JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

AMERICAN WATER RESOURCES ASSOCIATION

Representing the Connectivity of Upland Areas to Floodplains and Streams in SWAT+

Katrin Bieger , Jeffrey G. Arnold, Hendrik Rathjens, Michael J. White, David D. Bosch, and Peter M. Allen

Research Impact Statement: We propose and test a simple concept to incorporate hydrologic connectivity in
watershed models, which improves the simulation of processes controlling the response of watersheds to rainfall
events.

ABSTRACT: In recent years, watershed modelers have put increasing emphasis on capturing the interaction of
landscape hydrologic processes instead of focusing on streamflow at the watershed outlet alone. Understanding
the hydrologic connectivity between landscape elements is important to explain the hydrologic response of a
watershed to rainfall events. The Soil and Water Assessment Tool+ (SWAT+) is a new version of SWAT with
improved runoff routing capabilities. Subbasins may be divided into landscape units (LSUs), e.g., upland areas
and floodplains, and flow can be routed between these LSUs. We ran three scenarios representing different
extents of connectivity between uplands, floodplains, and streams. In the first and second scenarios, the ratio of
channelized flow from the upland to the stream and sheet flow from the upland to the floodplain was 70/30 and
30/70, respectively, for all upland/floodplain pairs. In the third scenario, the ratio was calculated for each
upland/floodplain pair based on the upland/floodplain area ratio. Results indicate differences in streamflow were
small, but the relative importance of flow components and upland areas and floodplains as sources of surface
runoff changed. Also, the soil moisture in the floodplains was impacted. The third scenario was found to provide
more realistic results than the other two. A realistic representation of connectivity in watershed models has
important implications for the identification of pollution sources and sinks.

(KEYWORDS: watersheds; runoff; surface water hydrology; simulation; SWAT+; connectivity; uplands; flood-
plains.)

INTRODUCTION Understanding the hydrologic connectivity between

In recent years, watershed modelers have put
increasing emphasis on simulating internal water-
shed processes as realistically as possible (Yen et al.
2014; Arnold et al. 2015). They are no longer satisfied
with achieving a good fit of observed and simulated
discharge at the watershed outlet without under-
standing the interaction of landscape hydrologic pro-
cesses that produce the observed streamflow signal.
landscape elements and how it varies spatially and
temporally is important to explain the hydrologic
response of a watershed to rainfall or snowmelt
events and critical to guiding model development
(Jencso et al. 2009). Bracken and Croke (2007)
defined hydrologic connectivity as the “passage of
water from one part of the landscape to another,”
generating a watershed runoff response. Incorporat-
ing concepts of hydrologic connectivity in watershed
models potentially improves process representation

Paper No. JAWRA-18-0017-P of the Journal of the American Water Resources Association (JAWRA). Received February 2, 2018; accepted
December 21, 2018. © 2019 American Water Resources Association. Discussions are open until six months from issue publication.
Blackland Research & Extension Center (Bieger), Texas A&M AgriLife, Temple, Texas, USA; Grassland, Soil and Water Research Labora-
tory (Arnold, White), USDA-Agricultural Research Service, Temple, Texas, USA; Environmental Modeling (Rathjens), Stone Environmental,
Montpelier, Vermont, USA; Southeast Watershed Research Laboratory (Bosch), USDA-Agricultural Research Service, Tifton, Georgia, USA;
and Department of Geology (Allen), Baylor University, Waco, Texas, USA (Correspondence to Bieger: kbieger@brc.tamus.edu).
Citation: Bieger, K., J.G. Arnold, H. Rathjens, M.J. White, D.D. Bosch, and P.M. Allen. 2019. “Representing the Connectivity of Upland
Areas to Floodplains and Streams in SWAT+.” Journal of the American Water Resources Association 1–13. https://doi.org/10.1111/1752-1688.
12728.

