02' 42" w 
Savage River Near Barton 39? 34' 05" N 79? 06' 10" w 
73 
Appendix II. LULC classes, codes, and id key 
Class l : Low intensity developed LuCode: 2 
Class 2: High intensity developed LuCode: 3 
Class 3: Agriculture (hay/pasture/crop) LuCode: 5 
Class 4: Forest (e vergreen/deciduous/mixed) LuCode: 10 
Class 5: Active surface mines LuCode: 13 
Class 6: Abandoned surface mines LuCode: 14 
Class 7: Reclaimed surface mines LuCode: 15 
Class 8: Spoil (tailings) LuCode: 16 
Class 9: Landfill LuCode: 17 
LULC classes were divided into their respective classes with the following 
descriptions (adapted from the Multi-Resolution Land Characteristics 
Consortium-MRLC): 
Class l: Low intensity developed (approximately 50 - 80% constructed material; 
approximately 20-50% vegetation cover; generally a high percentage of 
residential development). 
Class 2: High intensity developed ( 80 - 100 % constructed material; less than 20 % 
vegetation; generally commercial development or dense residential). 
Class 3: Agriculture (areas that are primarily hayed or grazed. Includes pastures, row 
crops, and hay). 
Class 4: Forest (greater than 50% forest cover; includes a wide grouping of forest types: 
deciduous, conifers, both conifers and deciduous, forested wetlands, > 50 % 
revegetation on reclaimed mines). 
Class 5: Active surface mines (areas currently being surface mined; visible coal seam, 
haul roads, and equipment). 
Class 6: Abandoned surface mines (open pits; often scattered shrubs in pits; no mining 
equipment visible). 
Class 7: Reclaimed surface mines (significant signs of recent reclamation; diversion 
ditches present; impressions in soil from reclamation equipment). 
Class 8: Spoil (primarily from deep mines; tailings, gob piles, deep mine openings). 
Class 9: Landfill (active landfilling; equipment and solid waste visible). 
74 
Appendix III. PC-IHACRES Model Results 
====-===================----===========-------------- - -
MATl 2000 
IHACRES for WINDOWS, Version 1.02 
FILE : D:\PROGRAMS\IHACRES\MATTHEW\M2000.SUM 
Date created :02/26/02 
Time created :15:08:49 
Record start date :01/06/2000 
Record start time :01:00 
Record end date :27/09/2000 
Record end time :12:00 
Record time interval 1 hourly 
Number of records 2844 
results. 
CONTAINS : Summary of model 
Re f e rence Temperature = 20.00 8700 (2844), subints= 
l,Time Delay= 
Vers ion 1.02, Subperiod= 1,Range= 
5857 to 
BO Const 
1 Bias xl ul 
%ARPE T.C. Al 
f TauW %Run D -.529 .005 
1.00 90 12.22 9.8 .o .04 1.57 ***** .005 
1.00 90 12.22 .682 . 13 -.442 ***** 
.693 .19 1. 8 
.o .05 1.23 
1.00 90 12.22 . 0 .06 1.17 -.426 ***** 
.005 
.5 .005 
1.00 90 12.22 .694 .20 -.424 ***** .4 .o .06 1.17 .005 
1.00 90 12.22 .694 . 20 . 0 .06 1.18 -.429 ***** 
.694 .19 1. 3 1.00 90 12.22 
- --- - ---=======================--------------------------== 
MATl 2001 
IHACRES for WINDOWS, Version 1.02 
FILE : D:\PROGRAMS\IHACRES\MATTHEW\M2001.SUM 
Date created :02/26/02 
Time created :15:23:27 
Record start date :01/06/2001 
Record start time :01:00 
Record end date :25/09/2001 
Record end time :01:00 
Record time interval 1 hourly 
Number of records 2785 
CONTAINS : Summary of model results. 
VReerfseiroenn c1e.0 2T,e mspuebrpaetruiorde==  12,R0a.0ng0e =14617 to17401(2785), subints= 1,Time Delay= 
Al BO Const 1 
%Run Bias xl 
ul %ARPE T.C. 
