Jesse López

Oregon Health & Science University

The Columbia River drains more than 250,000 square miles of Canada and the United States and meanders more than 1,200 miles through Washington and Oregon. Its estuary – the final, broad stretch of more than 100 miles as it flows to the Pacific Ocean – hosts an abundance of creatures, from microscopic plants to salmon and seals.

This is Jesse López’s laboratory. The Department of Energy Computational Science Graduate Fellowship (DOE CSGF) recipient is part of a team researching the estuary’s biogeochemistry – its physical, chemical, biological and geological processes and reactions. What they learn will help preserve its health and productivity.

López, a graduate student at Oregon Health & Science University, is part of a group led by António Baptista, who oversees the Center for Coastal Margin Observation & Prediction (CMOP), which studies the Columbia River estuary (CRE).

The group, Baptista says, wants to “understand estuaries as bioreactors – places where significant biogeochemical transformations occur” with help from microbes that decompose dead plants and animals. These reactions supply fundamental nutrients for plant and animal life.

But the CRE, unlike most other estuaries, has “very fast water-flushing, so there’s not a lot of time for transformations to occur,” Baptista says. This short water residence time inhibits reactions, López says, yet “there’s definitely a lot of fish and there’s a lot of biogeochemical activity in the system, so where does it happen?”

The answer, the researchers hypothesize, is biological hotspots, where microbial communities overcome the limitations of short water residence. López focuses on one candidate: the estuarine turbidity maxima (ETM).

As the name suggests, the ETM is a cloudy mix of sediment, plant and animal detritus and microbes that digest organic material. The ETM moves with tides and river flow but usually follows the point where salty ocean water meets river water.

Lopez/DEIXIS

CMOP models suggest that residence time through most of the estuary ranges from hours to a few days. In the ETM, it’s longer, but just how much is something López hopes to learn.

“What we need from Jesse is a quantitative description of these estuarine turbidity maxima from the perspective of their genesis, their dynamics, so that others then can look at the biogeochemistry inside,” Baptista says. The team wants models that can predict the effects of such factors as climate change, upstream dam management and coastal earthquakes.

It’s complicated, López says. “You have the ocean water that’s trying to go upstream and generally you have the river water that’s going downstream, so you have this convergence. It’s at this point where you have high levels of turbidity” and mixing. The computer models must capture it all, including sediment, detritus and microbes.

One of CMOP’s tools is SELFE (pronounced “self”), a code that models circulation in layered fluids. Although originally designed to simulate the CRE, researchers around the world use SELFE on similar problems.

The code uses tracer fields, applying the rules of physics and solving fluid dynamics equations to track qualities like salinity and temperature over time. López’s research introduced tracers for sediment – a more complex input because sediment sinks. His algorithms add a term for settling velocity and deal with other factors, like how sediment moves along the bed of the river or ocean.

With the models’ help, “we have a much better understanding of the physics and the dynamics that are causing the ETM to exist,” López says. He and his colleagues are estimating how long water stays in the ETM compared to elsewhere in the estuary and how much the ETM contributes to productivity.

But López was frustrated because SELFE’s performance stalled when running on more than 128 processor cores. He used his 2013 Argonne National Laboratory practicum to address the problems. Working with Jed Brown, an assistant computational mathematician, he analyzed SELFE’s performance and improved its scalability and workflow. He linked it to code libraries that SELFE can summon for routines that perform functions like solving equations. He also addressed input and output bottlenecks.

The improved code drastically cut the time to solution for most simulations, Baptista says. Researchers now can more easily simulate estuary circulation over multiple years, helping them understand the ecosystem’s variability.

Using modeling and other tools, the research group hopes to clarify some of the CRE’s basic properties, including whether it releases carbon or absorbs it from the environment. Researchers also want to understand when, which and why parts of the estuary generate their own energy while others depend on energy from elsewhere.

López is thrilled that HPC can help answer these some of these questions. He’s also happy to see his efforts bear fruit. “I work directly with folks engaged with policy and management” of the CRE. “Our model results are actually used in the system and have implications immediately. That’s incredible motivation” to continue his research.

Read the entire article in DEIXIS, the DOE CSGF annual. [PDF, pages 9-10]

Image caption: This simulation of estuarine dynamics shows the intrusion of dense salt water (colored contours) from the ocean on the left creating upstream velocity near the bed (in red) and downstream velocity elsewhere (in blue). The opposing velocity fields converge near where salinity is low. Particles are trapped there and suspended sediments concentrate (filled contours) in the estuarine turbidity maximum (ETM). Credit: Jesse López.