Researchers have used a deep learning model to understand the predictive value of temperature, light and flow for driving oxygen in US rivers
The amount of dissolved oxygen in a river is a matter of life or death for the plants and animals living within it, but this oxygen concentration varies drastically from one river to another, depending on their unique temperature, light and flow.
To better understand which factor has the greatest impact on the concentration of dissolved oxygen, researchers at Penn State used a deep learning model to analyse data from hundreds of rivers across the United States.
Oxygen concentration is an important measure of water quality because fish and other aquatic organisms require dissolved oxygen to breathe, according to Wei Zhi, assistant research professor of civil and environmental engineering at Penn State and first author of the study.
Zhi said: “Studies have shown that three major factors – flow, temperature and sunlight – influence the amount of dissolved oxygen found in a river or stream.
“We wanted to know, at the US continental scale, which of these competing drivers was dominant.”
According to corresponding author Li Li, Barry and Shirley Isett Professor of Civil and Environmental Engineering at Penn State, the common perception is that all three factors matter: how quickly a stream flows impacts how fast oxygen in the air can dissolve in the water; temperature affects how much oxygen the water can pull from the air; and the level of sunlight shining into the water affects how much oxygen the plants in the water can make on their own.
Zhi added: “It is challenging, however, to figure out which of these factors is the most important at a continental scale because of different amounts of monitoring data in different rivers at different times.
“There has been little consistency in the way dissolved oxygen concentrations have been measured in different rivers.
“For example, some rivers were measured only in the 1980s in the summers, and some rivers were measured only in the 2000s in the spring.”
- Existential threats to iconic River Nile identified
- Revisiting 1964 and a unique river mussel study
- UK study highlights mental health and wellbeing benefits of rivers and canals
Temperature outweighs
Using 40 years of data from 580 rivers across the contiguous US – each with unique temperature, flow and sunlight conditions – the researchers trained a long short-term memory deep learning model to figure out the relationship between the weather conditions and dissolved oxygen.
Li said: “Traditionally, it has been very difficult to predict the dissolved oxygen levels on such a large scale, simultaneously with one model, but with a deep learning and big data approach, we can do that.
“Deep learning models enable large-scale systematic analysis of patterns and drivers.”
The model revealed that, at a continental scale, temperature outweighed light and stream flow in controlling the dissolved oxygen dynamic.
Light was the second-important factor on dissolved oxygen levels, while stream flow had minimal influence, according to the findings.
Zhi said: “Temperature is the predominant driver of daily dissolved oxygen dynamics in US rivers.
“Fairly accurate predictions of oxygen concentration can be made by temperature alone.
“Dissolved oxygen is declining in warming rivers, which has important implications for water security and ecosystem health in the future warming climate.”
The study is published in Nature Water.
Image: Raquette River, New York State. © Jondude11 (CC BY-SA 3.0) https://creativecommons.org/licenses/by-sa/3.0/legalcode.
