The Open Hydrology Journal, 2007, 1, 19-24 19

WaterBase: SWAT in an open source GIS

C. George1 and L.F. Leon*2

1
International Institute for Software Technology, United Nations University, Macao, China
2
Department of Biology, University of Waterloo, Ontario, Canada

Abstract: WaterBase is a project of the United Nations University. Its aim is to advance the practice of Integrated Water
Resources Management (IWRM) in developing countries, by providing (a) free, open source tools for modeling and decision
support (b) a collection of IWRM resources: web sites, tools, literature, training material, etc. and (c) a community of partners
who can provide advice, support, contribute to tools and resources. A first step in the project is a tool to provide Geographic
Information System (GIS) support and a setup interface for the Soil and Water Assessment Tool (SWAT). This paper
describes the design of this tool, called Map Window SWAT (MWSWAT).

INTRODUCTION* users the possibility of adapting or extending them. This

possibility ranges from the localization of the interface to the
The WaterBase project (http://www.waterbase.org) of the
local language to the adding of significant functionality. The
United Nations University is aimed in particular, though not
use of open source tools also implies the use of the
exclusively, at developing countries. Predictive modelling and
corresponding open standards, such as those supported by the
decision support for water management in developing
Open Geospatial Consortium (OGC) [3].
countries are plagued with a number of related problems such
as: lack of money, lack of expertise, inadequate training There are a number of open source GIS systems available
capacity, and dependence on experts from other countries. At (OS GIS) [4]. WaterBase eventually chose MapWindow
the same time water resources are under increasing pressure (http://www.mapwindow.com) for three reasons. First, and
and aquatic ecosystems are being damaged by actions of critical when choosing any open source project, it was under
people who lack the resources to explore the consequences of active development. Second, unlike most open source
decisions before they are made. WaterBase aims to improve projects, it is native to Microsoft Windows, which is the
this situation by providing (a) tools for decision support, (b) operating system we expect most of our users to be currently
resources such as web sites, documentation, training material, using and accustomed to. Third, it had just been chosen by the
and case studies, and (c) a community of partners who can Environmental Protection Agency (EPA) in the US as the
advise and support other partners, and who can contribute to basis for version 4 of BASINS (EPA-Basins) [5], which gave
the tools and resources. us confidence in its future support. There were also technical
issues to be considered, such as whether MapWindow, could
SWAT (Srinivasan and Arnold, 1994) [1]; (Neitsch et al.,
support watershed delineation, and how easy it was to write an
2005) [2], has a substantial reputation as a model to quantify
extension for it, but technical problems can often be
the impact of land management practices in large, complex
overcome, while the basic issue of whether your chosen GIS
watersheds, and has been used in many developing countries
system will still be available and supported in 5 years time is
as well as in its home country, the US. Like other modeling
the most important issue.
tools, it requires a lot of data about terrain, landuse, soil, and
climate. There are two essential components needed to set up As it happens, MapWindow does have a watershed
SWAT models: (a) a GIS system to support the storage and delineation tool, using the TauDem software (Tarboton, 2001)
display of the relevant maps, and to perform the terrain [6]. In fact Taudem’s use of the Dinf approach to slope
analysis needed to delineate watersheds, to identify the stream directions, instead of the normal D8, promised better
reaches and the associated subbasins, etc., and (b) a watershed delineation than found in the current ArcSWAT
component that can generate all the files needed by SWAT, interface. MapWindow is also intended to be extensible
partly from the input maps and analyses, and partly by manual through the use of “plug-in” architecture, so it was in fact
editing. technically suitable. So an interface for setting up SWAT was
created based on MapWindow, and called MWSWAT. The
There is a substantial price tag on the current commercial
rest of this paper describes some of the details of MWSWAT
GIS system that is currently used by SWAT. The WaterBase
and what else needs to be done to provide decision support for
project decided to identify a suitable free, open source GIS
IWRM.
system, and then to produce the additional component that
would support the generation of SWAT input data. The use of DATA SOURCES
open source is important: it gives users confidence that tools
There is a considerable amount of data available on the
will not suddenly disappear with their original writers, or one
web, and MWSWAT is designed from the start to take
day become something you have to pay for, and also gives
advantage of that issue in mind. In particular it will be
*
delivered along with global data:
Address correspondence to this author at the Department of Biology,
University of Waterloo, Ontario, Canada; E-mail: [email protected] 1.-DEM maps: SRTM project (SRTM, 2004) [7]

1874-3781/07 2007 Bentham Science Publishers Ltd.