f TauW D 
1.00 1 Tw(tk) is 
less than 1 
1.00 11 14.95 
.007 
.38 83.3 -.8 .02 
2.09 -.619 ***** 
1.00 11 14.95 .614 
1.00 21 14.95 72.9 -.6 .02 1. 79 -.573 ***** 
.006 
1.00 21 14 . 95 .692 .35 
1.00 31 14.95 70.1 -.6 .02 1. 70 -.556 ***** 
.005 
1.00 31 14.95 .720 .33 
1.00 41 14.95 .005 
14.95 .729 .31 69.5 
-.6 .02 1. 67 -.550 ***** 
1.00 41 
1.00 51 14.95 .005 
14.95 .730 .29 69.9 -.6 
.02 1. 67 -.549 ***** 
1.00 51 
1.00 61 14.95 ***** 
.726 .27 70.7 -.6 
.02 1. 68 - .551 
1.00 61 14.95 
.004 
1.00 71 14.95 .02 1. 69 -.554 ***** -.6 
1.00 71 14.95 .720 .24 
71. 7 
.004 
1.00 81 14.95 .02 1. 71 -.558 ***** 
1.00 81 14.95 . 713 .22 
72. 9 -.6 
.004 
75 
1.00 91 14.95 .02 1.74 -.563 ***** 
1.00 14.95 .704 .20 
74.1 -.6 
91 
.004 
==-----==-==============--~=-=---------==================== 
NEFl 2000 
IHACRES for WINDOWS Version 1.02 
FILE : D:\PROGRAMS\IHACRES\TRBNEF1\N200SUM.SUM 
D~te created :02/26/02 
Time created :15:08:49 
Record start date :01/06/2000 
Record start time :01:00 
Record end date :27/09/2000 
Record end time :12:00 
Record time interval 1 hourly 
Number of records 2844 
CONTAINS : Summary of model results. 
Reference Temperature = 20 . 00 8689 (2833), Subi nts= l,Tirne Delay= to 
Version 1 . 02, Subperiod= 1,Range= 5857 
1 T.C. Al BO Const 
f Tauw %Run D Bias xl 
ul %ARPE 
1.00 61 108.94 . 0 .04 1. 58 -.531 3.400 
.050 
1.00 61 108.94 .664 .14 10.6 
------==----=================--=-=======-=-----------------
NEFl 2001 
IHACRES for WINDOWS, Version 1.02 
Date created :02/26/02 
Time created :15:19:35 
Record start date :01/06/2001 
Record start time :01:00 
Record end date :27/09/2001 
Record end time :01:00 
Record time interval 1 hourly 
Number of records : 2833 
FILE D:\PROGRAMS\IHACRES\TRBNEF1\N2001.SUM 
CONTAINS : Summary of model results. 
Reference Temperature = 20.00 to17449(2833), Subints= l,Tirne 
Version 1. 02, Subperiod= 1,Range=14617 
Delay= 1 1/c Tq Ts Vs 
f TauW %Run D Bias xl ul %ARPE 
1.00 60 9.97 .1 . 0 .06 341. 5 .81 27.28 .458 1.00 60 9.97 .755 .01 
1.00 61 9.97 343.4 .81 27.26 .458 
1.00 61 9.97 .755 .01 .1 .0 .06 
76 
Appendix IV. LULC changes over time by category (area in hectares) 
1962 
LID HID AG FOR ACTMIN ABANMIN RECMIN SPOIL LFIL SUM 
LID 64.0 0.8 18.0 24.8 0.0 0.3 0.0 0.0 0.0 108 
HID 25.2 261.1 26.5 30.1 0.0 0.0 0.0 0.2 0.0 343 
AG 57.6 47.7 
QC) 2998.1 2549.2 25.3 279.1 0.0 32.4 0.0 5990 
~ 
~....  FOR 12.8 1.4 556.9 11214.4 77.3 340.5 0.0 62.2 0.0 12265 
ACTMlN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 
ABANMIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 
SPOIL 0.1 1.4 3.9 42.1 0.3 0.2 0.0 22.2 0.0 70 
SUM 160 312 3603 13861 103 620 0 117 0 18776 
1982 
LID HID AG FOR ACTMIN ABANMlN RECMIN SPOIL LFIL SUM 
LID 87 .0 18.3 12.4 39.9 0.0 0.0 0.0 0.0 0.0 158 
HID 11.7 260.8 16.7 22.5 0.0 0.0 0.0 0.0 0.0 312 
-.J C: 
-.J 
N AG 131.4 34.5 1583.8 1283.8 165.8 12.5 387.4 5.4 0.0 3605 zc  
\0 
~....  FOR 62.7 17.5 639.6 11619.3 371.8 78.0 1053.8 18.1 0.0 13861 :) ~ 
ACTMlN 0.0 0.0 1.7 48.4 11.6 7.7 33.4 0.1 0.0 103 C,  
ABANMIN 1.1 0.0 34.9 264.1 98.1 60.9 161.6 0.0 0.0 621 
SPOIL 2.4 0.1 3.9 77.1 3.8 4.8 12.3 12.6 0.0 117 ;; 
SUM 296 331 2293 13355 651 164 1649 36 0 18775 '"  
1997 
LID HID AG FOR ACTMIN ABANMIN RECMlN SPOIL LFILL SUM 
LID 174.4 34.8 26.9 58.2 0.2 0.0 1.7 0.0 0 .0 296 
HID 47.7 259.6 4.0 19.6 0.0 0.1 0.1 0.0 0.0 331 
AG 70.3 45.5 1222.3 736.6 39.6 5.0 173.3 0.4 0.0 2293 
N 
00 FOR 170.5 63.0 397.3 11563.4 185.6 68.0 862.7 12.5 31.2 13354 
~....  ACTMlN 0.7 0.0 33.8 140.4 75.7 1.8 398.5 0.4 0.0 651 
ABANMIN 0.3 0.0 3.5 92.1 0.0 10.6 57.1 0.3 0.0 164 
RECMIN 0.3 0.1 99.0 532.4 97.9 18.0 898.6 2.1 0.0 1648 
SPOIL 0.1 0.1 5.0 18.5 0.0 1.3 7.5 3.7 0.0 36 
SUM 464 403 1792 13161 399 105 2400 19 31 18774 
REFERENCES 
Allan, J.D., Erickson, D.L., and J. Fay. 1997. The influences of watershed landuse on 
stream integrity across multiple spatial scales. Freshwater Biol. 37:149-161. 