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 1 JAWRA

 BIEGER, ARNOLD, RATHJENS, WHITE, BOSCH,
and the predictive capability of models (Detty and
McGuire 2010).
Aiming at improving our understanding of the pro-
cesses controlling watershed runoff, McGlynn and
McDonnell (2003) and McGlynn et al. (2004) exam-
ined how the most basic units of a watershed, i.e.,
hillslopes and riparian zones, store, receive, and deli-
ver water during and between precipitation events.
Riparian zones differ from hillslopes based on their
proximity to streams, topography, soils, and vegeta-
tion (McGlynn and McDonnell 2003), which impacts
their hydrology and connectedness to streams. Ripar-
ian zones are often characterized by higher soil mois-
ture and prolonged periods of saturation (McGlynn
and McDonnell 2003). They have the potential to buf-
fer the delivery of water from hillslopes to streams
and thus exert a large influence on the hydrologic
response of a watershed to precipitation events
(Jencso et al. 2010; McGuire and McDonnell 2010).
Hydrologic connectivity is influenced by many fac-
tors including storm characteristics, antecedent wet-
ness conditions, topography, soils, and vegetation
(Bracken and Croke 2007). However, our knowledge
of the dominant controls on hydrologic connectivity of
different landscape elements and their interactions
over space and time is still incomplete. Several
researchers have analyzed the connectivity of hill-
slopes to riparian zones and streams along individual
transects or in small watersheds (McGlynn and Seib-
ert 2003; Jencso et al. 2009), but the transferability
of the detailed knowledge available at small spatial
scales to larger watersheds or different landscapes is
limited (Jencso et al. 2009).
The Soil and Water Assessment Tool (SWAT;
Arnold et al. 1998) is a widely used, semi-distributed
watershed model that has been applied to a wide
range of spatial and temporal scales, environmental
conditions, land management practices, and land use
and climate change scenarios. It has proven to be an
effective and comprehensive tool for simulating
streamflow and pollutant transport, but a major limi-
tation of the model is that transport and deposition
processes in the landscape are not explicitly accounted
for. SWAT simulates runoff generation processes at
the level of hydrologic response units (HRUs), but
water and pollutant yields from the HRUs are
summed at subbasin level and added directly to the
corresponding stream. However, several approaches
to improving the simulation of landscape processes in
SWAT have been reported in the literature.
Arnold et al. (2010) developed a first version of a
SWAT landscape model by dividing a watershed in
Central Texas into three landscape units (LSUs), the
divide, hillslope, and valley bottom, and routed surface
runoff, lateral flow, and groundwater flow across a rep-
resentative catena. They only defined one HRU per
JAWRA
AND ALLEN
LSU but concluded that the catena approach was
promising for large-scale applications if HRUs were
simulated within each LSU. Bosch et al. (2010) tested
the catena approach on a hillslope in South-Central
Georgia. They found results to be encouraging and the
SWAT landscape model to be an important step toward
a more realistic representation of landscape flow and
transport processes. Sun et al. (2016) applied the
catena approach to a floodplain along a short section of
the Garonne River in Southern France by dividing it
into three LSUs representing the 1-year, 2- to 5-year,
and 10-year flood areas. Their main objective was to
simulate the exchange of surface water and groundwa-
ter in the alluvial plain using Darcy’s equation. SWAT-
LUD (Soil and Water Assessment Tool-Landscape Unit
Darcy) results showed good agreement with results
provided by a complex hydraulic model. A denitrifica-
tion module was successfully added to the SWAT-LUD
model by Sun et al. (2017). Hoang et al. (2017) intro-
duced another modified version of SWAT called
SWAT_LS and tested it in the Odense River Basin in
Denmark. Within SWAT_LS, the watershed was
divided into upland areas and riparian zones and the
Riparian Nitrogen Model was added to simulate deni-
trification processes within the riparian zone. The
model was able to simulate water and nitrogen trans-
port from upland areas to riparian zones and to esti-
mate nitrogen removal in the riparian zones.
Evenson et al. (2015) focused on improving the
simulation of the effects of geographically isolated
wetlands on watershed hydrology. They developed a
SWAT model featuring spatially explicit representa-
tion of geographically isolated wetlands and simula-
tion of subsurface inflow into them. The model was
applied to the Nahunta Watershed in North Carolina
and results indicated that geographically isolated
wetlands substantially impacted the watershed
hydrology in the study area, which confirmed the
importance of integrating them in watershed models
and watershed management plans. A further refine-
ment of the modified SWAT model was presented by
Evenson et al. (2016) and successfully tested in the
Pipestem Creek Watershed in North Dakota. This
model facilitated simulation of flow between geo-
graphically isolated wetlands within a fill-spill net-
work and flow from uplands to geographically
isolated wetlands.
A new version of the model, SWAT+, was designed
to improve the runoff routing capabilities while main-
taining computational efficiency and ease of model
use (Bieger et al. 2017). In SWAT+, subbasins may be
divided into two or more LSUs. HRUs are defined
after LSU delineation, so each LSU will have its own
set of HRUs. The functionality to define multiple
HRUs per LSU represents an advancement of
SWAT+ compared to the SWAT landscape version
2 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
 REPRESENTING THE CONNECTIVITY
used by Arnold et al. (2010) and Bosch et al. (2010).
Flow generated in the HRUs is summed at LSU level
and can be routed from the LSU to any other spatial
object within the watershed, i.e., another LSU, a
channel, wetland, pond, reservoir, or aquifer. This is
an important prerequisite for improving the simula-
tion of water transfer at watershed scale and the
interactions of storm characteristics and spatial pat-
terns of watershed properties it is controlled by.
SWAT+ uses so-called connect files to define gravity-
controlled connections between different spatial
objects within a watershed. Surface runoff, lateral
flow, and tile flow can be sent separately or combined
as total flow and fractions of flow can be sent to dif-
ferent spatial objects, which allows for a very flexible
definition of connectivity in the model. The use of
spatial objects and connect files facilitates the inte-
gration of model modifications as developed by Even-
son et al. (2015, 2016) and Sun et al. (2016, 2017).
Geographically isolated wetlands can be defined as
LSUs or even as HRUs, as flow can be routed at
HRU level as well. SWAT+ LSUs can also be used to
discretize the floodplain into flood areas as suggested
by Sun et al. (2016). Additionally, the model has the
capability to route overbank flow to LSUs within the
floodplain, where it can evaporate, transpire, or infil-
trate. Retention and release back to the stream is
controlled by a user-defined lag factor.
In this study, we divided the subbasins in SWAT+
into two LSUs representing hillslopes and riparian
zones, the most basic units of a watershed according
to McGlynn and McDonnell (2003) and McGlynn
et al. (2004). Accordingly, the spatial discretization of
the watershed is very similar to the one presented by
Hoang et al. (2017). We use the terms uplands and
floodplains instead of hillslopes and riparian zones as
this terminology was established during SWAT+
model development. Rathjens et al. (2016) tested four
methods for delineating LSUs and concluded that two
of them, the Slope Position method and the Variable
Storage method, provided an adequate discretization
of the watershed into uplands and floodplains for
SWAT+ modeling purposes.
Currently, detailed observed data on connectivity
are not available for most watersheds, hampering
model validation. Nevertheless, we believe it is cru-
cial to critically analyze and test the new capabilities
of SWAT+ and discuss the opportunities and chal-
lenges arising from an improvement of the represen-
tation of internal watershed processes in the model.
Therefore, the objectives of this study were to (1) run
scenarios in SWAT+ with different ratios of surface
runoff sent from the uplands to the streams and
floodplains, (2) evaluate their impacts on the relative
role of runoff originating from uplands and flood-
plains and the buffering of runoff by the floodplains,
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
OF UPLAND AREAS TO FLOODPLAINS AND STREAMS IN SWAT+
and (3) discuss the challenges associated with repre-
senting hydrologic connectivity in SWAT+.
MATERIALS AND METHODS
Brief Introduction to SWAT+
SWAT+ is a completely restructured version of
SWAT that was developed to address present and
future challenges in water resources modeling and
management and to meet the needs of the growing
worldwide user community (Bieger et al. 2017).
SWAT+ has more flexibility in watershed discretiza-
tion and configuration than previous versions of
SWAT. In SWAT+, the different elements of a water-
shed, e.g., LSUs, HRUs, aquifers, ponds and reser-
voirs, inlets, point sources, and channels, are defined
as spatial objects. The user can define the hydrologic
interactions between different spatial objects to repre-
sent the physical characteristics of a watershed as
realistically as possible. SWAT+ uses a so-called con-
nect file to define the routing of flow from one spatial
object to one or more others. Each spatial object has
its own connect file. This is a substantial improve-
ment compared to previous versions of SWAT, where
all routing commands were summarized in one file
and routing of fractions of the same hydrograph to
different objects was only possible by modifying the
source code. Croke et al. (2005) distinguished
between two types of surface runoff connectivity,
direct connectivity via new channels and gullies and
diffuse connectivity via sheet flow. In the SWAT+
connect files, these two types of connectivity can be
considered. The user can send a defined fraction of
surface runoff generated in the uplands directly to
the stream, while the remainder is routed to the
floodplain. There, it provides an additional input of
water, thereby impacting soil water contents, percola-
tion, and evapotranspiration in the riparian zone.
Accordingly, the connect file controls the connectivity
of upland areas to the adjacent floodplain and the
corresponding stream. The structure of the connect
files is the same for all spatial objects that are
defined in SWAT+ (Table 1). Figure 1 shows a
schematization of the connectivity between LSUs,
channels, reservoirs, and aquifers as defined in the
SWAT+ connect files used in this study.
Study Area
The Little River Experimental Watershed (LREW)
is a 334 km2 large watershed located in the Upper
3 JAWRA
 BIEGER, ARNOLD, RATHJENS, WHITE, BOSCH, AND ALLEN

