Wednesday, 31 January 2024
Hall E (The Baltimore Convention Center)
The Laguna Madre is a shallow estuarine system off the coast of South Texas that is important for the development of juvenile green sea turtles and economically valuable fishes. During the winter season, cold fronts pass through the Laguna Madre, causing the water temperature to cool rapidly. If water temperatures reach ≈8°C for at least 24 hours, it may cause marine life to become hypothermically-stunned (Shaver et al., 2017). Locals consider this a cold-stunning (CS) event. During CS events, sea turtles and fish can suffer from severe cold-related illnesses or even death.
Currently, there is an operational model that predicts water temperatures in the Laguna Madre up to 120-hour lead times (https://cbigrid.tamucc.edu/tpw/). The predictions are then communicated to community stakeholders so they may call a voluntary halt in the navigation of the barges, as well as organize rescue efforts for the sea turtles with the help of trained volunteers. Currently, the operational model does not capture uncertainty information. White et al. (2023) utilized an ensemble approach to capture the uncertainty of the water temperature predictions by perturbing air temperature predictions used as inputs in the model.
The goal of this research is to explicitly model uncertainty in the predicted water temperature, using a Probability Neural Network, in hopes of better communicating the uncertainty information to the stakeholders. Results indicate that the new method captured and visualized the uncertainty in the form of a probability region. The model is being evaluated by comparing the percentage of the observed measurements that fall within the probability region constructed around the predicted mean and +/- one predicted standard deviation.
Shaver, D. J., Tissot, P. E., Streich, M. M., Walker, J. S., Rubio, C., Amos, A. F., ... & Pasawicz, M. R. (2017). Hypothermic stunning of green sea turtles in a western Gulf of Mexico foraging habitat. PLoS One, 12(3), e0173920.
Currently, there is an operational model that predicts water temperatures in the Laguna Madre up to 120-hour lead times (https://cbigrid.tamucc.edu/tpw/). The predictions are then communicated to community stakeholders so they may call a voluntary halt in the navigation of the barges, as well as organize rescue efforts for the sea turtles with the help of trained volunteers. Currently, the operational model does not capture uncertainty information. White et al. (2023) utilized an ensemble approach to capture the uncertainty of the water temperature predictions by perturbing air temperature predictions used as inputs in the model.
The goal of this research is to explicitly model uncertainty in the predicted water temperature, using a Probability Neural Network, in hopes of better communicating the uncertainty information to the stakeholders. Results indicate that the new method captured and visualized the uncertainty in the form of a probability region. The model is being evaluated by comparing the percentage of the observed measurements that fall within the probability region constructed around the predicted mean and +/- one predicted standard deviation.
Shaver, D. J., Tissot, P. E., Streich, M. M., Walker, J. S., Rubio, C., Amos, A. F., ... & Pasawicz, M. R. (2017). Hypothermic stunning of green sea turtles in a western Gulf of Mexico foraging habitat. PLoS One, 12(3), e0173920.