Andreassian, V., Perrin, c., Michel, C., Usart-Sanchez, I., and J. Lavabre. 2001. Impact 
of imperfect rainfall knowledge on the efficiency and the parameters of 
watershed models. J. Hydrol. 250 (l-4): 206-223. 
Barnhise1, RI. and J.M. Hower. 1997. Coal surface mine reclamation in the eastern 
United States: the revegetation of disturbed lands to hay land/pasture or 
cropland. Advances in Agron. 61:233-275. 
Bell , J.C., Cunningham, R.L., and C.T. Anthony. 1994. Morphological characteristics of 
reconstructed prime farmland soils in western Pennsylvania. J. Environ. Qual. 
23:515-520. 
Beven, K., and P. Germann. 1982. Macropores and water flow in soils. Water 
Resources Research 18: 1311-1325. 
Banta, J.V. , Amerman, C.R., Dick, W.A., Hall, G.F., Harlukowicz, T.J., Razem, A.C., 
and N.E. Smeck. 1992. Impact of surface coal mining on three Ohio watersheds-
physical conditions and ground-water hydrology. Water Resour. Bull. 28:577-
596. 
Banta, J.V. , Amerman, C.R., Harlukowicz, and W.A. Dick. 1997. Impact of coal 
surface mining on three Ohio watersheds-surface water hydrology. J. Amer. 
Water Resour. Assoc. 33:907-917. 
Burges, S.J., M.S. Wigmosta, and J.M. Meena. 1998. Hydrological effects of landuse 
change in a zero-order watershed. J. Hydrol. Eng. 3:86-97. 
Burt, T.P., and W.T. Swank. 1992. Flow frequency responses to hardwood-to-grass 
conversion and subsequent succession. Hydrological Processes. 6: 179-188. 
Burton, T.A. 1997. Effects of basin-scale timber harvest on water yield and peak 
streamflow. J. Amer Water Resour. Assoc. 33:1187-1196. 
Bussler, B.H., Byrnes, W.R. , and Pope, P.E., and W.R. Chaney. 1984. Properties of 
minesoil reclaimed for forest landuse. Soil Sci. Soc. Am. J. 48:178-184. 
Chapman, T.G. I 996a. Common unitgraphs for sets of mnoff events. I. Unitgraph 
identification from streamflow data. Hydrol. Processes. 10:773-782. 
78 
II I I ? 11111? 11?11a uuu au,???? ??? ,. .,, ??-? ?? ?-? 
Chapman, T.G. 1996b. Common unitgraphs for sets of runoff events.2.Comparisons and 
inferences for rainfall loss models. Hydro!. Processes. 10:783-792. 
Chong, S.K., and P.T. Cowsert. 1997. Infiltration in reclaimed mined land ameliorated 
with deep tillage treatments. Soil and Tillage Res. 44:255-264. 
Chiew, F.H.S., Stewardson, M.J., and T.A. Mcmahon. 1993. Comparison of 6 rainfall-
runoff modeling approaches. J. Hydro!. 147 (1-4): 1-36. 
Chow, V. T., Maidment, D.R. and Mays, L. W. Applied Hydrology: McGraw-Hill 
1988 
Dietrich, C.R. 1996. Linear rainfall-runoff system identification with ramifications for 
modeling watershed processes. Hydro! Proc. 10:893-902. 
