Developing a flexible framework for producing river water temperature projections: summary


1. Summary of the Research Report of the Group of Chief Scientists

This project examined alternative approaches to model future river water temperatures for England. The results will help the Environment Agency develop temperature projections and inform measures to help adapt to a changing climate.

2. Background

Climate change is expected to modify freshwater ecosystems and water quality. Understanding how and where changes can take place and their likely magnitude will help predict changes in water quality and target management activity.

Patterns of water temperature change may not be the same as for projected air temperature change, for which information is readily available. This study aimed to understand the best approaches the Environment Agency could use to model future river water temperature projections for England to better inform future management actions.

3. Approach

The project reviewed the approaches available to model river temperature and grouped them into 4 categories: statistical, process-based, machine learning and hybrid.

The review included a review of the data and modeling requirements, as well as the advantages and limitations of each approach for producing river water temperature projections for England under climate change scenarios. . The most appropriate methodology for developing these projections given the current availability of data has been identified.

4. Results

Several promising approaches have been identified that could be suitable for modeling the temperatures of English rivers. “Spatial network models” (a statistical approach) represent the current best approach with promising machine learning / artificial intelligence techniques for future development. Data, purpose of application, and scale all limit the potential applicability of different approaches.

Decision trees have been developed to select water temperature models based on a methodological approach and the availability of data in space and time. A potential modeling approach, mixed effects regression, has been identified to make water temperature projections in English based on data availability constraints and the scale required for the estimates.

Additional steps required to develop a water temperature projection for the entire river network have also been considered.

5. Conclusion

The ability to generate water temperature projections for England under climate change scenarios is limited by the data available. However, a flexible modeling approach has been identified for making projections at certain locations, which can be extended as more data becomes available.

The study gives the Environment Agency an approach to produce an important source of data on climate impacts that now allow us to understand and prepare for climate impacts on water quality and river ecosystems.

6. Project details

This summary relates to the information from project SC200008, reported in detail in the following output:

  • Report: SC200008 / R
  • Title: Definition of a flexible framework for producing river water temperature projections
  • Project manager: Judy England, Chief Scientist’s Group

This project was carried out by the Research, Analysis and Assessment group of the Environment Agency, which provides scientific knowledge, tools and techniques to enable us to protect and manage the environment as effectively as possible.

Information: [email protected]

© Environment Agency


About Author

Leave A Reply