Using Satellite Imagery for Red Tide Detection
A comparative analysis for Tampa Bay, Florida
Tampa Bay
Tampa Bay is a 400-square-mile natural bay located on the west coast of Florida that is connected to the Gulf of Mexico. Most of the bay is fairly shallow (average depth of 12 feet) and serves as estuarine habitat for a large variety of fish and wildlife.
The area is home to more than 4 million people and has a military base, an international airport, a port, and numerous industrial facilities (Wikipedia). The area also serves as a tourist destination for nearly 15 million people annually (WUSF News), and they often come seeking warm water and beautiful beaches.
What is Red Tide?
Harmful algal blooms (HABs) are an annual occurrence along the Gulf Coast of Florida. Karenia brevis, the species of algae that creates these events, known as red tide, discolors the coastal waters creating a reddish tint caused by light absorption in the K. brevis cells (Hu et al, 2022). Historically, they were most often observed during summer and fall (Carvalho et al, 2011) yet blooms are not limited to occurring during those time periods (NOAA, 2016).
While it is often thought that increasing nutrient loads from urban development entering Tampa Bay is contributing to increased occurrences of red tide, Florida's red tide events originate 10-40 miles offshore. However, the nutrients may serve as "food" for K. brevis during HABs promoting further growth during an event (Mote, 2023). Red tide impacts adversely effect marine life, human health, and local tourism. (Karki, 2018).
Detecting Red Tide
Red tide creates a brevotoxin that is both present in the water and in aerosols from wave action. This brevotoxin is harmful to humans and marine life (Pierce et al, 2005). Concentrations as low as 5,000 cells/L cause the closure of shellfish beds. Once concentrations reach 250,000 cells per liter, fish kills are common. More than 100 locations throughout the state are monitored via in situ water samples on a weekly, twice-monthly, or monthly rate in order to detect HABs (FFWCC, 2023).
When K. brevis concentrations reach 50,000 cells per liter, chlorophyll, a metric that can be calculated from satellite images (Hu et al, 2022), can be used to detect red tide events. However, chlorophyll production is not limited to K. brevis (Stumpf et al, 2003), therefore this metric cannot be used alone to detect red tide event. It should be used only in conjunction with in situ testing to confirm or dismiss the satellite results.
Satellite Imagery
The goal of this project was not only to assess the ability of satellite images for detecting red tide events but also to compare different satellite imagery for such purposes. Imagery taken October 2021, December 2022, and March 2023 was used. These were months with documented red tide in Tampa Bay per the Florida Fish and Wildlife Conservation Commission (FFWCC) (2023). While other dates of imagery were available, many proved to be too cloudy to be useful for this project.
A Landsat 8 image was used for October 2021, while Landsat 9 images were used for December 2022 and March 2023. Landsat images were accessed through the U.S. Geologic Survey's data website, Earth Explorer (USGS).
Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua images were used for the same time periods as the Landsat imagery. The images were accessed through the Google Earth Engine Code Editor platform (GEE, 2023).
Methods
The image to the left shows chlorophyll concentration (CI) from March 6, 2023 as calculated manually in QGIS using the raster calculator to apply the formula below to the individual satellite bands. This calculation had to be performed for Landsat imagery, but not MODIS.
CI was determined using the following formula:
CI=Rrs(λgreen)−[Rrs(λblue)+(λgreen−λblue) / (λred−λblue) ∗ (Rrs(λred)−Rrs(λblue))] (NASA, 2023)
The chlorophyll-a (the main form of chlorophyll found in algae) had already been calculated for the MODIS images so it did not have to be calculated manually. Code was used to display the chlorophyll-a concentrations directly from the MODIS imagery via Google Earth Engine (GEE, 2023). Only CI was calculated for the Landsat images, not chlorophyll-a due to concerns about the validity of the chlorophyll-a algorithm to these Landsat datasets.
Analysis Results
The images to the left show the MODIS chlorophyll-a levels compared to the Landsat CI, where red indicates higher levels while blue lower levels. While both were successful in showing higher chlorophyll levels indicating potential red tide during documented events, their results vary significantly.
The first significant difference noted is the very course spatial resolution, 1km, of the MODIS imagery compared to the Landsat imagery at 30m spatial resolution (Sabins, 2020). While 1km resolution makes sense for analyzing very large study areas, such as the entire state of Florida, it does not provide the level of detail needed to assess red tide occurrence at a more local scale.
Second, the MODIS imagery has a return interval of 1-2 days, while Landsat has a return interval of 16 days (Sabins, 2020). This difference in return intervals makes comparison between the two more difficult since the images may not occur on the same day. Also, with a lower temporal resolution, Landsat imagery becomes less useful if trying to monitor red tide relative to in situ sampling and for providing the public with up-to-date information. Therefore, Landsat would be more appropriate for studying past events, while MODIS could be used for monitoring current events. Furthermore, neither satellite can penetrate clouds so their presence can further impact the usefulness of the imagery (NSICD, 2023; Sabins, 2020).
Finally, while both imagery sources provided evidence of higher chlorophyll levels, indicating the possibility of red tide, they show different magnitudes and spatial extents of higher chlorophyll. MODIS is showing higher levels within the center and eastern part of the bay, while Landsat shows high levels along the coastline and barrier islands.
Further Research
While satellite imagery can provide useful information when looking to study red tide events, the issues of varying spatial and temporal resolutions, along with cloud cover, can make near real-time assessments difficult without in situ measurements to verify imagery findings.
Had more time been allocated to the project, the use of other satellite imagery, such as Sentinel-2 data could have been examined as well. Chlorophyll-a could have been calculated for the Landsat imagery, hopefully making the variation shown between Landsat and MODIS chlorophyll levels more directly comparable.
Finally, historical analysis of red tide events in conjunction with sea surface temperature measurements could have been investigated to see if a correlation exists between rising temperatures and red tide occurrence, intensity, and duration.