Dingman, S.L., 1994. Physical Hydrology. Prentice-Hall, Upper Saddle River. 
Dunker, R.E., Hooks, C.L., Vance, S.L., and R.G. Darmody. 1995. Deep tillage effects 
on compacted surface-mined land. Soil Sci. Soc. Am. J. 59:192-199. 
Dunne, T., and L.B. Leopold. 1978. Water in environmental planning. San Francisco: 
W.H. Freeman. 818 pp. 
Fors tall, R.L. 1995. Population of Counties by Decennial Census: 1900 to 1990. 
Online report: http://www.census.gov/population/cencounts/md 190090. txt 
Freeze, R.A. 1974. Streamflow generation. Reviews of geophysics and space physics. 
124:627-647. 
French, M.N., and W.F. Krajewski. 1994. A model for real-time quantititative rainfall 
forecasting using remote sensing. Water Resour. Res. 30: 1075-1083. 
Gebert, W.A., and W.R. Krug. 1996. Streamflow trends in Wisconsin's driftless area. 
Water Resour. Bull. 32:733-744. 
Gifford, G.F., Humphries, W., and R.A. Jaynes. 1984. A preliminary quantification of 
the impacts of aspen to conifer succession on water yield - II. Modeling results. 
Water Resour. Bull. 20: 181-186. 
Guebert, M.D. and T.W. Gardner. 2001. Macropore flow on a reclaimed surface mine: 
infiltration and hillslope hydrology. Geomorphology. 39:151-169. 
79 
Hansen, D .P. , Ye, w., Jakeman, A.J., Cooke, R., and P. Shanna. 1996. Analysis of the 
effect of rainfall and streamflow data quality and watershed dynamics on 
streamflow prediction using the rainfall-runoff model IHACRES. Environ. 
Softw. 11 (1-3): 193-202. 
Hewlett, J.D., and A.R. Hibbert. 1967. Factors affecting the response of small 
watersheds to precipitation in humid areas, in Proceedings of the International 
Symposium on Forest Hydrology, pp. 275-290, Pennsylvania State University, 
University Park. 
Hornbeck, J.W. , Pierce, R.S., and C.A. Federer. 1970. Streamflow changes after forest 
clearing in New England. Water Resour. Res. 6: 1124-1132. 
Hypio, S.J. 2000. Streamwater nitrogen export from the Savage River watershed, 
westen Maryland. Masters Thesis. University of Maryland, College Park, MD. 
Jones, J.A., and G.E. Grant. 1996. Peak flow responses to clear-cutting and roads in 
small and large basins, western Cascades, Oregon. Water Resour. Res. 32:959-
974. 
Kochenderfer, J.N., Edwards, P.J. , and F. Wood. 1997. Hydrologic impacts of logging 
an Appalachian watershed using West Virginia's best management practices. 
Northern J. Appl. For. 14:207-218. 
Kokkonen, T., Koivusalo, H., and T. Karvonen. 2001. A semi-distributed approach to 
rainfall-runoff modeling - a case study in a snow affected watershed. Environ. 
Modell. Softw. 16 (5): 481-493. 
Kuhnle, R.A., Bingner, R.L., Foster, G.R., and E.H. Grissinger. 1996. Effect of landuse 
changes on sediment transport in Goodwin Creek. Water Resour. Res. 32:3189-
3196. 
Langbein, W.B. 1940. Channel storage and unit hydrograph studies; BOS, Amer 
Geophys. Union Trans. 21:620-627. 
Letcher, R.A., Schreider, S.Y., Jakeman, A.J., Neal, B.P., and R.J. Nathan. 2001. 
Environmetrics. 12 (7): 613-630. 
Littlewood, LG., Down, K., Parker, J.R., and D.A. Post. 1997. IHACRES vl.0 User 
Guide. Center for Ecology and Hydrology, Wallingford, UK & Integrated 
Watershed Assessment and Management Centre, Australian National 
University, Canberra, 97 pp. 
80 
?? 1 , ? ..,.,.' -;f,~IY' ....Q f:. MD COLLEGS: PAAlf 
Maryland Department of Natural Resources. 2001. Georges Creek Watershed 
Characterization. ( draft). 73pp. 
Mcsweeney, K. and I.J. Jansen. 1984. Soil structure and associated rooting behavior in 
minesoils. Soil Sci. Soc. Am. J. 48: 607-612 
Mwendera, E.J., and M.A Mohamed-Saleem. 1997. Infiltration rates, surface runoff, 
and soil loss as influenced by grazing pressure in Ethiopian highlands. Soil Use 
Manag. 13:29-35. 
