Professional Documents
Culture Documents
67
(a)
(b)
Figure 1: (a) DEM and subwatersheds and (b) wetlands and landuse of the study watershed
The watershed contains numerous wetlands in the history. However, these wetlands have lost
more than 60% of their areas over last two centuries mainly due to drainage for agricultural
purpose. According to their connectivity to the streams, wetlands can be classified into isolated
wetlands and riparian wetlands. In the study area, isolated wetlands are usually bogs and fens
that have no significant inflows or outflows and support marshlike vegetation, while riparian
wetlands are usually swamps and bottomlands that are dominated by trees and shrubs. These
areas are periodically inundated by water over the year and waterlogged conditions are
dominant in their ecosystem. At present, the total wetland area is about 39.2 km2 being 10.7%
of the watershed area. Among them, riparian wetlands have 10.2 km2 and isolated wetlands
have 26.4 km2, which are 27.8% and 72.2% respectively of the wetland area in the watershed
(Table 1). Several large swamps exist in the upper part of the watershed (Figure 1b). The
largest one has an area of 2.1 km2. Riparian buffer width in the watershed varies generally
from 10 to 50 meters for the main stream, and 3 to 10 meters for the small tributaries.
The geospatial data of 1010 m DEM, stream network, and wetland coverage were provided
by the Grand River Conservation Authority, Ontario. These dada are used as inputs to the
interface to delineate different drainage areas in the watershed. The landuse and riparian
buffer data was classified from SPOT and Landsat imageries acquired in August of 2004 and
2005 respectively (Bonnycastle 2006). The soil type data was obtained from the Canadian Soil
Information System (CANSIS) database of Agriculture and Agri-Food Canada. The digital data
of DEM, soil type and landuse are combined together to derive spatial model parameters for
estimating runoff and sediment input into wetlands and riparian buffers.
68
69
70
71
After loading the drainage division extension into an ArcView project, a new view is created,
and the user is asked to provide necessary base data into the new view including a DEM grid,
and coverages of digitized stream network and wetland distribution. From the views menu,
user can perform individual tasks such as filling sinks, determining flow direction and
accumulation, delineating streams and tributaries, defining stream links, orders and
subwatersheds. The major tasks are undertaken using the drainage division request, by which
different drainage areas are divided using the approaches described above.
The outputs of the ArcView project are grids of drainage areas for isolated wetlands,
concentrated flow, riparian wetlands and non-wetland riparian buffers. Intermediate outputs of
the project include grids of isolated wetlands, riparian wetlands, streams, small tributaries or
ditches, stream links, stream orders, and subwatersheds. These datasets can be used for
additional spatial analysis, model parameter derivation and layout visualization. Characteristics
of different drainage areas within each subwatershed can be computed after the request of
drainage division including area, perimeter, maximum flow length, average slope and elevation,
major soil type and landuse. These output parameters are stored in ArcView attribute tables
and text files for use in the hydrologic modeling. The interface also has functions for
characterizing wetlands, channels and riparian buffer geometric parameters, such as normal
and maximum volume and surface area of wetlands; bankfull width, depth and discharge;
width, slope, soil and landuse of riparian buffers. These parameters together with parameters
of different drainage areas are used for hydrologic simulations of wetlands and riparian buffers.
The ArcView Interface allows for the results to be displayed graphically and in map form.
These results can then be modified and exported using corresponding functions in ArcView.
Results and Discussion
Division of drainage areas
To demonstrate the applicability of the drainage division tools, the developed ArcView
extension is loaded into the SWAT project of the Fairchild Creek watershed, and the derived
grids of different drainage areas are used for further analysis of the SWAT runoff output. Using
the ArcView interface, the digitized stream network is burned into the 1010 m DEM of the
watershed. The stream network is delineated using a threshold area value of 700 ha, which
means a visible stream channel is initiated when the cells drained area becomes greater than
700 ha. The same threshold value is used in the SWAT stream network delineation for
modeling purpose. A threshold area value of 50 ha is chosen to delineate ditch network in the
study watershed. These ditches are assumed to be small tributaries to the streams during a
flood event and dried up during non-flood season (Figure 2). As a result, 31 subwatersheds
are divided based on the stream links (Figure 1a).
By incorporating the grids of wetland distribution, stream and ditch network, flow direction, and
the subwatershed, the drainage areas of isolated wetlands, concentrated flows, riparian
wetlands and non-wetland buffers are obtained using the developed ArcView interface. In
order to be compatible with the SWAT simulation results, the contributing areas draining to
isolated wetlands, concentrated flows, riparian wetlands, and riparian buffers are lumped
within each subwatershed. To do so, the identification numbers of these areas are assigned
the same number as their corresponding subwatershed and reach segment. The drainage
division results of subwatershed 3 and 7 are presented in Figure 3 and Figure 4.