TABLE 1. Description of columns in the Soil and Water Assess-
ment Tool+ (SWAT+) connect files defining the routing of runoff
from one spatial object to a user-defined number of other spatial
objects.
Column name Description
OUT_TOT Total number of objects receiving runoff from
this object. The following four columns
are repeated accordingly.
OBJ_TYP The type of the receiving object
(e.g., LSU, CHA, RES, AQU).
OBJ_ID The ID of the receiving object.
HYDRO_TYP The type of hydrograph to be sent to the receiving
object (e.g., SUR, LAT, RHG).
FRAC The fraction of the hydrograph to be sent to the
receiving object. Fractions of the same
hydrographs can be sent to multiple receiving
objects as long as the fractions add up to 1.
Note: LSU, Landscape Unit; CHA, channel; RES, reservoir; AQU,
aquifer; SUR, surface runoff; LAT, lateral flow; RHG, recharge.
Suwannee River Basin in South-Central Georgia (Fig-
ure 2). It is a USDA-ARS (Agricultural Research Ser-
vice of the United States [U.S.] Department of
Agriculture) regional experimental watershed estab-
lished in the late 1960s. Accordingly, long-term
hydrologic and climatic data are available for the
watershed (Bosch et al. 2007a).
The climate in the LREW is humid subtropical
with long, hot, and humid summers and short, mild
winters. High intensity, convective thunderstorms
frequently occur in summer, while frontal storms
with moderate rainfall amounts are typical of the
winter and spring months (Bosch et al. 1999). Annual
precipitation averages 1,208 mm and the annual
mean temperature is 19.1°C. Average annual evapo-
transpiration is estimated to be 69% of average
annual precipitation (Sheridan 1997).
The LREW is characterized by broad floodplains
and gently sloping uplands. Elevations range from 82
to 148 m m.s.l. Bosch et al. (2006) estimated that the
percentages of forest, agricultural land, urban areas,
and water are 50%, 41%, 7%, and 2%, respectively.
The dominating agricultural crops are cotton, peanuts,
and corn. Upland forests consist of pine, while the
dense riparian vegetation in the floodplains is made
up of a mix of hardwoods and evergreens. Typical soil
types in the LREW are loamy sands and sandy loams
(Sullivan et al. 2007). The soils are underlain by the
Hawthorn formation, which serves as an aquiclude
and thus restricts deep seepage to regional aquifers.
Accordingly, lateral movement of shallow baseflow
from the uplands to the floodplain is promoted (Sheri-
dan 1997), which leads to saturation excess conditions
along the streams (Bosch et al. 2010).
SWAT+ Setup and Scenarios
The data used for constructing the SWAT+ model
for the LREW and their sources are listed in Table 2.
Watershed delineation resulted in a total of 262 sub-
basins. All subbasins were subdivided into upland
areas and floodplains. The LSUs were delineated

FIGURE 1. Schematization of connectivity between LSU, which contain the hydrologic response units (HRU), CHA, RES, and AQU as
defined in the SWAT+ connect files. The arrows show the hydrograph type (TOT = surface runoff plus lateral flow for LSUs, reservoir out-
flow for RES, and streamflow for CHA; SUR; LAT; RHG; and the fraction that is being routed. The fractions are examples taken from the
connect files used in this study. The color of the arrows indicates in which connect file the connectivity is defined.

JAWRA 4 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

 REPRESENTING THE CONNECTIVITY OF UPLAND AREAS TO FLOODPLAINS AND STREAMS IN SWAT+

subbasin was divided into three upland and three

floodplain LSUs (left and right bank of the stream
and channel head). In all other subbasins, two upland
and two floodplain LSUs (left and right bank of the
stream) were defined. The watershed delineation
resulted in a total of 1,248 LSUs and 13,727 HRUs.
There were 609 upland/floodplain pairs, as 30 sub-
basins lay completely within the floodplain and thus
did not have an upland.
A peanut–corn–cotton–cotton rotation was imple-
mented for all cropland in the watershed. All areas
that were identified as water in the land use or the
soil map were defined as ponds in SWAT+. Their vol-
ume and drainage areas were estimated based on
guidelines for pond design published by the USDA-
NRCS (1997). The ratio of the drainage area of all
ponds in a LSU to the total area of the LSU was used
to determine the ratio of surface runoff and lateral
flow sent directly to the ponds. The remaining lateral
flow was routed to the adjacent floodplain, while the
remaining surface runoff was partly sent to the flood-
plain and partly to the corresponding stream. There
was one aquifer per LSU and all groundwater flow
from upland aquifers was routed to the adjacent
FIGURE 2. Location, topography, stream network, and floodplains floodplain aquifer. Flow from ponds was sent to the
(FP) of the Little River Experimental Watershed (LREW). corresponding stream. Accordingly, streams received
groundwater flow from floodplain aquifers only, lat-
eral flow from floodplain LSUs and ponds, and sur-
using the Slope Position method that was tested for face runoff from floodplain LSUs, upland LSUs, and
SWAT+ in the LREW by Rathjens et al. (2016). The ponds.
method first identifies valley bottoms and ridge tops Daily precipitation data from 27 stations within
based on a digital elevation model and subsequently and near the watershed and temperature, solar radi-
the slope position of each raster cell, i.e., its relative ation, relative humidity, and wind speed from one
position between valley bottom and ridge top. A user- station (Bosch et al. 2007b) were used to drive the
specified slope position threshold defines the bound- model. Average daily streamflow at the watershed
ary between upland and floodplain (Rathjens et al. outlet was provided by the USDA-ARS Southeast
2016). In the LREW SWAT+ setup, each headwater Watershed Research Laboratory (Bosch and Sheridan
2007). SWAT+ was run for 25 years including a 5-
year warm-up period from 1988 through 1992, a 10-
TABLE 2. Spatial input data used to set up the SWAT+ model for
the LREW.
year calibration period from 1993 through 2002, and
a 10-year validation period from 2003 to 2012. The
Data Resolution (m) Source Hargreaves equation was used for calculating poten-
tial evapotranspiration. Channel routing was simu-
Topography 30 United States Geological Survey
(USGS), National Elevation
lated using the Muskingum routing method.
Data. Accessed January 22, Calibration of the model was performed manually
2015, https://gdg.sc.egov and resulted in a very good model performance at the
.usda.gov/. watershed outlet.
Land use 30 USGS, National Land Cover Data Three scenarios were simulated to test how differ-
(NLCD), 2011 Edition. Accessed
January 22, 2015,
ent ratios of channelized overland flow that connects
https://gdg.sc.egov.usda.gov/. the uplands directly to the streams and sheet flow
Soil 30 U.S. Department of that travels through the floodplains impact the simu-
Agriculture-Natural Resource lation of streamflow and variables describing internal
Conservation Service, Soil Survey watershed processes (Table 3). For the first scenario
Geographic. Accessed
January 22, 2015, https://gdg.
(70CHA/30LSU), a ratio of 70% channelized overland
sc.egov.usda.gov/. flow and 30% sheet flow was assumed. Thus, 70% of
surface runoff generated in the uplands was sent