Rantz, ~-E. 197 I . Suggested criteria for hydrologic design of stor?1-drainage facilitie~ 
m the San Francisco Bay Region, California; U.S. Geological Survey Open File 
Report, Menlo Park, CA. 
Ritter, J.B., and T.W. Gardner. 1993. Hydrologic evolution of drainage basins disturbed 
by surface mining, central Pennsylvania. Geol. Soc. Amer. Bull. 105:101-115. 
Rose, S., and N.E. Peters. In press. Effects of urbanization on streamflow in the Atlanta 
area (George USA): a comparative hydrological approach. Hydrol. Proc. 
Schreider, S. Y., Jakeman, A.J., and A.B. Pittock. 1996. Modelling rainfall-runoff from 
large watershed to basin scale: The Goulburn Valley, Victoria. Hydrol. Proc. 
10 (6): 863 -876. 
Schreider, S.Y., Young, P.C., and A.J. Jakeman. 2001. An application of the Kalman 
filtering technique for streamflow forecasting in the Upper Murray Basin. Math. 
Comput. Model. 33 (6-7): 733-743. 
Schueler, T. 1994. The importance of imperviousness. Watershed Prot. Tech. 1 (3): 
100-111 
Sefton, _C.E.M., and D.B. Boorman. 1997. A regional investigation of climate change 
impacts on UK streamflows. J. Hydro]. 195:26-44. 
Sherman, L.K. I 932. Stream-flow from rainfall by the unit-graph method, Eng. News-
Rec. 108:501-505. 
Smith, 1.A., and W.F. Krajewski. 1991. Estimation of the mean field bias of radar 
rainfall estimates, J. Applied Meteor. 30:397-412. 
Smith 1.A., Seo, D.J., Baeck, M.L. , and M.D. Hudlow. 1996. An intercomparison study 
of NEXRAD precipitation estimates. Water Resour. Res. 32 (7): 2035-2045. 
81 
Snyder, F.F. 1938. Synthetic unit hydrographs; EOS, Amer Geophys Union Trans. 
19:447-454. 
Swanson, R.H. 1984. Managing lodgepole pine: the species nad its management. 
Symposium Proceedings, May 8-10, 1984, Spokane, Washington, pp. 305-313. 
Swift, L.W., Swank, W.T., Mankin, J.B., Luxmoore, R.J., and R.A. Goldstein. 1975. 
Simulation of evapotransporation and drainage from mature and clear-cut 
deciduous forests and young pine plantation. Water Resour. Res. 11 :667-673. 
Thomas, R.B., and W.F. Megahan. 1998. Peak flow responses to clear-cutting and roads 
in small and large basins, western Cascades, Oregon: a second opinion. Water 
Resour. Res. 34:3393-3403. 
Troendle, C.A., and R.M. King. 1987. The effect of partial and clearcutting on 
streamflow at Deadhorse Creek, Colorado. J. Hydrol. 90:145-157. 
United States Congress. 1977. Surface mining control and reclamation act of 1977. 95th 
Congress, Public Law 95-87. 
United States Department of Agriculture. 2001. Fact Finders for Agriculture. 
http://www.usda.gov/nass. 
United States Environmental Protection Agency. Multi-Resolution Land 
Characteristics-MRLC Homepage. Revised: Friday, March 30, 2001 13:07:50. 
http://www.epa.gov/mrlc/. 
United States Soil Conservation Service 1972. Hydrology, Section 4, National 
Engineering handbook, Washington, DC. 
United States Soil Conservation Service. 1974a. Soil survey of Allegany County, 
Maryland. 83pp. 
United States Soil Conservation Service. 1974b. Soil survey of Garrett County, 
Maryland. 83pp. 
United States Geological Survey. 1983. Hydrology of area 6, eastern coal province, 
Maryland, West Virginia, and Pennsylvania. Water Resources Investigations: 
Open-File Report 83-33. 
Whipkey, R.Z. 1965. Subsurface stormflow from forested slopes. Int. Assoc. Sci. 
Hydrol. Bull. 10 (2):74-85. 
82 
Ye, W., Bates, B.C., Viney, N.R., Sivapalan, M., and Jakeman, A.J. 1997. Performance 
of conceptual rainfall-runoff models in low-yielding ephemeral watersheds. 
Water Resour. Res. 33 (1): 153-166. 
Ye, W., Jakeman, A.J. , and P.C. Young. 1998. Identification of improved rainfall-
runoff models for an ephemeral low-yielding Australian watershed. Environ. 
Modell. Softw. 13 (l): 59-74. 
83