Description of Data Sets:
In this project, we utilized NASA’s Landsat 8 satellite with the Operational Land Imager (OLI) and Thermal Infrared (TIRS) sensor, and the European Space Agency’s (ESA) Sentinel-2 satellite with the MultiSpectral Instrument (MSI) sensor to improve an algal event detection system for the Highland Lakes chain in Austin, Texas. Earth observation (EO) data were coupled with in-situ sampling data provided by the City of Austin Department of Watershed Protection and the Lower Colorado River Authority. The near-surface concentration of chlorophyll-a (chl-a) was retrieved from OLI and MSI, through a machine learning approach. We also achieved detection of cyanobacteria mats by implementing a modified cyanobacteria index, the Broad Wavelength Algae Index (BWAI), utilizing OLI and MSI. The water surface temperature was retrieved from OLI’s TIRS bands. Turbidity levels were also estimated from MSI and OLI, based on the widely used Normalized Turbidity Index (NDTI) . Both OLI and MSI data were obtained from the Google Earth Engine Data Catalog, which freely hosts petabytes of EO data ready for processing in the cloud. The chosen image collections in Earth Engine were already atmospherically corrected and pre-processed to surface reflectance to work within the capabilities of the Google Earth Engine API. Lastly, the in situ chl-a and water surface temperature were compared with remote sensing derived products, for validation and uncertainty quantification. The EO-derived products enable near real-time monitoring of environmental proxies relevant to algal events, allowing end-users to move beyond traditional monitoring methods, which has focused on data collection at single sites.
Scientific Potential of Presentation:
Beginning in 2019, harmful algal events have caused canine deaths in both Lady Bird Lake and Lake Travis located near Austin, Texas. These two reservoirs are part of the larger Highland Lakes chain, managed by the City of Austin Department of Watershed Protection and the Lower Colorado River Authority, which fulfill municipal, commercial, and agricultural water demands. Given the recent increase in favorable environmental conditions for algal events in Central Texas, NASA DEVELOP partnered with these organizations to improve their monitoring and detection of algal events, through the application of EO data and machine learning.
We developed a user-friendly toolkit, named the Lake Algal Monitoring Dashboard (LAMDA), that enables near real-time monitoring of environmental proxies relevant to algal event presence in the Highland Lakes chain, and will ultimately support water management, decision making, and risk communication. Environmental proxies in this tool include spatially and temporally varied chlorophyll-a (chl-a) concentrations, cyanobacteria detections, turbidity, and water surface temperature. The Landsat 8 Operational Land Imager (OLI) platform and the Sentinel-2 MultiSpectral Instrument (MSI) were used with a combined revisit time of up to ~3 days and < 30 m per pixel products. Chl-a concentrations were estimated using a novel data-driven approach, Mixture Density Network (MDN), which is a class of neural networks that estimates multimodal Gaussian distributions over a range of solutions. Recent publications suggested MDNs outperform most state-of-the-art benchmarks for chl-a estimation. Cyanobacteria detection was accomplished using the Broad Wavelength Algae Index (BWAI), which can differentiate algal blooms from algal proliferations (mats).
The City of Austin Department of Watershed Protection and the Lower Colorado River Authority have been collecting in situ water samples and routinely monitoring lake conditions for decades, but limitations such as cost and lab delays result in insufficient monitoring coverage from field sampling alone. Therefore, the availability and processing ease of continually updating remote sensing datasets allow end users to move beyond traditional field-based methods for water quality monitoring, which has previously been focused on data collection typically at a limited number of sites and bi-monthly intervals. Second, the use of machine learning has provided confident chlorophyll-a concentrations, which will help the decision makers better understand which sites are chlorophyll-a hotspots and help them decide further sampling efforts. Third, benthic algae proliferations have been recently found to produce neurotoxins and the production of these cyanotoxins has been concomitant with the observed increase of benthic algae proliferations in multiple Highland Lakes since 2019. Cyanobacteria detections enable the detectable differentiations between phytoplankton and benthic algae proliferations, which are only detectable through remote sensing technology. Furthermore, observing turbidity and temperature patterns and/or trends through spectral indices will help the partners understand the hydraulic system as well as observe any environmental connections. Overall, our project enables the partners in Austin, Texas, to implement EO products into their current decision-making framework regarding lake management and protections, which would save costs from lab water testing and provide timely warnings.
Interdisciplinary Team Submission
Current Insitute of Study/Organization: Pennsylvania State University
Currently Pursuing: Doctoral
- Grand Prize