1. Hydrologic Modeling of Development Effects
on an Untouched Coastal Watershed
Julianna Corbin, Hunter Morgan, Evan Patrohay, Ty
Williams
Clemson University, Clemson, SC
December 3, 2020
4. Background
● Above image shows the urban centers of the US
○ Southeastern Trend:
■ Higher densities
■ More urban centers
● Below image shows state of forests in the US
○ Almost everywhere in the southeast
contains forests that are regularly cut
○ With population growth trends, some of
this forest is never replanted
● EPA projects 2,029 mi² of forest in SC to be
cleared for urban use by 2050
5. Background
● Charleston Metro Area (CMA) population = 802,122 people
● The area is growing at:
○ ×3 the population of the USA
○ ×2 the population of South Carolina
● 30 new residents move in per day
○ Charleston (+13%)
○ North Charleston (+16%)
○ Mount Pleasant (+32%)
Percent growth from 2010-2019
6. Background
● The city is bounded by the Atlantic
Ocean to the east
● Constrained by the harbor
○ Ashley, Cooper, Wando Rivers
● Much of the land area is marshy and
not fitting for development
Means the city has to expand to the north & west!
Towards the Francis Marion Forest.
FMNF
7. Background
● Mission of the Santee Experimental Forest:
○ Understand coastal plain forest hydrology
○ Silviculture and forest management
● Now, that Charleston, SC is encroaching into
the Francis Marion National Forest:
○ Wish to understand how will
imperviousness impacts forest hydrology
and ecosystem functions
8. Background
● Watershed of interest: WS80
○ 1st order watershed
○ ~ 400 acres
○ Drains into a tributary of
the Cooper River
○ Last logged in 1937
○ Most untouched
watershed in SEF since
that time
9. Background - Urbanization and Stormwater Flooding
● Natural land
○ Vegetation and soil intercept and soak up rainfall and slow runoff
○ Momentum of runoff is reduced and roots anchor the soil, reducing erosion
● Developed land
○ Rainfall hits impervious surfaces intensifies runoff
■ Rooftops, roads, parking lots, pavement
○ Natural water cycle is changed
○ Streets collect stormwater and channel it into waterways
○ Pollutants from urban surfaces are collected and transported into lakes, streams, and the ocean
■ Decreased water quality
○ Storm drains directly transport runoff and pollutants to bodies of water
○ Greater frequency and severity of flooding, channel erosion, and destruction of aquatic
ecosystems
11. Rationale
● Running urban hydrologic analyses is important for future land development
○ How does impervious surfaces affect hydrology here?
● WS80’s preservation makes it an excellent “control” for these hydrologic analyses
○ Assumed a typical Atlantic Coastal Plain Forested Watershed
○ Running peak flow / runoff modeling will give best approximation of a natural
watershed in this region
● Comparing results with current literature will expand datasets and increase
confidence in the mission of Santee Experimental Forest
12. Objectives
The objectives of this project are to:
1. Assess the pre-development hydrologic conditions of Watershed 80, located within the
USDA Forest Service’s Santee Experimental Forest northeast of Charleston, SC, using three
different stormwater hydrology models
(The Rational Method, NRCS TR-55 model, and USGS Regional Regression Equations).
1. Simulate runoff volume and peak flow rate with varying levels of urbanization, defined by 0%,
5%, 10%, and 15% imperviousness organized by soil classification on Watershed 80, using
the same models and compare the values obtained to real-life data.
2. Design a culvert to transport the amount of peak discharge at the outlet of watershed 80
calculated from the model with the highest peak-flow at a 100-yr return period.
