Ground Water Modeling Using Cloud and Grid resources

The following tasks are suggested to perform in the next few mounts:

Task 1- Science in the Clouds: Demonstrated Use Cases

Groundwater modeling using modflow (the most widely used groundwater simulation model): can be used for water sustainability research, planning and management.

  • Large-scale ensemble runs
  • Coupled with optimization runs
  • Geospatial analysis and visualization
  • Coupled with large amount satellite data (GRACE data) and other geospatial analysis
  • Can be linked to climate model, land use model, surfacewater model, etc.
  • One case from Arizona (has six alternative conceptual models)
  • Another case from Texas (using Genetic Algorithms + Modflow + GRACE satellite data)


Information about Modflow
In terms of information abut Modflow, here are a few papers and links that might be of interest:

  1. Modflow web site where you can download the executable and source code: Modflow web site
  2. An easy-to-read paper about modflow development history: The History of Modflow
  3. A very detailed description of the software including solvers, physical parameters etc.: The Modflow-2000 Manual
  4. The modeling case from Arizona (The current case study we are carrying out): Modeling_Report_21.pdf


Modflow On Azure

  • Our work on integrating CyberGIS infrastructure and Modflow is on this page


Task 2- Hybrid Cyberinfrastructure (HPC/Cloud Integration): Options, Performance, Cost

Some Initial targets could be:

  • (TeraGrid, Condor and Open Science Grid)+Cloud
  • SAGA, BigJob API, ...

Example Reference: http://www.cct.lsu.edu/~sjha/select_publications/bigjob-cloudcom10.pdf

What are the use cases that are best suited for using such a hybrid CI. The following constraints shoud be considered:

  • Software licenses(eg. Matlab, ArcGIS): cannot obtain license automatically in a cloud environment.
  • Software migration/portability issues: too expensive to port to a different platform; not available in a cloud platform
  • On-demand, event-driven quality of service requirement: need to get resources unexpectedly (surge computing)
  • Is desktop dropbox style “drop-and-compute” desirable for users to interact with across desktop/TeraGrid/Cloud infrastructure boundaries?