In this conversation, Rich Wildman, PhD, PE, addresses how water quality modelers can help target data collection to improve salmon habitat and comply with state and federal regulations.
About the Speaker
Rich is an environmental engineer with 20 years of experience focusing on water quality in reservoirs, rivers, and stormwater. His clients include Seattle City Light, Seattle Public Utilities, the Orange County Department of Public Works, the Tennessee Valley Authority, and industrial stormwater clients in Washington.
Q&A
When we talk about modeling for dam operators in the hydropower sector, what does that mean? What can a model do?
A model can help dam operators understand physical, chemical, and biological processes in the reservoir. It can help us characterize the water quality both within the reservoir itself and in downstream dam releases. This will afford dam operators the ability to optimize for certain water quality or biological attributes downstream, such as the transport of nutrients, or the temperature of the water, or fish habitat, including the promotion of salmon spawning. Additionally, we can use the computer model to run scenarios to test potential new operational strategies before implementing them in real life. This approach is useful for achieving management goals and complying with regulatory objectives.
How do you go about collecting the data that goes into that model?
Water quality models require a substantial amount of data, and collecting that data can involve multiple strategies, such as installing meteorological stations, gauging stations, or temperature measurement stations on tributaries to the reservoir. Field crews might also need to go out in boats every three to five weeks to collect water samples and vertical profiles of water quality parameters. The model also benefits from the extremely valuable data collects from a reservoir’s existing monitoring program.
Once samples are collected, they are analyzed in a laboratory, which incurs additional expenses. But with this data, we can create a model to help us understand much more deeply. For example, we input meteorological data and flow rates into a reservoir model, which then provides us with temperature and water quality characteristics in the reservoir and in the downstream releases from the dam.
How can an operator use modeling data to inform their decisions?
There are two main ways. First, the data helps us understand what is happening throughout the system. Second, it allows us to assess the effects of the hydropower project or its operations on water quality in the reservoir and downstream. This is particularly important for a FERC relicensing effort, where the regulatory body (FERC) needs to know the impact of the hydropower project on the river.
The data is fed into the model, which then provides output that helps us understand conditions between monitoring points. For example, if you have three monitoring stations across a 10-mile-long reservoir, the model can provide water quality characteristics at every point and depth in the reservoir, as long as we trust its output.
This is where the data becomes invaluable. It not only drives the model but also allows us to verify that the model’s output represents reality. The data helps us check the model’s accuracy and fine-tune it to deliver the best possible results.
In our industry, there can sometimes be a disconnect between those who collect field data and those who use computer modeling to create recommendations based on that data. Where do you think this disconnect originates?
I’ve experienced both sides of this. I’ve run models, managed water quality projects, and spent time in the field getting my hands wet and my feet dirty.
The general vibe is that field personnel are the ones touching the water and sediment and therefore the ones observing and experiencing the trends they aim to characterize, while modelers are dealing with the data in a way that is, by design, a simplification and approximation. Field personnel see firsthand the nuances that can explain variations in the system, while modelers tend to have a strong mechanistic understanding of the macro forces driving the trends.
From my experience, the strongest professionals are those who have experience in both fieldwork and modeling, because they can couple their hands-on experience with the abstract representation on a computer.
How do you establish the priorities for data collection in a monitoring program? What determines what they need to monitor?
The natural environment is complex, and when we think about the scientific method, as we learned it in middle school, we talk about having an independent variable and a dependent variable. For example, you might do a simple experiment where you grow a plant in the classroom. You water one plant more than the other. The first plant grows more, and the second plant grows less. You’ve varied the variable and observed an outcome.
In the lake, there are many variables changing all the time, and we cannot control them. We don’t have a way of determining whether it’s going to be a rainy summer or a windy winter. We just have to let it happen and try to react to it. The scientific method aims to reduce any human-caused variability on top of that. A good monitoring program standardizes as much as possible within logistical and safety constraints, in order to ensure consistency. It also involves creating a logistically reasonable plan that field crews can implement. A monitoring plan needs to be clearly written so that different people follow it the same way, ensuring that data is comparable across time and space, which allows us to draw meaningful conclusions.
