 |
| Titel |
Calibration and Execution of SWAT Models over GRID Architecture |
| VerfasserIn |
Victor Ioan Bacu, Denisa Rodila, Lukasz Kokoszkiewicz, Elham Rouholahnejad, Danut Mihon, Ann van Griensven, Karim Abbaspour, Dorian Gorgan |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250046017
|
|
|
|
|
|
| Zusammenfassung |
| The presentation concerns with key concepts and architectures supporting calibration and
execution of SWAT (Soil Water Assessment Tool) [1] models over Grid infrastructures. The
Grid capabilities are required due to the large number of runs required in both calibration and
execution processes.
The assessment of the sustainability and vulnerability in the Black Sea Catchment is one
of the goal of the FP7-funded enviroGRIDS [2] project (April 2009 – March 2013). By
developing a Spatial Data Infrastructure for this catchment region, different scenarios based
on environmental changes and water quality models will be simulated.
One of the water quality models that will be used is SWAT (Soil Water Assessment Tool).
SWAT is a model designed to estimate impacts of land management practices on water
quantity and quality in complex watersheds. The SWAT model requires specific information
about weather, soil properties, topography, vegetation, and land management practices of
the watershed. The information is collected from stations distributed all over the
water basin. The density of these stations is closely related with SWAT input data
correctness.
The SWAT application will be integrated as a module in the enviroGRIDS portal. This
application will allow the calibration of the SWAT models and the execution of different
scenarios based on a calibrated SWAT model. The calibration process will use the SUFI2 [3]
uncertainty analysis routine. The flow for calibrating a model is the following:
randomly generate uncertainty model parameters, run the SWAT model in several
iterations until the objective function is meet and extract the required outputs that
correspond to the observed data from the model output files. Different scenarios could be
developed by hydrologic experts for executing on a calibrated model and the results of
executions could be displayed in different outputs and formats depending on the user’s
requirements.
Both the calibration and the execution require a great number of input and output data
files. All the SWAT model files are stored on the Storage Element inside the Grid
infrastructure. The input files are uploaded on Storage Element by the hydrologic model
developers. The calibration output is used for defining scenarios and for running these
scenarios. The visualization and report generation modules use the output data files of the
scenario execution.
In order to run the SWAT calibration on the Grid, a parallel approach at the data level
should be used. The Grid based approach will speed-up the entire calibration flow of the
model [4]. This means that the process could be split in multiple sub-processes that could run
simultaneously on Grid nodes. In conclusion, this solution could significantly reduce the total
execution time compared to the same process that runs on a single machine that in some cases
could take hours or even days. The execution over the Grid infrastructure is done by using
Ganga [5] and Diane [6] gLite’s software packages. Those packages are used as an upper
layer above the gLite middleware and provide useful functionalities in managing and
monitoring the jobs submitted on the Grid. The large user community of SWAT inside
the enviroGRIDS consortium may greatly benefit from a gridified version of the
software.
[1]. Soil and Water Assessment Tool, http://www.brc.tamus.edu/swat/index.html
[2]. enviroGRIDS project, http://envirogrids.net/
[3]. Karim C. Abbaspour, Jing Yang, Ivan Maximov, Rosi Siber, Konrad Bogner, Johanna
Mieleitner, Juerg Zobrist, Raghavan Srinivasan, Modelling hydrology and water quality in the
pre-alpine/alpine Thur watershed using SWAT, Journal of Hydrology (2007) 333, 413–
430
[4]. S. G.Yalewa, A. van Griensvena, L. Kokoszkiewiczb, Parallel computing of a large
scale spatially distributed model using the Soil and Water Assessment Tool, 2010
International Congress on Environmental Modelling and Software Modelling for
Environment’s Sake, Fifth Biennial Meeting, Ottawa, Canada
[5]. GANGA software, http://ganga.web.cern.ch/ganga/ |
|
|
|
|
|
|