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 5 JAWRA

 BIEGER, ARNOLD, RATHJENS, WHITE, BOSCH,
TABLE 3. Description of scenarios.
Scenario Description
70CHA/30LSU 70% of SUR generated in the upland areas are
sent directly to the CHA, 30% to the FP.
30CHA/70LSU 30% of SUR generated in the upland areas are
sent directly to the CHA, 70% to the FP.
VAR_RATIO For each pair of upland and FP, the ratio of
channelized overland flow and sheet flow is
calculated based on the relative size of the
upland and FP area.
directly to the channel and 30% was sent to the flood-
plain. The second scenario (30CHA/70LSU) was
designed to contrast the first scenario, so the ratios of
channelized overland flow and sheet flow were
reversed, resulting in only 30% of surface runoff gen-
erated in the uplands being sent directly to the chan-
nel and 70% being sent to the floodplain. The third
scenario (VAR_RATIO) was based on the assumption
that the amount of runoff from an upland area that
can be retained in the floodplain depends on the rela-
tive size of upland and floodplain. In this scenario,
the proportions of upland surface runoff sent to the
channel and the floodplain varied across the
FIGURE 3. Simulated daily streamflow at the LREW outlet during one year of the calibration period (1997) and during a low flow period in
July/August 1997 for the three scenarios.
JAWRA
AND ALLEN
watershed depending on the relative area of upland
and floodplain as suggested by G€ untner and Bron-
stert (2004). For example, if the upland comprised
60% of the total area of an upland/floodplain pair and
the floodplain 40%, then 60% of the surface runoff
generated in the upland would be sent directly to the
channel and 40% to the floodplain. This method was
also successfully used by Hoang et al. (2017). Routing
of lateral and groundwater flow was the same in all
three scenarios. The scenarios were intentionally kept
very simple as this was considered expedient for a
first assessment of the new landscape routing func-
tionality in SWAT+.
RESULTS
The differences in streamflow at the LREW outlet
between the three scenarios were very small. The
peaks were slightly lower in the 30CHA/70LSU sce-
nario and slightly higher in the VAR_RATIO scenario
than in the 70CHA/30LSU scenario (Figure 3). Differ-
ences in streamflow between the three scenarios were
6 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
 REPRESENTING THE CONNECTIVITY
more pronounced during low flow periods than during
wet periods. Out of 609 upland/floodplain pairs in the
LREW, 430 (71%) had an upland area percentage of
more than 70% and uplands covered more than 90%
of the total area of 163 (27%) of the upland/floodplain
pairs (Figure 4). Therefore, the ratio of channelized
overland flow that was sent directly to the streams
and thus not buffered by the floodplain was higher in
the VAR_RATIO scenario than in the 70CHA/30LSU
scenario. In contrast, more water was traveling
through the floodplain in the 30CHA/70LSU scenario,
resulting in slightly lower streamflow peaks. The
Nash–Sutcliffe efficiency (NSE), coefficient of determi-
nation (R2), and percent bias (PBIAS) for daily
streamflow at the watershed outlet indicate a very
good model performance for all three scenarios
(Table 4). There were no differences in the NSE and
the R2 values between the three scenarios, but the
PBIAS was slightly higher for the 30CHA/70LSU sce-
nario and slightly lower for the VAR_RATIO scenario
than for the 70CHA/30LSU scenario.
The average annual amounts of flow delivered
from the landscape to the stream were 264.84,
261.63, and 266.1 mm for the 70CHA/30LSU, the
30CHA/70LSU, and the VAR_RATIO scenarios,
respectively. While the differences in total flow
FIGURE 4. Distribution of the fraction of upland (UL) area in all
UL/FP pairs in the LREW.
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
OF UPLAND AREAS TO FLOODPLAINS AND STREAMS IN SWAT+
TABLE 4. Model evaluation statistics during the calibration and
validation periods for the three scenarios (NSE, Nash–Sutcliffe effi-
ciency; R2, coefficient of determination; PBIAS, percent bias).
70CHA/30LSU 30CHA/70LSU VAR_RATIO
Calibration
NSE 0.89 0.89 0.89
R2 0.90 0.90 0.90
PBIAS 5.21 5.88 4.92
Validation
NSE 0.83 0.83 0.83
R2 0.83 0.83 0.83
PBIAS 0.93 1.88 0.58
delivered to the stream were very small, the relative
contributions of the flow components and source
areas within the watershed differed notably between
the three scenarios. In the 30CHA/70LSU scenario,
the ratios of lateral flow and especially groundwater
flow were higher than in the 70CHA/30LSU scenario,
whereas they were slightly lower in the VAR_RATIO
scenario. Accordingly, the ratio of surface runoff was
lowest in the 30CHA/70LSU scenario and highest in
the VAR_RATIO scenario (Figure 5). There were
notable differences in the source areas of surface run-
off delivered to the streams. In the 70CHA/30LSU
and VAR_RATIO scenarios, 65.61% and 75.52% of
the total surface runoff entering the stream came
from the upland areas of the watershed, respectively.
In contrast, the floodplains contributed 67.18% of the
total surface runoff delivered to the streams in the
30CHA/70LSU scenario.
Figure 6 shows the percentage of total surface
runoff delivered to the stream that originates from
the upland LSUs for the three scenarios. The LSUs
with a very low percentage of surface runoff enter-
ing the streams from the uplands in all three sce-
narios (as indicated by the red color in the maps)
have a large volume of ponds relative to the area of
the LSU. Therefore, most of their surface runoff was
routed to the ponds and very little surface runoff
was distributed between channelized overland flow
to the streams and sheet flow to the floodplains. In
the 70CHA/30LSU scenario, the percentage of total
surface runoff delivered to the stream originating
from the uplands ranged between 0% and 86.27%.
In the 30CHA/70LSU scenario, the highest percent-
age was only 56.09%, whereas in the VAR_RATIO
scenario, the uplands contributed up to 97.66% of
the total surface runoff entering the streams. The
map for the VAR_RATIO scenario in Figure 6 shows
that the percentage of total surface runoff delivered
to the stream that originated from the upland areas
was smaller along the lower reaches of the Little
River than along the headwater streams, indicating
a lower degree of upland connectedness to the
7 JAWRA
 BIEGER, ARNOLD, RATHJENS, WHITE, BOSCH, AND ALLEN

FIGURE 5. Basin average flow volumes and percentages of total average annual flow to the CHA originating from SURQ from UL and
FP, LATQ, and groundwater flow (GWQ) for the three scenarios. In the 70CHA/30LSU and 30CHA/70LSU scenarios, 70% and 30% of
SUR generated in the UL are routed directly to the CHA, respectively, and the remainder is routed to the corresponding FP.
In the VAR_RATIO scenario, the ratio of SURQ routed to the CHA and the FP varies based on the UL/FP area ratio. SURQ, surface runoff;
LATQ, lateral flow.