The Open Hydrology Journal, 2007, 1, 19-24 20

Fig. (1). Main Digital Data Pre-Processed in MapWindow

2.-Land: Global Land Cover Facility (Hansen,1998) [8] 2. HRU definition

3.-Soil maps: FAO (FAO/UNESCO, 2003) [9] 3. SWAT setup and run
4.- Precipitation and temperature data (NCDC) [10] This overall design concept is clear from the main
The increasing availability of such data opens a number MWSWAT form. For example Fig. (2) shows the main
of possibilities for its exploitation beyond water resource control form when the first two steps are completed and the
management. Additionally, users should not be restricted to third is ready to be started.
such data, because where local data exists it will generally be
finer grained and more accurate. But at the same time they
should not be prevented from doing some simulations even
when there is no local data. Fig. (1) shows an example of the
main digital source data (DEM, land and soil) being pre-
processed in MapWindow, from selection of files, clipping
and re-projecting.
DESIGN PHYLOSOPHY
Setting up a SWAT run is complicated. Generating a
thousand input files is not unusual, and so there are a vast
number of parameters to consider. The user can therefore
easily get lost in the process, and we need to keep a balance
between simplicity of the interface and access to everything Fig. (2). Main MWSWAT Form
the user might need to see and perhaps change. The first
priority is therefore to try to create a simple model of the WATERSHED DELINEATION
process that the users may have in their minds. We based the Watershed delineation uses a plug-in included in Map-
interface around three basic steps: Window as can be seen in Fig. (3). First the digital elevation
1. Watershed delineation. (DEM) is chosen, and options to burn in existing streams,

1874-3781/07 2007 Bentham Science Publishers Ltd.

The Open Hydrology Journal, 2007, 1, 19-24 21
and/or to use a mask for the watershed, may be selected. Then SWAT SETUP AND RUN
the threshold (minimum area to be designated as drainage for
The final step is to read the meteorological data, write the
a stream) is chosen. Finally outlets and inlets are selected,
SWAT files in the proper format and run the model. The
from an existing shapefile, or by creating one interactively.
SWAT Setup and Run form allows the user to select weather
sources (currently weather stations, plus precipitation and
temperature gauges), to choose the period of simulation, and
make a number of other choices as can be seen in Fig. (6).
Users can also choose to make detailed edits to the input
files using the SWAT Editor, can run SWAT itself, and can
save the output from the latest SWAT run. In brief, this form
is the main control of the model itself. It is fully based on the
current SWAT development and if required can be modified to
keep pace with future changes. Because the entire application
is open source, advanced users can even have access to the
source code and compile their own DLL for further
development.
DISPLAYING SWAT OUTPUT
Once the model produces the output files, the Swat2Dat
tool allows the selection of the desired sub-basin to use and
produces ascii files from the SWAT reach output file. This
can be directly imported into a spreadsheet for further analysis
and plotting. The export tool will be eventually incorporated
into the MWSWAT interface. For now it is an independent
program that works also for all the different versions of
ArcView SWAT. In order to verify the validity of the
application, a test watershed (San Juan River in central
Mexico) was prepared from scratch for two years of
meteorological data. The data used in the present example in
the MapWindows tool is from the web data described above.
Fig. (3). MapWindow Watershed Delineation Plug-in The model output is shown in Fig. (7). The current ArcMap
version for SWAT was also tested for a basin with calibration
After the watershed delineation process is completed, all
data available. It is worth noticing that both interfaces
the layers are displayed in the MapWindow view. A
ultimately run the same and most recent version of the model
delineated watershed is illustrated in Fig (4) with all the sub-
Swat2005. The results, in the testing basin, are as expected
watershed notation in place.
very similar for both platforms with respect to flows.
HRU CREATION Sediment yield has some differences that are under
investigation. Nutrients also behave very similar, but are also
SWAT uses the Hydrological Response Units (HRUs) as under scrutiny. The user lookup tables used are compared to
the basis for its modeling. HRUs may be formed per sub-
identify potential errors. A further study is in process to
basin (where a sub-basin is the area that drains into a reach of
validate the output between platforms.
the stream network), or as a division of a sub-basin based on a
particular combination of landuse, soil, and slope range. The CONCLUSIONS AND FUTURE WORK
Create HRUs form allows users to first select the landuse and
MWSWAT is complete at the time of writing, and will
soil maps, together with database tables (lookup tables), which
very soon be released. The next immediate technical task is to
relate the categories used in these maps to SWAT landuse and
provide some decision support capability, especially some
soil categories.
graphical support for viewing the SWAT outputs. This will
Then users can select intermediate slope percentages so as certainly include capability for drawing graphs or histograms,
to form bands of slopes. At this point the maps are read. especially for comparing outputs from runs with different
Then the user can choose singe HRUs (i.e. one per subbasin) input parameters, and also for showing, for example,
or multiple HRUs. In the second case the user removes small watersheds coloured according to user-chosen characteristics
HRUs, either by a selecting a minimum area, or by selecting such as sediment output. Another technical aim is to support
minimal percentages for landuse, soil and slope. Users may other kinds of models, such as event-based models which can
optionally also select subbasins at whose exit points reservoirs analyse the effects of storm events.
are situated, may choose to subdivide landuses into others, and
MWSWAT is the first of, we hope, many tools to support
may choose to exempt some landuses from the thresholds. In
IWRM. The next objective is to form a community of
Fig (5) the main form for HRU generation is shown.
partners who are interested in using and/or contributing tools

