PONG.TAMU.EDU - Rob Hetland's home page

In another instance of acronym first, I decided to call my research group in the Oceanography department at Texas A&M the Physical Oceanographic Numerical Group (PONG). I know it's silly, but I like it.

I use computer models to understand physical and biological processes in the coastal ocean. I have applied three dimensional hydrodynamic models, in particular ROMS, to a variety of coastal environments, such as the Gulf of Maine and the Gulf of Mexico, and an even wider range of idealized configurations.

The majority of my current projects focus on buoyancy driven flow in estuaries and the continental shelf. Regions of interest include the Gulf of Maine and the Gulf of Mecixo.

Buoyancy-driven flow in the coastal ocean

I have been working on a number of different idealized buoyancy-driven flow problems. I have used salinity coordinates to examine wind mixing in a buoyant plume in the coastal ocean. I have used layer models to investigate mixing in the supercritical outflow region of narrow river outflows. I have also used idealized models to study buoyancy driven flow in estuaries, reexamining classic theories of estuarine circulation. I have also examined realistic buoyancy-driven flow in the Gulf of Maine and the Gulf of Mexico. These studies have focused on either model skill, or physical/biological interactions (such as the hypoxia study described below). My most relavent papers regarding this topic are listed here:

• Hetland, R. D. Water mass modification in near-field river plumes , J. Phys. Oceanogr., submitted. [PDF]
• Hetland, R. D. Water mass structure of wind-forced river plumes, J. Phys. Oceanogr., in press. [PDF]
• Hetland, R. D. AND R. P. Signell, Modeling coastal current transport in the Gulf of Maine, Deep See Res. II, accepted. [PDF]
• Hetland, R. D. AND W. R. Geyer, An idealized study of long, partially mixed estuaries, J. Phys. Oceanogr., 34, 2677-2691. [PDF]

Hypoxia on the Texas/Louisiana shelf

We are developing models of hypoxia on the Louisiana/Texas continental shelf as part of a project funded by the NOAA coastal ocean program. Preliminary efforts have focused on hydrodynamic modeling of the Mississippi/Atchafalaya river plume system. The structure of the density field is important in determining the amount of stratification, a key ingredient in the formation and maintanence of hypoxia; currents are important in transporting biological matierial both above and below the pycnocline.

Real-time surface current predictions over the Texas shelf

The Texas Automated Buoy System (TABS) measures surface currents for use in oil spill response and mitigation. We run a real-time now-cast/forcast of currents over the same region, forced by surface winds and climatological heat fluxes. We publish maps of the surface currents every day. Much more about the inner workings of creating an automated system like this can be found on Steve Baum's TGLO/TABS Tiki. All of the scripts we use to download and regrid the wind data, run the model, and produce the various outputs are freely available. This project is funded by the Texas General Land Office.

Here are some of the tools I use to get my work done. Most of these involve either tools to aid in integrating or analyzing numerical simulations of coastal ocean flow.

Cluster computing

We have been working on distributed computing, using commodity-off-the-shelf components. The cluster pictured to the left was built using inexpensive small form factor machines, connected with gigabit ethernet. This configuration performs remarkably well, given its low cost. We use NPACI ROCKS cluster software as a base, shuttle computers for the nodes (because of the small form factor and excelent cooling system), and a gigabit switch to link the whole thing together. For ROMS applications, we use lammpi as the MPI library. This tends to be about 10 to 15 percent faster than other libraries, such as MPICH. In the future, we plan to transition to using AMD based computers, connected with infiniband. The pathscale compiler seems very promising for this.

Python

Python is an extremely powerful, freely available scripting language similar to Perl or Ruby. What makes python interesting for me is the large number of mathamatical and scientific add-on packages available. The most essential is some sort of package creating and operating on large, multidimensional arrays, such as or Numeric. This, in addition to plotting packages such as matplotlib, makes python a viable replacement for similar commercial such as MATLAB. You can find out more by visiting my python page, were I describe the tools you need (and provide tools I have written) to perform model analysis using python.