Similarly, you wouldn’t schedule sampling for the last weekend of every month because holidays like Thanksgiving and Christmas would interfere. Instead, you need a consistent and clearly written monitoring program so that different people follow it the same way. This means data are comparable across time and space, allowing us to draw more meaningful conclusions.
Is there a way that modeling programs can be adapted to be useful for capital improvement projects?
Monitoring programs tend to be broad because it’s unclear which processes will matter most, so they aim to cover a wide range to explain whatever might come up. If you have a specific capital improvement program, such as our work for Seattle City Light, you might need a more detailed and intensive monitoring program or a model focused on a specific location or time. Often, a capital improvement program leads to changes, so you might use a model to test the likely effects of those changes.
The model can also be an important after a project is complete. The monitoring crew will continue monitoring, possibly keeping specific locations around the project for a few more years. This helps check if the model scenario matches monitoring results.
Another thing we consider is cost. Monitoring programs can be expensive, especially for remote locations where hydroelectric dams are. As modelers, we always need to be asking ourselves, how much data do we really need? To manage costs and effectiveness, we need to carefully evaluate the data to decide which elements of a monitoring program to enhance and which to discard. As we collect data, we begin to see that some things aren’t going to change much from year to year, and therefore, cutting that data collection can save a lot of overall field effort, not to mention money, while targeting our data collection more effectively.
What is a specific example where this kind of adaptation in action has best served the client?
Our extensive work with Seattle City Light. The City has been running a robust monitoring program across Ross Lake, Diablo Lake, and Gorge Lake, all part of the Skagit River hydroelectric project in northern Washington.
SCL initially invested in measuring algae biovolume and species in the reservoir, and we conducted taxonomy measurements of algae and zooplankton to understand the species. These measurements are expensive and time-consuming because field crews need to process samples in the evening after being on the water, and laboratories need to examine the samples under a microscope to identify the species present.
We concluded after the first year that algae and zooplankton concentrations are quite low in the Skagit River reservoirs, and that, therefore, it made sense to stop measuring these variables. We proposed instead to invest more into additional vertical profiles at more locations to measure temperature, dissolved oxygen, specific conductance, and pH. These are simpler measurements; temperature, specifically, is crucial for salmon; Downstream salmon and trout prefer cold water. As a result of the cost-savings of terminating the other monitoring, we were able to collect information at a few more locations to help calibrate the temperature model.
If a dam operator or a similar client was interested in replicating what Seattle did here with Geosyntec, what would be the first steps to take?
The first step is to define the goals of the entire effort. We need to determine the questions we’re trying to answer and decide whether those questions are best addressed with data or with the output from a computer model. If all we need to do is characterize the reservoir at three different times a year, we may not need a model at all and can design a straightforward monitoring program. We do this regularly.
If a question does require a model, understanding the purpose of the model will allow us to develop it. The first step is to define why you need a model and then let your modelers provide input on the goals and activities of the monitoring program.
We love it when we get to shape the data collected. If we can have some input into the data collected as part of a monitoring program, even better. It makes the modeling much more successful because it’s based on more accurate and useful data.
What’s one thing that might surprise dam operators about water quality modeling?
Modeling is more approachable than some people believe.
Modeling can seem like a black box, but it doesn’t have to be that way! We can break down the modeling process, explain what’s happening at each step, and show results from each step. Most of us like nothing more than to get into the details. I’m happy to show the actual equations if people are interested and explain the specific values that go into these equations to produce a temperature output, for example.
Sometimes the most interesting outcomes are when a model’s results don’t match the data, and we can explain why that is. If your modelers truly understand their models and can empathize with the dam manager or their field-based colleagues, they can clearly explain what the model is doing. We do this step-by-step, presenting clear figures that provide evidence for different processes, to show how we reach the answers we’re interested in.
More Information
For consultation, please contact Rich at RWildman@Geosyntec.com.
Hear Rich speak about Geosyntec’s services for dam water quality in reservoirs: Services for Dams Water Quality in Reservoirs | Rich Wildman (youtube.com)
Watch Rich explain the science and approaches to temperature total maximum daily load (TMDL) regulations: Water and Natural Resources: The Science and Approaches to Temperature TMDLs | Rich Wildman (youtube.com)