1874-3781/07 2007 Bentham Science Publishers Ltd.

The Open Hydrology Journal, 2007, 1, 19-24 22

Fig. (4). Delineated Watershed

Fig. (5). Create HRUs for MWSWAT

1874-3781/07 2007 Bentham Science Publishers Ltd.

The Open Hydrology Journal, 2007, 1, 19-24 23

Fig. (6). SWAT Setup and Run Dialog

Fig. (7). Example of SWAT Output

1874-3781/07 2007 Bentham Science Publishers Ltd.

The Open Hydrology Journal, 2007, 1, 19-24 24
and other resources to the project The existence of an active [2] Neitsch, S.L., Arnold, J.G., Kiniry, J.R. and Williams, J.R,
collection of users and developers will also be a critical factor (2005), Soil and Water Assessment Tool Theoretical
Documentation and User's Manual, Version 2005, GSWR
in finding organizations to support the project financially. Agricultural Research Service & Texas Agricultural Experiment
Station, Temple Texas.
ACKNOWLEDGMENTS [3] OGC, Open Geospatial Consortium,
http://www.opengeospatial.org.
The WaterBase project is a project of the United Nations [4] OS GIS, Open Source GIS, http://opensourcegis.org/
University, International Institute for Software Technology [5] EPA-Basins: Better Assessment Science Integrating Point and
(http://www.iist.unu.edu) and the International Network for Nonpoint Sources (BASINS). The BASINS home page is
Water, Environment and Health (http://www.inweh.unu.edu). http://www.epa.gov/ost/basins.
[6] Tarboton, D. and D. P. Ames. Advances in the mapping of flow
We have received collaborative support from the Daniel Ames networks from digital elevation data. In World Water and
and the MapWindow team at Idaho State University, from Environmental Resources Congress. ASCE, May 2001. See
David Tarboton at Utah State University, from Dave Swayne http://hydrology.neng.usu.edu/taudem/
[7] SRTM (2004), DEM data from International Centre for Tropical
and his group at the University of Guelph, from Karim Agriculture (CIAT), available from the CGIAR-CSI SRTM 90m
Abbaspour of Eawag, and, not least, from Raghavan Database: http://srtm.csi.cgiar.org.
Srinivasan and the SWAT team at Texas A&M University. [8] Hansen, M., R. DeFries, J. Townshend, and R. Sohlberg. 1 Km
Land Cover Classification Derived from AVHRR, 1998.
REFERENCES http://glcf.umiacs.umd.edu/data/landcover.
[9] FAO/UNESCO. Digital Soil Map of the World and Derived Soil
[1] Srinivasan, R. and J.G. Arnold. (1994). Integration of a Basin- Properties. Rev. 1. (CD Rom), 2003. Available from
Scale Water Quality Model with GIS. Water Resources Bulletin. http://www.fao.org/catalog/what_new-e.htm
Vol. 30, No. 3., June 1994. Pgs. 453-462. [10] NCDC, National Climatic Data Center from
http://www.ncdc.noaa.gov/oa/mpp/freedata.html

1874-3781/07 2007 Bentham Science Publishers Ltd.