Deep Learning-Based Algal Bloom Prediction for Lake Okeechobee Using Multi-Source Data Fusion

By Lindsay Steis
Slide 1: Title slide presenting Deep Learning-Based Algal Bloom Prediction for Lake Okeechobee Using Multi-Source Data Fusion

Slide-1

REU Scholar: Lindsay Steis

REU Mentors: Dr. Yufei Tang & Yingqi Feng

Slide 2: Overview of Harmful Algal Blooms including definitions and characteristics

Slide-2

Harmful Algal Blooms (HABs)

  • Overgrowth of algae
  • Large accumulation of phytoplankton
  • Eutrophication: excess nutrients
    • Rapid reproduction
    • Microcystis aeruginosa
Slide 3: Economic and ecological impacts of harmful algal blooms with focus on Lake Okeechobee

Slide-3

Economic & Ecological Impacts

  • Microcystin: Hepatotoxin
    • Drinking water supply
    • Closed tourist locations
    • Marine aquaculture & fisheries
  • Hypoxia - low O2
    • Mammal mortality - dead zones

Maps showing Lake Okeechobee: 2nd largest freshwater lake

Slide 4: Satellite sensors used for remote sensing with challenges in data collection

Slide-4

Satellite Sensors

  • Used for remote sensing (RS) images
  • Missing information - discontinuity
    • Dead pixels
    • Thick clouds
    • Sun glint
    • Water turbidity

(3) Lake Okeechobee set of six RS images from satellites 3A, 3B

Slide 5: Multi-source prediction model combining satellite data with simulated data and hydrodynamic models

Slide-5

Multi-Source Prediction Model

  • Hybrid dataset
    • Satellites – RS
    • Simulated data
    • Hydrodynamic-biological model
  • Forecasting prediction
    • Single-day
    • Rolling window

(3) True color RS image of central HAB

Slide 6: Advanced deep learning model using ConvLSTM with training and validation results

Slide-6

Advanced Deep Learning Model

  • Convolutional Long-Short Term Memory (ConvLSTM)
    • Captures temporal & spatial correlations in data simultaneously
    • 14-14 & 14-1 predictions

Two graphs showing Training & validation loss for 14-14 and 14-1: Convergence is observed showing model is learning from data

Slide 7: Prediction results for 7-7 model showing performance metrics

Slide-7

Prediction Results for 7-7

  • Root Mean Square Error (RMSE): 0.0033599667
  • Peak Signal-to-Noise Ratio (PSNR): 49.47330001523929
  • Structural Similar Index Measure (SSIM): 0.9916645337059432
Slide 8: Prediction results for 14-2 model with improved performance metrics

Slide-8

Prediction Results for 14-2

  • RMSE: 0.0029430606
  • PSNR: 50.624015366344025
  • SSIM: 0.9950491670550674
Slide 9: Challenges with HAB model including blur issues and data reconstruction needs

Slide-9

Challenges with HAB Model

  • Blurs in ground truth and prediction images
    • Model rolling prediction & downsampling
  • Increase RS usable data
    • Patching & reconstruction
Slide 10: Future work plans including dataset comparison and publication status

Slide-10

Future Work

  • Paper being reviewed
    • IEEE Journal
    • Listed in acknowledgement section
  • Yingqi changing mask
    • Match size for phys. and hybrid datasets
  • Upload to website - forecasting

Image: Hybrid Dataset: (83, 88)

Image: Phys. Dataset: (112, 112)

Slide 11: Thank you slide with questions prompt

Slide-11

THANKS!

Questions?

Slide 12: References page with three academic citations

Slide-12

References

(1) Galoustain, G. (2020, August). FAU awarded $2.2 million to monitor algal blooms in Lake Okeechobee. Florida Atlantic University. https://www.fau.edu/newsdesk/articles/habs-lake-okeechobee.php

(2) Lake Okeechobee Aquatic Plant Management Interagency Task Force. (2021). University Of Georgia - Center For Invasive Species And Ecosystem Health. https://www.floridainvasives.org/okeechobee/about/

(3) Tang, Y., Feng, Y., Fung, S., Ruiz Xomchuk, V., Jiang, M., Moore, T., & Beckler, J. (2022, July). Deep learning-based algal bloom prediction for Lake Okeechobee using multi-source data fusion. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 1-13. *Under review*

Last slide: Contains plain text stating 'End of presentation. Click the right arrow to return to beginning of slide show.'

End of Presentation

Click the right arrow to return to the beginning of the slide show.

For a downloadable version of this presentation, email: I-SENSE@FAU.

Additional Information
The Institute for Sensing and Embedded Network Systems Engineering (I-SENSE) was established in early 2015 to coordinate university-wide activities in the Sensing and Smart Systems pillar of FAU’s Strategic Plan for the Race to Excellence.
Address
Florida Atlantic University
777 Glades Road
Boca Raton, FL 33431
i-sense@fau.edu