EARLY HEARING DETECTION AND INTERVENTION VIRTUAL CONFERENCE
MARCH 2-5, 2021

(Virtually the same conference, without elevators, airplane tickets, or hotel room keys)

<< BACK TO AGENDA

5/25/2021 | 2:00 PM - 3:30 PM | Predicting lake temperature to understand ecological change using process-guided deep learning models | Virtual Platform

Predicting lake temperature to understand ecological change using process-guided deep learning models

Climate change impacts on lakes are complex and variable. Lake temperature data are lacking on the spatiotemporal scales and resolution required to understand how changing temperatures influence ecological processes. We integrated deep neural networks with knowledge of lake physics to develop process-guided deep learning models of lake temperature. We applied these models to 638 lakes in Minnesota, USA to predict continuous, daily water temperature profiles from 1980-2018. The median lake-specific root-mean square error was 1.75°C. On average, lakes got warmer, although rates of change varied. Surface and bottom temperatures increased most in fall months, and both surface and bottom temperatures of lakes cooled on average in April and May. The number of days in which lakes contained optimal temperatures for cold, cool, and warmwater fish increased on average, while the area of optimal thermal habitat generally decreased during the same period. Spawning dates occurred earlier for spring and summer spawners and later for fall spawners, although these trends varied among lakes. These models advance understanding of lake temperature response to climate change and demonstrate that the impacts of these changes on fish and other biota are not straightforward.

Climate change
Machine learning
Biological effects

Presentation:
This presentation has not yet been uploaded.

Handouts:
Handout is not Available

Transcripts:
CART transcripts are NOT YET available, but will be posted shortly after the conference

Presenters/Authors

Gretchen Hansen (), University of Minnesota, ghansen@umn.edu;