stream due to the wider floodplains in this part of DISCUSSION

the watershed.
Taking one randomly selected pair of upland and
floodplain as an example, Figure 7 shows that soil
moisture was generally higher and exhibited a stron-
ger reaction to rainfall events in the floodplains than
in the uplands. The differences between the three
scenarios were relatively small compared to the dif-
ferences between upland and floodplain, but consis-
tent with the differences in the ratios of surface
runoff entering the floodplain from the upland. Dur-
ing wet periods, the soil water content in the flood-
plain was generally so high that the differences
between the three scenarios were negligible. During
dry periods, there was a stronger increase in the soil
water content after rainfall events in the 30CHA/
70LSU scenario than in the 70CHA/30LSU scenario
and especially the VAR_RATIO scenario. The upland
soil moisture was the same in all three scenarios as
the upland water balance is calculated in SWAT+
before the water routing from upland to floodplain
LSUs takes place.
The differences between the simulated streamflow
at the outlet of the LREW for the three scenarios
were small as indicated by the hydrographs (Fig-
ure 3) and the model evaluation statistics (Table 4).
All three scenarios provided very good simulations of
streamflow at the watershed outlet, but they differed
considerably regarding their representation of inter-
nal watershed processes and runoff sources (Fig-
ure 5). When larger amounts of water are sent from
the uplands to the floodplains as in the 30CHA/
70LSU scenario, the relative roles of surface runoff
and groundwater flow change. The proportion of sur-
face runoff decreases, while the proportion of ground-
water flow increases, resulting in only small changes
in the total volume of water delivered to the stream.
This is a good example for the equifinality of hydro-
logic models and confirms the importance of a realis-
tic representation of runoff generation processes
within a watershed.

JAWRA 8 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

 REPRESENTING THE CONNECTIVITY OF UPLAND AREAS TO FLOODPLAINS AND STREAMS IN SWAT+

FIGURE 6. Percentage of total SURQ delivered to the stream originating from UL LSU in the 70CHA/30LSU (left), 30CHA/70LSU (middle),
and VAR_RATIO (right) scenarios. In the 70CHA/30LSU and 30CHA/70LSU scenarios, 70% and 30% of SUR generated in
UL LSU are routed directly to the CHA, respectively, and the remainder is routed to the corresponding FP. In the VAR_RATIO scenario, the
ratio of SUR routed to the CHA and the FP varies based on the UL/FP area ratio.

No observed data are available to verify the three
scenarios and decide which one simulates the hydro-
logic connectivity of uplands, floodplains, and streams
in the LREW most realistically. However, several
experimental studies have demonstrated that the vol-
ume of runoff entering the floodplain from upland
areas varies spatially along the stream network. This
suggests that the VAR_RATIO scenario might be
superior to the other two scenarios. Also, McGlynn
and Seibert (2003) found that the connectivity
between uplands and streams is larger in headwater
catchments with narrow riparian zones than along
higher order streams with wide floodplains, a pattern
that is more obvious in the spatial distribution of the
upland runoff entering the stream in the VAR_RA-
TIO scenario than in the 70CHA/30LSU and 30CHA/
70LSU scenarios (Figure 6). Baseflow is estimated to
contribute about 53% of the streamflow in the LREW
(Bosch et al. 2017). All three scenarios overestimate
the proportion of groundwater flow, but the overesti-
mation is lowest in the VAR_RATIO scenario, which
again suggests that this scenario may be slightly
more realistic than the other two.
Figure 7 shows that distinguishing upland areas
and floodplains in SWAT+ leads to an improved rep-
resentation of the saturation status of these two basic
LSUs. McGlynn and McDonnell (2003) stated that
riparian zones are often characterized by higher soil
moisture and prolonged periods of saturation. Surface
runoff from upland areas contributes to the satura-
tion of floodplains (D’Orico and Rigon 2003; Vidon
and Hill 2004). In upland areas, the soil moisture def-
icit is higher between rainfall events (McGlynn and
McDonnell 2003). This is confirmed by the differences
in simulated soil moisture between upland and flood-
plain LSUs and by the change in soil water contents
in response to rainfall events during dry periods
when the floodplain areas in the LREW are not fully
saturated. A good representation of floodplain charac-
teristics and processes is crucial in the LREW. The
soils in the watershed are underlain by the imperme-
able Hawthorn formation, which restricts deep
seepage to regional aquifers. Accordingly, lateral
movement of shallow baseflow from the uplands to
the floodplain is promoted (Sheridan 1997), creating
saturation excess conditions along the streams (Bosch
et al. 2010). Shirmohammadi et al. (1986) demon-
strated that the hydrologic behavior of the watershed
is strongly affected by the storage capability of the
channel alluvium, which is highly dependent on
antecedent moisture conditions. Also, the high soil
moisture levels in the floodplain impact denitrifica-
tion processes and thus potentially influence water
quality.