13. Approaches
Task 01: To obtain aerial, visual, soil, and elevation maps of Watershed 80
Task 02: To procure rainfall intensity data and stream flow data from the SEF website
Task 03: To delineate sub-basins and drainage area on GIS and Civil 3D
Task 04: To simulate peak flow and runoff values of the watershed using WinTR-55
Task 05: To perform the same calculations using the USGS Regression Equation
Task 06: To utilize the Rational Method for further calculations of peak flow and runoff values
Task 07: To repeat tasks 4-6 with three increasing iterations of imperviousness
Task 08: To design a culvert for a 100-year storm
14. Deliverables
1. A mapped and delineated GIS file
2. Pre-development WinTR-55 hydrology values
and graphs
3. Pre-development USGS Equation values and
graphs
4. Pre-development Rational Method values and
graphs
5. Three sets of post-development graphs/values from
respective models
6. HY-8 Culvert design, including 3D model
20. The Southeast United States
● Approximately 55% of the Southeast US is forested
○ Timber production has increased dramatically
■ Doubled in the past few decades
■ Est. 4.9 million hectares of former forest
expected to be lost to development by
2020
○ Shift in maintenance has created a significant
nonpoint source pollution rise of sediments
● Some of this forest is not returning due to population
settlements
21. The Southern Coastal Plain
● Southern Coastal Plain
○ Barrier islands, coastal lagoons, marshes,
flat plains, swampy lowlands
○ Characterized by wet soils, low elevation,
little relief
● Location of WS80 and Charleston
23. Urban Stormwater Flow
● Urbanization presents significant danger to the
integrity of surrounding streams
○ Riparian buffers can be wiped out from
intense peak discharge values
○ Floodplain damage occurs during large
storm events
● Damaged buffers lead to increased exposure,
UV radiation and higher water temperatures
○ Threatens natural habitats
24. Urban Stormwater Flow
● Urban runoff contributes to polluted streams
and decreased water quality
○ Harmful to natural habitats
○ Trash and debris flows from streets into
the stream system
● Natural stream systems meander and contain
suitable animal habitats
25. Stormwater Flooding in Charleston
● Charleston flooded 1 out of 5 times in 2019
● Harbor flood gauges exceeded 7.0 feet 89 times
○ Flooding occurs at 7.0 ft
27. WS8O
● Characteristics:
○ Is 23% wetland
○ Slopes <1%
○ Soils: Primarily sandy loams, clay subsoils -- poorly drained
○ Vegetation: Loblolly pine, longleaf pine, cypress, sweet gum
● Weather characteristics:
○ Mean annual rainfall → 54 inches
○ Mean annual temperature → 64°F
● Monitored for:
○ Meteorological data, flow gauging, groundwater, water quality
○ Contains a v-notch weir at outflow
28. Literature Review - Climate Change Risks
● Risk Management must take into account:
○ Biophysical factors influencing future stormflow
○ Socioeconomic factors influencing community adaptation to alterations in stormflow
● Projections:
○ 4-12°F increase in southeast US temperatures by 2100
○ More extreme hydrological events → frequent +10 yr storms
○ Population growth + urban land cover increase 101-192% by 2060
30. Materials
● Santee Experimental Forest published datasets
○ GIS
○ Weather stations
● Mapping using ArcGIS software
● Mapping using AutoCAD Civil 3D
● Peak flow modeling using WinTR-55
● Low-impact design testing using L-THIA
● Data processing using Microsoft Excel
● Peak flow comparisons using NOAA published datasets
31. Soil Data for Watershed WS80
● NRCS resources used in collecting necessary soil
data
● Intended use is for proposal of development on
soils of different hydrologic soil groups for
comparison
32. Watershed Delineation
● Published data sets were utilized to input