Subwatershed 3 is a headwater subwatershed situated in the upper left part of the study area
(Figure 1a). The subwatershed covers an area of 13.7 km2 with an average slope of 2.3% and
an average elevation of 274 m. The lengths of streams and ditches are 2.4 km and 11.2 km
respectively. Isolated wetlands have cover surface areas of 1.7 km2 and are distributed mostly
in the upper and middle parts of the subwatershed (Figure 3). These wetlands are assumed to
have no surface flow connection to the streams under low flow condition, and form three major
concentrated flow tributaries when water storage in the isolated wetlands exceed their normal
72
73
74
75
100
Observed
Simulated without wetlands
Simulated with wetlands
80
60
40
20
Jan-99
Apr-99
Jul-99
Oct-99
Feb-00
May-00
Aug-00
Dec-00
76
3.0
2.5
Scenario 1
Scenario 2
Scenario 3
2.0
1.5
1.0
0.5
0.0
1999
2000
2001
2002
2003
Figure 6. Simulated annual sediment yield at the watershed outlet for various scenarios (1)
neglecting the effect of riparian buffers, (2) considering upland contribution areas to the
riparian buffers and (3) assuming all subwatershed area drains to riparian buffers
Obviously, simulation without considering the effects of stream riparian buffers gives the
highest sediment yield from the watershed, which is about 2.5 t/ha per year over the period
1999 - 2003. A good match can be achieved between simulated and observed sediment yield
by adjusting model parameters during calibration. However, this may lead to less estimation of
soil erosion or over estimation of in-stream sediment deposition because the effects of riparian
buffers are not taken into account. The second scenario simulates the effects of riparian
buffers with upland input from the divided riparian buffer contributing areas, while sediment
from outflow of isolated wetlands and from concentrated flow is assumed to bypass riparian
buffers without deposition. The simulated sediment yield in this case is about 2.1 t/ha per year
from the watershed, which reduces 14.6% of the sediment yield compared to the first scenario.
The third scenario assumes all surface runoff generated from the subwatershed contributes to
the stream riparian buffers. The simulated sediment yield in this case is about 0.7 t/h per year
from the watershed, which reduces 73.1% of the sediment yield compared to the first scenario.
Obviously, the function of riparian buffers to reduce sediment and nutrient loads is over
estimated because the effects of concentrated flow are not accounted for. This result is in
agreement with the statements of Dosskey et al. (2002) and Verstraeten et al. (2006) that
riparian buffers are highly effective at a field scale to reduce sediment load, but are less
effective at a watershed scale due to (1) overland flow convergence, which reduces the
sediment trapping efficiency, and (2) part of the sediment bypasses riparian buffers through
ditches, sewers and road surfaces. Because concentrated flow of surface runoff from
agricultural fields limits the capability of stream riparian buffers to remove sediment and
pollutants considerably, division of their contributing areas from the subwatershed may help
the model to give a reasonable evaluation result.
Discussion
The division of different contributing areas within a complex watershed is accomplished using
the geospatial data of DEM, stream network and wetland distribution. Therefore, the quality
and resolution of the DEM data have important influence on the drainage division result.
Quality refers to the accuracy of the elevation data, and resolution refers to the horizontal grid
spacing and vertical elevation increment. High quality and resolution of the DEM results in
more accurate division of the different contributing areas, and low DEM quality and resolution
77
78
Conclusions
In this study an ArcView GIS interface is developed for watershed delineation taking account
of isolated wetlands, concentrated flow, riparian wetlands and non-wetland riparian buffers.
The different surface contributing areas to these landscape features are delineated based on
geospatial data of DEM, wetland distribution and stream network. An approach of assigning
NODATA to the cells of isolated wetlands, tributary ditches and riparian wetlands is used to
determine flow direction grids and their contributing areas within a watershed. The drainage
delineation can be implemented for each individual wetland, tributary ditch and stream
segment, and also can be lumped into the four groups of drainage areas within each
subwatershed. The drainage division map of the watershed is obtained by overlapping the
grids of the four contributing areas. The interface can be loaded into any ArcView based
hydrologic models as an extension for site specific characterization of surface contributing
areas and preparation of spatial parameters for modeling purpose.
The use of the developed interface is demonstrated by coupling a watershed hydrologic model,
SWAT, and a riparian ecosystem management model, REMM, to evaluate the impacts of
isolated wetlands and riparian buffers on runoff attenuation and on sediment reduction at a
watershed scale. The interface is used to provide drainage and site characteristic parameters
in the SWAT and REMM modeling process. Simulation results from SWAT indicate that
accounting for isolated wetland processes in the model gives a better representation of the
outflow dynamics at the watershed outlet. More reasonable results from REMM are obtained
by computing upland inputs from the delineated contributing areas to the riparian buffers.
However, uncertainties may arise from the threshold values for delineating stream and ditch
networks. The choice of these threshold values for delineation of different contributing areas
within a watershed needs to be carefully justified for varies topographic and hydrologic
conditions.
The proposed approach takes advantages of spatially distributed data and GIS for watershed
analysis, and provides a tool for drainage division and parameter characterization of isolated
wetlands, riparian wetlands and riparian buffers. Coupled with a spatially distributed hydrologic
model, the developed interface can benefit the model potentially to (1) better simulate
hydrologic processes in a watershed containing a large numbers wetlands, (2) evaluate the
performance of wetlands and riparian buffers in reducing runoff, sediment and nutrient load,
and (3) assess the impacts of alternative wetland and riparian buffer restoration scenarios on
flow and water quality improvement. In addition, the approach makes a contribution in linking
watershed models and field-based models through divided contributing areas, and has a
potential to be used for evaluating various conservation practices in agricultural watersheds.
Acknowledgements
This research is jointly supported by Ducks Unlimited Canada, Agriculture and Agri-Food
Canada and Canadian Water Network. We thank Shane Gabor for research coordination, Dr.
Xixi Wang for research design suggestions, and the Grand River Conservation Authority for
providing data.
References
Arnold, J.G., Srinivasan, R., Muttiah, R.S. and Williams, J.R. (1998) Large area hydrologic
modeling and assessment-Part I: model development, Journal of American Water
Resources Association, 34(1), 73-89.
Arnold, J.G. and Fohrer, N. (2005) SWAT2000: current capabilities and research opportunities
in applied watershed modeling, Hydrological Processes, 19(3), 563-572.
79
80