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 9 JAWRA

 BIEGER, ARNOLD, RATHJENS, WHITE, BOSCH,
FIGURE 7. Soil water (SW) content in a randomly selected UL/FP pair during the year 1997 (PCP, precipitation).
Figure 5 demonstrates that the proportion of
streamflow that is contributed by groundwater flow is
higher in the 30CHA/70LSU scenario than in the
70CHA/30LSU and VAR_RATIO scenarios and the
surface runoff proportion is lower. This suggests a
buffering function of the floodplain. The surface runoff
entering the floodplain from the uplands is added in
SWAT+ to the precipitation falling on the floodplain
area. Therefore, the water added to the floodplain as
surface runoff from the upland might be evaporated,
infiltrate in the soil, and eventually recharge the
groundwater, or contribute to surface runoff generated
in the floodplain. The small changes in streamflow and
the relatively high ratio of groundwater flow in the
30CHA/70LSU scenario indicate that in the current
setup of SWAT+ and with the current parameteriza-
tion, most of the water reaching the floodplain from
the upland areas recharges the groundwater and even-
tually becomes groundwater flow. Accordingly, the
runoff signal from the uplands is modulated and
obscured by the floodplain. While this is important for
water quantity, it is expected to have an even bigger
impact on water quality as it affects sediment deposi-
tion and denitrification processes.
McGlynn and McDonnell (2003) found in a small
headwater catchment in New Zealand that the buffer-
ing potential of the riparian zone depended on the
JAWRA
AND ALLEN
hillslope runoff relative to the size of the riparian zone
and the flowpaths the runoff takes through the ripar-
ian zone. McGlynn and Seibert (2003) also interpreted
the ratio of upland and floodplain areas as an indicator
for the buffer capacity of the floodplain, but stress that
additional variables need to be taken into considera-
tion as well. Based on their findings, the VAR_RATIO
scenario can be assumed to provide more realistic sim-
ulations of floodplain buffering mechanisms than the
70CHA/30LSU and 30CHA/70LSU scenarios as the
ratio of upland surface runoff sent directly to the chan-
nel and upland surface runoff sent to the floodplain
varies spatially in the VAR_RATIO scenario based on
the proportions of upland areas and floodplains. In the
70CHA/30LSU and 30CHA/70LSU scenarios, the ratio
of channelized overland flow and sheet flow disregards
the relative sizes of uplands and floodplains. Thus, it is
not taken into account that surface runoff flowing from
a relatively large upland into a relatively small flood-
plain will most likely be too concentrated to be buffered
by the floodplain, whereas the chances for runoff to
slow down and eventually infiltrate are much higher
when the upland is relatively small and the floodplain
relatively large. Whether runoff infiltrates or not is a
key factor influencing hydrologic connectivity
(Bracken and Croke 2007; Zimmermann et al. 2014)
and the function of the floodplain as a sink for
10 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
 REPRESENTING THE CONNECTIVITY
pollutants. The proportions of surface runoff sent to
the floodplain and to the stream in SWAT+ can be used
to control the relative importance of concentrated and
dispersive overland flow and the degree of lateral
buffering in the floodplain.
The watershed response to rainfall/snowmelt events
is controlled by many variables, including the hillslope
curvature, slope, elevation, aspect, rainfall variability,
soil and bedrock properties, vegetation, and antece-
dent wetness (Jencso et al. 2009). Their relative influ-
ence and their interactions can be highly variable in
space and time. Different areas within a watershed
can become connected or disconnected seasonally or
even episodically during runoff events (McGuire and
McDonnell 2010). Detty and McGuire (2010) observed
a strong seasonal signature in the connectivity
between hillslopes and riparian zones in a watershed
in the Hubbard Brook Experimental Forest in New
Hampshire. Storm intensity and duration are impor-
tant factors controlling connectivity as well (Bracken
and Croke 2007). However, many storms last less than
a day and their duration and temporal variability are
not adequately captured by the time steps of observa-
tions and model simulations. Also, when there is only a
small number of climate stations over a large area, the
spatial variability of storm events within a watershed
is not sufficiently captured.
Jencso and McGlynn (2011) systematically analyzed
the impact of different factors describing the terrain,
geology, and vegetative cover on the spatial and tempo-
ral differences in stream network connectivity and con-
nectivity yield in the Tenderfoot Creek Experimental
Forest in Montana. They found that the explanatory
power of the variables they considered in their analysis
varied annually and seasonally. The upstream accu-
mulated area was the main factor controlling the
extent of connectivity, but other metrics of surface
topography, vegetation, and geology impacted the con-
nectivity yield by reducing or enhancing water flow
across connected LSUs. Their relative importance var-
ied depending on catchment wetness states. Studies
like this can help to identify the dominant variables
influencing the hydrologic response of a watershed and
to develop generalized conceptualizations of connectiv-
ity that can be used for model parameterization. Accu-
rate simulation of hydrologic connectivity and its
controlling factors is particularly important for the
identification and management of Critical Sources
Areas of pollution (Shore et al. 2013), pollutant sinks,
or when changes in water transfers at the watershed
scale due to changes in climate or land use are investi-
gated (Bracken et al. 2013). The value of scenario sim-
ulations is strongly increased when hydrologic
processes within the watershed are simulated accu-
rately and not only the streamflow hydrograph at the
watershed outlet (Smith et al. 2013).
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
OF UPLAND AREAS TO FLOODPLAINS AND STREAMS IN SWAT+
Many researchers call for a closer integration of
experimental studies and model development and
applications (Cloke et al. 2003; Smith et al. 2013). The
connectivity of uplands to streams is still not well
understood and quantified (Bracken and Croke 2007;
McGuire and McDonnell 2010) and advances in pro-
cess understanding are not adequately generalized to
be integrated into widely applicable hydrologic models
(McDonnell et al. 2007). Field studies at small to large
spatial scales are required to improve our understand-
ing of the factors controlling runoff generation in
watersheds (Jencso and McGlynn 2011) and to develop
general principles describing hydrologic connectivity
that are applicable across all environments (Hopp and
McDonnell 2009; Bracken et al. 2013). Knowledge
from plot-scale studies should be transferred to the
landscape scale with care though (Lane et al. 2009).
While internal watershed hydrologic response
dynamics should be simulated as accurately and realis-
tically as possible, it is important to maintain model
simplicity and parsimony and avoid introducing addi-
tional uncertainty when the model is applied in water-
sheds with limited information available for
parameterization (Ambroise 2004; Jencso and McG-
lynn 2011). Running models at spatial scales that are
fine enough to represent the heterogeneity of variables
