EnviroGRIDS Project - Gridification of SWAT
What is SWAT?
The Soil and Water Assessment Tool (SWAT) is a physically-based continues-event hydrologic model developed to predict the impact of land management practices on water, sediment and agricultural chemical yields in large, complex watersheds with varying soils, land use, management conditions over long period of time. For simulation, a watershed is subdivided into a number of homogeneous subbasins (hydrologic response units or HRUs) having unique soil and land use properties. The input information for each subbasin is grouped into categories of weather; unique areas of land cover, soil and management within the subbasin; ponds/reservoirs; groundwater; and the main channel of reach, draining the subbasin. The loading and movement of runoff, sediment, nutrient and pesticide loadings to the main channel in each subbasin in simulated considering the effect of several physical processes that influence the hydrology.
Objectives of SWAT
- Simulation of processes at land and water phase
- Spatially distributed (different scales)
- Semi physically based / empirical approaches
- Simulation of changes (climate, land use, management etc.)
- Water quantities, incl. different runoff components
- Water quality: Nutrients, Sediments, Pesticides, Bacteria, (algae and oxygen), etc.
Gridification of SWAT
The main purpose of SWAT gridification is to enable Grid-computing for SWAT software.
For now, main objectives are:
- Enabling SWAT simulation to run using Ganga application
- Enabling SWAT uncertainty analysis to run using Ganga application.
- To build some kind of middle-ware for sending input and running SWAT specified jobs.
SWAT first grid run
SWAT simulation was run on LCG Backend (Balaton region model from enviroGRIDS project) using Ganga software. Lake Balaton region of research is about 2500km2 which is part of enviroGRIDS Black Sea research area (about 2 000 000km2). For Balaton Lake the amount of input data is 5.9MB and output is 49.3MB for 16 years simulation. Input and output data amount depends on research mesh resolution so it could be, as well, larger or smaller for any similar area.
Average one job execution time (for Balaton Lake using LCG backend) is about 40 minutes (strongly dominated by grid overhead).
Input and output data have very good compression factor which is about 7.
Model calibration and uncertainty analysis or calculating very large models splitted on subbasins requires very large number of swat runs which justifies using of computing grid.
SWAT2009 compilation
Written for Intel Fortran Compiler
Under Linux (c2d 2.0Ghz machine) compiling takes about 5 minutes 22 seconds
Next steps
Potential schemes for the gridification:
- gridififation of the SWAT-cup algorithms
- gridification of SWAT: sub-basins in parallel, followed by routing
- gridification of SWAT: HRU's in parallel, merged to sub-basins and followed by routing
Right now we are beginning working on scheme number 2.
Next step will be to provide some kind of service to send input and submit jobs.
Important Links
enviroGRIDS home page:
http://www.envirogrids.net/
SWAT home page:
http://www.brc.tamus.edu/swat/index.html
SWAT-CUP home page:
http://www.eawag.ch/organisation/abteilungen/siam/software/swat/index_EN
Download recourses
SWAT Source (and 32bit executable) for Linux:
ftp://ftp.brc.tamus.edu/pub/outgoing/bkomar/linux/swat/dev9.zip