layers into ArcGIS
● Data was also input into AutoCAD Civil
3D for delineation
● Drainage areas were collected based on
delineation
East Sub-Basin West Sub-Basin
Area [acres] Area [acres]
198 191
34. Modeling Method #1 - Rational Method
● Developed by Thomas Mulvaney in 1851 and introduced in the US by Emil Kuichling in 1889
● Empirical formula used to estimate peak runoff discharge (Q) in small watersheds
● Function of drainage basin size, characteristics, and precipitation
Thomas Mulvaney Emil Kuichling
35. The Rational Method Equation:
Q= CiA
where
Q = peak flow rate [ft3/s]
C = runoff coefficient [dimensionless]
i = average rainfall intensity-duration [in/hr]
A = drainage area [acres]
36. Runoff Coefficient (C)
● Related to the abstractive and
diffusive elements found throughout
drainage basins
● Attributed to basin size, shape,
topography, soil, geology, and land
use
● Range from 0 to 1
● Table of example C values from a
manual published by the SC
Department of Transportation
37. Rainfall intensity (i)
● Calculated by measuring
the amount of rainfall per
unit time in a specific
location
● The unit of time selected
for the Rational Method is
the same as the time of
concentration
● Rainfall intensity curve of
WS80 from NOAA
estimated i-values
38. Area (A)
● The purpose of the Rational Method is to estimate peak discharge
from smaller watersheds
○ Recommended that it be applied to watersheds with drainage areas up
200 acres
○ Valid up to 300 acres for low-lying tidewater areas
● WS80 exceeds the size limitations of the Rational Method
○ Area of about 389 acres
39. Assumptions and Limitations of the Rational Method
When applying the Rational Method it is assumed:
1. That precipitation is uniform over the entire basin
2. The precipitation does not vary with time or space
3. The duration of the storm is equal to the time of concentration
4. The designed storm of a specified frequency produces the design flood of the same frequency
5. That the basin area increases roughly in proportion to increases in length
6. The time of concentration is relatively short and independent of storm intensity
7. That the runoff coefficient does not vary with storm intensity or antecedent soil moisture
8. The runoff is dominated by overland flow
9. The basin storage effects are negligible
40. Modeling Method #2 - USGS Regression Equations
● In 2014, the USGS worked in collaboration with the SCDOT to gather data from
488 stream gauges across the east coast
● Done for 3 hydrologic regions (HRs):
○ HR1 - Piedmont
○ HR3 - Sand Hills
○ HR4 - Coastal Plain ← of interest for us
● Created specific regression equations based on:
○ Return period
○ Stream gauge peak flows
○ NOAA weather data
41. Modeling Method #2 - USGS Regression Equations
Peak Flow = 𝑓(Area, % Annual Exceedance (AEP), %
Imperviousness)
The regression equations used are governed by the following parameters:
42. Modeling Method #2 - USGS Regression Equations
DRNAREA = drainage area [mi²]
IMPNLCD06 = % impervious area [-]
DEVNLCD06 = % developed land
I24H50Y = 24-hr, 50 yr maximum
precipitation [in]
Special note: Given the acceleration of climate change, the I24H50Y is
variable and will not always accurately represent the climate of the region.
43. Modeling Method #2 - USGS Regression Equations
Variance of Prediction Standard Error of Prediction
where
γ2
xi
U
x‘i
is the model error variance
variables for site i, augmented by 1 as the
first element
is the covariance matric for the regression
is the transpose of xi
where
Sp,ave
AVP
is the average standard error of
prediction, in percent
the average variance of prediction
44. Modeling Method #2 - USGS Regression Equations
Average Variance of Prediction
and
Standard Error of Prediction
*based on hydrologic region and return period.