controlling runoff generation and at time steps that
are small enough to capture its temporal dynamics
remains challenging (Lane et al. 2009). While fully dis-
tributed models can represent the spatial variability of
runoff generation mechanisms and their controlling
factors, they are data intensive and computationally
demanding (Shore et al. 2013). Semi-distributed mod-
els like SWAT+ are computationally more efficient but
offer a more simplified representation of spatial com-
plexity. In this case study, HRUs were still lumped at
LSU level and only one ratio of channelized surface
runoff and sheet flow was defined per LSU, so the loca-
tion of individual HRUs within the LSU did not impact
their connectivity to the stream. Nevertheless, the
delineation of LSUs in SWAT+ allows the user to take
the unique physical nature of uplands and floodplains
into account and thereby improves the simulation of
variations in runoff generation.
CONCLUSIONS
We presented SWAT+ model output for three sce-
narios with different ratios of channelized surface
runoff, which is sent from upland areas to the
stream, and sheet flow, which is sent from the
uplands to the floodplains, and discussed their
impacts on the relative role of runoff originating from
11 JAWRA
 BIEGER, ARNOLD, RATHJENS, WHITE, BOSCH,
different source areas within the watershed and the
buffering of runoff by the floodplains. In the first two
scenarios (70CHA/30LSU and 30CHA/70LSU), the
ratio of channelized and sheet flow was 70/30 and 30/
70, respectively, for all upland/floodplain pairs. In the
third scenario (VAR_RATIO), the ratio was calculated
for each upland/floodplain pair based on the upland/
floodplain area ratio. Differences in simulated
streamflow were limited to small differences in peak
flows, but the relative importance of flow components
and upland areas and floodplains as sources of sur-
face runoff changed notably. The more water was
routed from the uplands to the floodplains, the larger
the contribution of groundwater and the smaller the
contribution of surface runoff to streamflow was.
Also, the soil water content in the floodplains was
impacted during the dry summer and fall months.
The variable ratio scenario was found to provide
slightly more realistic results than the other two.
SWAT+ is a publicly available watershed model
that has enhanced runoff routing capabilities com-
pared to previous versions of SWAT and is thus cap-
able to more realistically represent hydrologic
connectivity within a watershed. Testing and analyz-
ing new capabilities of watershed models is a critical
step in model development, even if no observed data
are available for verification of model output and
analysis is limited to a general check of plausibility
based on expert knowledge and soft data. The study
presented here demonstrated that SWAT+ was cap-
able of simulating transport from uplands to flood-
plains and streams in the LREW in the U.S. Coastal
Plain. Additional case studies in different environ-
mental settings are needed to further test and vali-
date the new capabilities of SWAT+ in watersheds
dominated by different runoff and routing processes.
Also, the factors controlling hydrologic connectivity
that have been identified in field studies all over the
world and available indexes should be evaluated in
detail to determine their suitability to be included in
SWAT+. Finally, the representation of overbank flow
and floodplain retention processes in SWAT+ should
be tested and improved to better account for interac-
tions between channel and floodplain. This would fur-
ther improve the consistency of simulated and
observed internal watershed processes and properties
influencing runoff generation and water transfer from
upland areas to the watershed outlet.
LITERATURE CITED
Ambroise, B. 2004. “Variable ‘Active’ versus ‘Contributing’ Areas or
Periods: A Necessary Distinction.” Hydrological Processes 18:
1149–55. https://doi.org/10.1002/hyp.5536.
JAWRA
AND ALLEN
Arnold, J.G., P.M. Allen, M. Volk, J.R. Williams, and D.D. Bosch.
2010. “Assessment of Different Representations of Spatial Vari-
ability on SWAT Model Performance.” Transactions of the Amer-
ican Society of Agricultural and Biological Engineers 53 (5):
1433–43. https://doi.org/10.13031/2013.34913.
Arnold, J.G., R. Srinivasan, R.S. Muttiah, and J.R. Williams. 1998.
“Large Area Hydrologic Modeling and Assessment Part I: Model
Development.” Journal of the American Water Resources Associ-
ation 34 (1): 73–89. https://doi.org/10.1111/j.1752-1688.1998.tb
05961.x.
Arnold, J.G., M.A. Youssef, H. Yen, M.J. White, A.Y. Sheshukov,
A.M. Sadeghi, D.N. Moriasi et al. 2015. “Hydrological Processes
and Model Representation: Impact of Soft Data on Calibration.”
Transactions of the American Society of Agricultural and Biolog-
ical Engineers 58 (6): 1637–60. https://doi.org/13031/trans.58.
10726.
Bieger, K., J.G. Arnold, H. Rathjens, M.J. White, D.D. Bosch, and
P.M. Allen. 2017. “Introduction to SWAT+, a Completely
Restructured Version of the Soil and Water Assessment Tool.”
Journal of the American Water Resources Association 53 (1):
115–30. https://doi.org/10.1111/1752-1688.12482.
Bosch, D.D., J.G. Arnold, P.G. Allen, K.-J. Lim, and Y. Shik. 2017.
“Temporal Variations in Baseflow for the Little River Experi-
mental Watershed in South Georgia, USA.” Journal of Hydrol-
ogy: Regional Studies 10: 110–21. https://doi.org/10.1016/j.ejrh.
2017.02.002.
Bosch, D.D., J.G. Arnold, M. Volk, and P.M. Allen. 2010. “Simula-
tion of a Low-Gradient Coastal Plain Watershed Using the
SWAT Landscape Model.” Transactions of the American Society
of Agricultural and Biological Engineers 53 (5): 1445–56.
https://doi.org/10.13031/2013.34899.
Bosch, D.D., and J.M. Sheridan. 2007. “Stream Discharge Data-
base, Little River Experimental Watershed, Georgia, United
States.” Water Resources Research 43: W09473. https://doi.org/
10.1029/2006WR005833.
Bosch, D.D., J.M. Sheridan, and F.M. Davis. 1999. “Rainfall Char-
acteristics and Spatial Correlation for the Georgia Coastal
Plain.” Transactions of the American Society of Agricultural
Engineers 42 (6): 1637–44. https://doi.org/10.13031/2013.13330.
Bosch, D.D., J.M. Sheridan, R.R. Lowrance, R.K. Hubbard, T.C.
Strickland, G.W. Feyereisen, and D.G. Sullivan. 2007a. “Little
River Experimental Watershed Database.” Water Resources
Research 43: W09470. https://doi.org/10.1029/2006WR005844.
Bosch, D.D., J.M. Sheridan, and L.K. Marshall. 2007b. “Precipita-
tion, Soil Moisture, and Climate Database, Little River Experi-
mental Watershed, Georgia, United States.” Water Resources
Research 43: W09472. https://doi.org/10.1029/2006WR005834.
Bosch, D.D., J.M. Sheridan, and D.G. Sullivan. 2006. “Hydrologic
Impact of Land-Use Changes in Coastal Plain Watersheds.”
Transactions of the American Society of Agricultural and Biologi-
cal Engineers 49 (2): 423–32. https://doi.org/10.13031/2013.20416.
Bracken, L.J., and J. Croke. 2007. “The Concept of Hydrological
Connectivity and Its Contribution to Understanding Runoff-
Dominated Geomorphic Systems.” Hydrological Processes 21:
1749–63. https://doi.org/10.1002/hyp.6313.
Bracken, L.J., J. Wainwright, G.A. Ali, D. Tetzlaff, M.W. Smith,
S.M. Reaney, and A.G. Roy. 2013. “Concepts of Hydrological
Connectivity: Research Approaches, Pathways and Future Agen-
das.” Earth-Science Reviews 119: 17–34. https://doi.org/10.1016/
j.earscirev.2013.02.001.
Cloke, H.L., J.-P. Renaud, A.J. Claxton, J.J. McDonnell, M.G.
