Hydrilla Hype: Mapping an Invasive Weed in Two Georgia Reservoirs

Category: Identifying Invasive Species Extent & Critical Species Habitat
Project Team: Southeast U.S. Ecological Forecasting III
Team Location: University of Georgia – Athens, Georgia

Hydrilla distribution prediction in Lake J. Strom Thurmond based on Secchi disk depth, October 2015. Image Credit: Southeast Ecological Forecasting III Team

Hydrilla distribution prediction in Lake J. Strom Thurmond based on Secchi disk depth, October 2015. Image Credit: Southeast Ecological Forecasting III Team

Shuvankar Ghosh
Austin Haney
Frank Braun
Zachary Conner
Christopher Cooper
Abhishek Kumar

Dr. Deepak Mishra (University of Georgia, Department of Geography)
Dr. Susan Wilde (University of Georgia, Warnell School of Forestry and Natural Resources)

Past/Other Contributors:
Caren Remillard (Center Lead)
Wuyang Cai
Pradeep Kumar Ragu Chanthar
Elizabeth Dyer
Peter Hawman
Brandon Hays
Benjamin Page
Linli Zhu


Hydrilla verticillata is an invasive aquatic plant which has rapidly spread through many inland water-bodies across the Southeastern United States, mainly through inadvertent transfer. Once in a water body, this invasive species generally out-competes native aquatic plants and becomes established as the most dominant vegetative species. Consumption of water for drinking, power generation, and recreational use of lakes has been threatened by the spread of Hydrilla. In recent years it was discovered that Hydrilla serves as a host for an epiphytic, toxic cyanobacteria (Aetokthonos hydrillicola) in some water bodies. Aetokthonos hydrillicola is now known to be the causative agent of the neurodegenerative disease avian vacuolar myelinopathy (AVM), which affects waterfowl, raptors, and amphibians. Using Landsat 8 Operational Land Imager (OLI) imagery, a rapid assessment tool was developed to accurately map the extent of Hydrilla on Lake Thurmond (Georgia and South Carolina) and Long Branch reservoir in Henry County, Georgia. This tool will act as the foundation for later models intending to predict future locations in need of Hydrilla management.

Previous story | Main Page | Next story


Daryl Ann Winstead (Mekong River Basin Agriculture) 18-08-2016, 13:07

Very interesting! I see that the team previously mentioned a 30m resolution. Has the team thought about using Sentinel-2 data in the future to possibly capture a higher resolution? Thanks in advance!

Christopher S Cooper 19-08-2016, 11:19

We most definitely considered other platforms and sensors, including Sentinel. NASA DEVELOP projects require the application of a NASA Earth Observation system, therefore Landsat was the most spatially and temporally appropriate. The Sentinel series of satellites are managed by the European Space Agency. We would consider adding Sentinel imagery into the project after the conclusion of the term, to provide a more complete time series.


Daryl Ann Winstead (Mekong River Basin Agriculture) 19-08-2016, 12:19

Awesome! Thank you for your response!

Brian Woodward 16-08-2016, 02:43

Hi team,

Great job! As a former Hydrilla hunter (my first job), I really appreciate this project. We’re conducting a project in the Fall in Ft. Collins looking at secchi depth using NASA EO– I’d love to discuss further with you soon.

All my best,

Christopher S. Cooper 16-08-2016, 08:12


Secchi Disk Depth was a very important variable in the creation of this model. We were able to use SDD to derive values of Percent Light through Water Column (PLW). The PLW values represent our main predictors for hydrilla growth.

I’d love to explain further so please feel free to contact me anytime.

Christopher S. Cooper

Emily Gotschalk 15-08-2016, 14:36

What is the spatial resolution of your model output? Do you think this model could be applied (1) to waterways that are turbid and (2) other species of invasive weeds?

Christopher S. Cooper 15-08-2016, 19:43

The spatial resolution is 30m x 30m, as this model was developed for use with Landsat 8 OLI imagery. The model can be applied to other lakes or waterways, but does not predict high growth in turbid waterways. The reason why it doesn’t tend to predict growth in highly turbid regions of a water body is due to the fact that the model is based on light availability through the water column. The more turbid a water body, the less light is available through the water. Finally, Landsat 8’s spectral resolution prevents the model from differentiating species of submerged aquatic vegetation. We recommend future research use a hyperspectral sensor (such as Hyperion) to possibly accomplish that goal.


Sara Lubkin 14-08-2016, 11:03

Interesting project! Has Hydrilla ever been successfully removed?

Austin Haney 15-08-2016, 18:50

Hi Sara,

Thanks for the kind words. In short, hydrilla has been removed in some areas. It is easiest (although still tough) to remove in small lakes and farm ponds. However, when you get to a scale as large as Lake J. Strom Thurmond, it is nearly impossible. Chemicals used to fight hydrilla spread have highly adverse effects to other species such as fish and native vegetation. They are also extremely expensive and time-consuming to apply to the water body. With our project, we hoped not to eradicate hydrilla, but to highlight areas where spread is most likely so those areas can be targeted for management.



Leave a Reply