● Used to create an estimated range of
peak flow values
● Based on inherent uncertainty of the
regression equations
45. Limitations of the USGS Regression Equations:
1. Generally for use in areas with <10% impervious area (rural areas)
a. Study used limit of 15% imperviousness
2. Drainage area should be > 0.1 mi²
a. This watershed is 0.6 mi² in area
3. Not appropriate where significant man-made structures alter flow
a. V-notch weir at outlet, assumed non-significant
4. Do not apply where tidal effects are found
Modeling Method #2 - USGS Regression Equations
46. Modeling Method #3 - WinTR-55
● Single event small watershed
hydrology analysis program
● First launched in 1975
○ Several updates since
● Managed by the Natural Resources
Conservation Service (NRCS)
47. Modeling Method #3 - WinTR-55
● User inputs
○ Sub-basin drainage areas
○ Sub-basin Curve Numbers
○ Time of Concentration
○ Rainfall
● Computes outflow of watershed
● Can incorporate outflow through
culverts
48. WinTR-55 Limitations
● Sub-basin areas should be at least one
acre
● No more than 25 square miles total
● Sheet flow must be less than 100 feet
● No more than ten sub-basins
49. Study of Low-Impact Design (LID)
● Long Term Hydrologic Impact Analysis (L-THIA)
○ Estimates changes in runoff from
past/proposed development
○ Based on climate data, soil type, & land use
● Enables study on how LID will reduce runoff based
on the % imperviousness of this study
58. Post Development Results
The following equation was used to calculate the weighted average C-value for the simulated
development areas:
CW = (C1A1 + C2A2 + … + CnAn)/(A1 + A2 + … An)
where
CW = weighted runoff coefficient [-]
C = runoff coefficient [-]
A = drainage area [acres]
59. Post Development Results
● Rational Method Equation: Q = CiA
Return period
[yr]
i [in/hr] A0
[acres]
A5% [acres] A10%
[acres]
A15% [acres] Cavg Cres Weighted
C5%
Weighted
C10%
Weighted
C15%
2 0.882
389.943 19.497 38.994 58.491 0.096 0.400 0.111 0.126 0.141
5 1.209
10 1.492
25 1.949
50 2.390
100 2.906
● INPUTS: C, i, and A
63. Results - USGS Regression Equations
Figure 1: Pre-development peak flow Figure 2: Increase in flow by imperviousness
64. Results - USGS Regression Equations
Trends:
● Peak flow and imperviousness have a highly linear relationship
● Greatest difference in flow occurs for a 25-yr storm (Δ109 cfs from 0% - 15%)
○ Smaller differences in flow at either extreme
65. Results - USGS Regression Equations
● Increase in peak flow is more
dramatic for smaller return-period
storms
○ Common storms will on average
cause more damage
● Flows from large return-period storms
are not affected as much
○ But their flow rates are more
unpredictable
66. USGS Regression Equations using NOAA
values
● NOAA 24-hr 50-yr intensity = 8.85 in
○ WS80 value was higher because
Hurricanes Joaquin & Matthews
and tropical storms factored in
● Results shown:
○ Peak flows reduced in every case
○ Reduced by close to 50% at and
above 50-yr storms
67. WinTR-55 Inputs
Land Use Information
● Used a Curve Number of
67 for pre development
and undeveloped
conditions
● Used Curve Number of
98 for simulated
developed area
Rainfall Data
● Used depths for 2, 5, 10,
25, 50, 100, and 200
year storms derived by
Amatya et al.
Time of Concentration
● Used a time of
concentration of 3 hours
derived by Amatya et al.
● Calculated a time of
concentration of 1.66
hours for simulated
development
73. Comparison of Models
Figure 1: Comparison of Models at Pre-development Figure 2: Comparison of Models at 10% Imperviousness
74. Results - Low Impact Design
Using:
● Bioswales
● Downspout disconnections
● Green roofs
● Porous pavement
● Natural resource conservation
Models functions by calculating reduction in curve
number by LID and applying to runoff equations.
75. With 50% LID
● Avg. runoff volume reduced from
0.24 → 0.21 acre-ft
● Annual runoff depth reduced from
4.84 → 4.28 in
● Unweighted residential depth
reduced from 8.38 → 4.69 in
76. With 100% LID
● Avg. runoff volume reduced from
0.24 → 0.17 acre-ft (-
20%)
● Annual runoff depth reduced from
4.84 → 3.59 in
(-16%)
● Unweighted residential depth
reduced from 8.38 → 0.09 in (-
98%)
Even on just 15% of the
land, LID practices are
proven to have a very
considerable impact on
runoff.
80. What’s Next?
● Re-evaluate WinTR-55 inputs
● Compare data between the three
modeling methods
● Size and design culvert
Photo from our socially distant September
24 site visit to SEF. Pictured from L to R,
Andy Harrison, Dr. Amatya, Ty, Julianna,
Hunter, and Evan.
82. Acknowledgements
● Dr. Christophe Darnault, Clemson University
● Dr. Rui Xiao, Clemson University
● Dr. Devendra Amatya, USDA Forest Services
● Andy Harrison, USDA Forest Services
● Dr. Andrzej Wałȩga, University of Agriculture,
Krakow, Poland