Anderson, J.R. Blake, and P.D. Bates. 2003. “The Effect of
Model Configuration on Modelled Hillslope–Riparian Interac-
tions.” Journal of Hydrology 279: 167–81. https://doi.org/10.
1016/S0022-1694(03)00177-X.
Croke, J., S. Mockler, P. Fogarty, and I. Tacken. 2005. “Sediment
Concentration Changes in Runoff Pathways from a Forest Road
12 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
 REPRESENTING THE CONNECTIVITY OF
Network and the Resultant Spatial Pattern of Catchment Con-
nectivity.” Geomorphology 68: 257–68. https://doi.org/10.1016/
j.geomorph.2004.11.020.
Detty, J.M., and K.J. McGuire. 2010. “Topographic Controls on
Shallow Groundwater Dynamics: Implications of Hydrologic
Connectivity Between Hillslopes and Riparian Zones in a Till
Mantled Catchment.” Hydrological Processes 24: 2222–36.
https://doi.org/10.1002/hyp.7656.
D’Orico, P., and R. Rigon. 2003. “Hillslope and Channel Contribu-
tions to the Hydrologic Response.” Water Resources Research 30
(5): 1113. https://doi.org/10.1029/2002WR001708.
Evenson, G.R., H.E. Golden, C.R. Lane, and E. D’Amico. 2015.
“Geographically Isolated Wetlands and Watershed Hydrology: A
Modified Model Analysis.” Journal of Hydrology 529: 240–56.
https://doi.org/10.1016/j.jhydrol.2015.07.039.
Evenson, G.R., H.E. Golden, C.R. Lane, and E. D’Amico. 2016. “An
Improved Representation of Geographically Isolated Wetlands
in a Watershed-Scale Hydrologic Model.” Hydrological Processes
30: 4168–84. https://doi.org/10.1002/hyp.10930.
G€
untner, A., and A. Bronstert. 2004. “Representation of Landscape
Variability and Lateral Redistribution Processes for Large-Scale
Hydrological Modelling in Semi-Arid Areas.” Journal of Hydrol-
ogy 297: 136–61. https://doi.org/10.1016/j.jhydrol.2004.04.008.
Hoang, L., A. van Griensven, and A. Mynett. 2017. “Enhancing the
SWAT Model for Simulating Denitrification in Riparian Zones
at the River Basin Scale.” Environmental Modelling & Software
93: 163–79. https://doi.org/10.1016/j.envsoft.2017.03.017.
Hopp, L., and J.J. McDonnell. 2009. “Connectivity at the Hillslope
Scale: Identifying Interactions between Storm Size, Bedrock
Permeability, Slope Angle and Soil Depth.” Journal of Hydrol-
ogy 376: 378–91. https://doi.org/10.1016/j.jhydrol.2009.07.047.
Jencso, K.G., and B.L. McGlynn. 2011. “Hierarchical Controls on
Runoff Generation: Topographically Driven Hydrologic Connec-
tivity, Geology, and Vegetation.” Water Resources Research 47:
W11527. https://doi.org/10.1029/2011WR010666.
Jencso, K.G., B.L. McGlynn, M.N. Gooseff, K.E. Bencala, and S.M.
Wondzell. 2010. “Hillslope Hydrologic Connectivity Controls
Riparian Groundwater Turnover: Implications of Catchment
Structure for Riparian Buffering and Stream Water Sources.”
Water Resources Research 46: W10524. https://doi.org/10.1029/
2009WR008818.
Jencso, K.G., B.L. McGlynn, M.N. Gooseff, S.M. Wondzell, K.E.
Bencala, and L.A. Marshall. 2009. “Hydrologic Connectivity
between Landscapes and Streams: Transferring Reach- and
Plot-Scale Understanding to the Catchment Scale.” Water
Resources Research 45: W04428. https://doi.org/10.1029/
2008WR007225.
Lane, S.N., S.M. Reaney, and A.L. Heathwaite. 2009. “Representa-
tion of Landscape Hydrological Connectivity Using a Topograph-
ically Driven Surface Flow Index.” Water Resources Research
45: W08423. https://doi.org/10.1029/2008WR007336.
McDonnell, J.J., M. Sivapalan, K. Vache, S. Dunn, G. Grant, R.
Haggerty, C. Hinz et al. 2007. “Moving Beyond Heterogeneity
and Process Complexity: A New Vision for Watershed Hydrol-
ogy.” Water Resources Research 43: W07301. https://doi.org/10.
1029/2006WR005467.
McGlynn, B.L., and J.J. McDonnell. 2003. “Quantifying the Rela-
tive Contributions of Riparian and Hillslope Zones to Catchment
Runoff.” Water Resources Research 39 (11): 1310. https://doi.org/
10.1029/2003WR002091.
McGlynn, B.L., J.J. McDonnell, J. Seibert, and C. Kendall. 2004.
“Scale Effects on Headwater Catchment Runoff Timing, Flow
Sources, and Groundwater-Streamflow Relations.” Water
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
UPLAND AREAS TO FLOODPLAINS AND STREAMS IN SWAT+
Resources Research 40: W07504. https://doi.org/10.1029/
2003WR002494.
McGlynn, B.L., and J. Seibert. 2003. “Distributed Assessment of
Contributing Area and Riparian Buffering along Stream Net-
works.” Water Resources Research 39 (4): 1082. https://doi.org/
10.1029/2002WR001521.
McGuire, K.J., and J.J. McDonnell. 2010. “Hydrological Connectiv-
ity of Hillslopes and Streams: Characteristic Time Scales and
Nonlinearities.” Water Resources Research 46: W10543. https://d
oi.org/10.1029/2010WR009341.
Rathjens, H., K. Bieger, I. Chaubey, J.G. Arnold, P.M. Allen, R.
Srinivasan, D.D. Bosch, and M. Volk. 2016. “Delineating Flood-
plain and Upland Areas for Hydrologic Models — A Comparison
of Methods.” Hydrological Processes 30 (23): 4367–83. https://d
oi.org/10.1002/hyp.10918.
Sheridan, J.M. 1997. “Rainfall-Streamflow Relations for Coastal
Plain Watersheds.” Applied Engineering in Agriculture 13 (3):
333–44.
Shirmohammadi, A., J.M. Sheridan, and L.E. Asmussen. 1986.
“Hydrology of Alluvial Stream Channels in Southern Coastal
Plain Watersheds.” Transactions of the American Society of
Agricultural Engineers 29 (1): 135–42. https://doi.org/10.13031/
2013.30116.
Shore, M., P.N.C. Murphy, P. Jordan, P.-E. Mellander, M. Kelly-
Quinn, M. Cushen, S. Mechan, O. Shine, and A.R. Melland.
2013. “Evaluation of a Surface Hydrological Connectivity Index
in Agricultural Catchments.” Environmental Modelling & Soft-
ware 47: 7–15. https://doi.org/10.1016/j.envsoft.2013.04.003.
Smith, T., L. Marshall, B. McGlynn, and K. Jencso. 2013. “Using
Field Data to Inform and Evaluate a New Model of Catchment
Hydrologic Connectivity.” Water Resources Research 49: 6834–
46. https://doi.org/10.1002/wrcr.20546.
Sullivan, D.G., H.L. Batten, D. Bosch, J. Sheridan, and T. Strick-
land. 2007. “Little River Experimental Watershed, Tifton, Geor-
gia, United States: A Geographic Database.” Water Resources
Research 43: W09471. https://doi.org/10.1029/2006WR005836.
Sun, X., L. Bernard-Jannin, C. Garneau, M. Volk, J.G. Arnold, R.
Srinivasan, S. Sauvage, and J.M. S anchez-Perez. 2016.
“Improved Simulation of River Water and Groundwater
Exchange in an Alluvial Plain Using the SWAT Model.” Hydro-
logical Processes 30: 187–202. https://doi.org/10.1002/hyp.10575.
Sun, X., L. Bernard-Jannin, S. Sauvage, C. Garneau, J.G. Arnold,
R. Srinivasan, and J.M. S anchez-Perez. 2017. “Assessment of
the Denitrification Process in Alluvial Wetlands at Floodplain
Scale Using the SWAT Model.” Ecological Engineering 103 (B):
344–58. https://doi.org/10.1016/j.ecoleng.2016.06.098.
USDA-NRCS. 1997. “Ponds — Planning, Design, Construction.”
Agriculture Handbook Number 590, 85 pp. https://www.nrcs.
usda.gov/Internet/FSE_DOCUMENTS/nrcs144p2_030362.pdf.
Vidon, P.G.F., and A.R. Hill. 2004. “Landscape Controls on the
Hydrology of Stream Riparian Zones.” Journal of Hydrology
292: 210–28. https://doi.org/10.1016/j.jhydrol.2004.01.005.
Yen, H., R.T. Bailey, M. Arabi, M. Ahmadi, M.J. White, and J.G.
Arnold. 2014. “The Role of Interior Watershed Processes in
Improving Parameter Estimation and Performance of Water-
shed Models.” Journal of Environmental Quality 43: 1601–13.
https://doi.org/10.2134/jeq2013.03.0110.
Zimmermann, B., A. Zimmermann, B.L. Turner, T. Francke, and
H. Elsenbeer. 2014. “Connectivity of Overland Flow by Drainage
Network Expansion in a Rain Forest Catchment.” Water
Resources Research 50 (2): 1457–73. https://doi.org/10.1002/
2012WR012660.
13 JAWRA