Welcome to our second installment of Puget Sound Nutrient Watch, an ongoing blog series that will focus on the excess nutrient problem in Puget Sound.
In this post we will be focusing on the Salish Sea Model and how scientific computer models help us better understand the world around us.
So...what is a scientific model?
Many children like to play with models — like doll houses, model cars or model airplanes. These toys are simplified versions of things we find in the world around us. These models might seem like they are just amusing toys on the surface, but they actually help you learn how the world works!
Similarly, scientists also use models. Scientists often rely on developing models to establish new science. Models are ideas that scientists use to explain patterns they observe in the world.
A computer model replicates some part of the environment in a way that helps us understand and predict potential changes. This allows scientists to ask the model questions like, "If the water temperature goes up three degrees, what happens to everything else?" If you were to change something in the model, it should tell you how changing the same thing in the environment would play out in nature.
Weather forecasting — for example — relies on computer models. Models are built to explain how aspects of the real world work and consist of ideas and concepts. Scientists use models to investigate the secrets of nature!
Characteristics of scientific models
Good scientific models should be:
- Grounded in scientific principles.
- Compared to a mechanism that is well understood.
- Calibrated to actual data so we can have confidence in the outputs.
- Be able to test if hypotheses are true or false.
New models are more likely to succeed if they dovetail or merge with existing scientific models. In fact — successful models often reveal that phenomena we once thought to be isolated are really connected incidents. Modern models use high-powered computers to compute millions of calculations!
It’s important to remember that a model is not the same as the real thing; they are similar in some respects, but not in all. A model becomes more refined and complex after many, many rounds of testing. Over time, the model looks less like the original simple model, and looks more like the real environment it is simulating.
We have confidence in a model's ability to simulate real life when it produces the same results that we find in actual data. If the model can accurately represent a known condition, we can use it to predict future conditions.
Why are models needed?
The most important advantage of computer modeling is that it gives us the ability to ask “what if?” questions about complex aspects of the world we can’t easily test in reality. Often, the focus of science is too small to be observed directly, or may be inaccessible for a direct visual study. For example, we may not be able to access the center of the earth to study it, but a computer model can give us a pretty good idea of what's happening!
New scientific discoveries depend upon scientists developing scientific models and interacting with them. Each hypothesis tested, simulation ran, and calibration made helps scientists better understand nature. After all, science is an attempt to explain our natural environment and make predictions about it.
What is the Salish Sea Model?
First of all, this model covers a huge area! The Salish Sea includes the Puget Sound, the Strait of Georgia, and the Strait of Juan de Fuca.
Our Environmental Assessment Program and staff from Pacific Northwest National Laboratory (PNNL) developed the Salish Sea water quality circulation model — called the Salish Sea Model for short — as a tool to broaden our understanding of how nutrients travel and ultimately affect water quality throughout Puget Sound.
We are using the Salish Sea Model to evaluate how current and potential future inputs of nutrients effect dissolved oxygen levels in the Salish Sea.
The Salish Sea model helps us understand questions like:
- Are human sources of nutrients in and around the Salish Sea significantly impacting water quality now? How bad might it get in the future?
- Where are the areas that are most sensitive to human impacts? When are those effects the most harmful?
- How much do we need to reduce human sources of nutrients to protect water quality in the Salish Sea?
We will be using the model to test the effects of short- and long-term actions to reduce anthropogenic — or human caused — sources of nutrients. We are aim to use this information to increase Puget Sound’s resilience to the effects of climate change and population growth. These model findings will help decision-makers use resources wisely and guide where additional study or action is necessary.
How have we used the model so far? Check out these publications:
Ask an engineer
We’ve asked one of our Salish Sea modeling engineers — Greg Pelletier — to participate in a Q&A about the Salish Sea Model, learn more from our interview!
What have we begun to learn from the Salish Sea Model?
We are starting to learn about the effect of regional anthropogenic sources of nutrients on changes in dissolved oxygen and acidification of the Salish Sea. We are finding that there are extensive areas of the Salish Sea that do not meet the water quality standards for dissolved oxygen. There are also areas where regional human-caused sources of nutrients appear to decrease dissolved oxygen to levels that do not meet the water quality standards.
How does this model help us understand ocean acidification?
The model allows us to look at how much impact is caused by regional anthropogenic sources and compared them with global sources. For example, in waters at the bottom of Puget Sound, the model predicts that the regional nutrient sources caused by humans has the highest impact when compared with estimates of combined global anthropogenic and nutrient impacts.
The model allows us to look at how much improvement would result from efforts to reduce nutrients to different levels. Also, the model allows us to look at the relationship between hypoxia (low dissolved oxygen) and acidification.
How long does does it take for us to run a scenario in the model?
The model uses a very powerful “cluster” computer at Battelle’s Pacific Northwest National Laboratory. It takes a couple of days for the computer to run all of the calculations for a single year for the original version of the model, and up to three days are required for our latest expanded version of the model.
We call each model run a scenario because it represents a particular set of assumptions. For example, one scenario is the existing conditions as they are in a particular year while another scenario is what we call “reference conditions” with human sources of nutrients removed.
In addition to the computer time, it takes time to make the input files for the model scenario – and to review the input files to assure their quality, which can vary depending on the complexity of a scenario. This step alone can take multiple people and several months.
After the model run is completed, time is required to convert the model output files into visualizations such as maps or animations of the predicted water quality, interpret what they are telling us, and discuss it with project team.
What do you like most about working with models?
The best part about working with models in my job is working with all of the great people we have on our modeling team. Our modeling team from Ecology includes Anise Ahmed, Cristiana Figueroa-Kaminsky, Sheelagh McCarthy, and Teizeen Mohamedali. We also collaborate with several great people from Battelle’s Pacific Northwest National Lab. It takes a team of people to do this work.
Different people do different parts of the model analysis. For example, some people work mainly in preparing model input information, while others work more on running the model or preparing the output visualizations. The entire modeling team works together to look at and discuss the model input files and the interpret the output visualizations. It is very rewarding to find out what the model runs tell us about the relationships between nutrient loading sources and changes in water quality throughout the Salish Sea.
Another amazing thing about working with models is its ability to predict water quality in areas where we have no monitoring data. We are confident of these predictions because the model performs well to replicate observed data when we compare it to areas where we have actual water quality measurements. Sometimes the results show us where we need to collect more data.
Visit our website and join our email list to get up-to-date information on the nutrient problem in Puget Sound. We want this to be a collaborative effort that brings all of the technical work that is happening on Puget Sound nutrients together. We need all hands on deck to find the best solutions for meeting water quality goals for Puget Sound.