This is the Harmful Algal Bloom (HAB) Prediction System for Charlotte County and surrounding water using the data collected from MODIS (Moderate Resolution Imaging Spectroradiometer). The satellite data, acquired daily, is downloaded and processed in our Linux data server. The location and the cell count of Karenia brevis for the time span ranging from January 2010 to July 2017 were obtained from Fish and Wildlife Research Institute (FWRI) harmful algal bloom (HAB) database hosted by Florida Fish and Wildlife Conservation Comission. Those data were gathered in response to the reported incidents and depended on projects as well as funding. MODIS archival data were processed for 2010 to 2017 to correspond with historical HAB incidents.

The historical occurrences were used as the response variables for all the temporal variables generated from satellite data and spatial variables collected from other sources. The variables we employed are: chlorophyll a (OC3M), chlorophyll a Generalized Inherent Optical Property (GIOP), chlorophyll a Garver-Siegel- Maritorena (GSM), sea surface temperature, normalized fluorescence line height, euphotic depth, diffuse attenuation coefficient, wind direction, wind speed, backscatter coefficient at 547 µm, distance from the mouth of the stream, and bathymetry. The cell count corresponding to 509 algal bloom events that occurred in different parts of the bay from 2010 to 2017 were used to generate the multivariate regression prediction model. The model was evaluated using an additional 127 reported HAB occurrences throughout the investigated period; the model was found to be successful at 90%, 6% were found to be false positives, and 4% of the reported HAB events were not detected by the model.

The Algal Bloom Prediction System for Charlotte County involves downloading of data, processing of derived variables and application of prediction model to predict the occurrences of bloom events for each consecutive day and finally the publication of the result to the web using ArcGIS server on daily basis. The MODIS data for every day is acquired at ~3 pm, is made available for download on NASA’s website at ~4 pm, and is processed for HAB occurrences and published on our website at ~7 pm. Currently, we are developing models that can predict HAB events several days in advance and in the near future, we will include nutrient measurements from field observation and remote sensing datasets from Unmanned Aerial Vehicles (UAVs). These new techniques will generate continuous temporal records for the investigated variables and will enhance model predictability given the minimal atmospheric contributions.