Abstract
This article describes the application of machine learning techniques to develop state-of-the-art detection and prediction system for spatiotemporal events found within remote sensing data; specifically, harmful algal bloom (HAB) events. We propose HAB detection system based on a ground truth historical record of HAB events, a novel spatiotemporal datacube representation of each event (from MODIS and GEBCO bathymetry data), and a variety of machine learning architectures utilizing the state-of-the-art spatial and temporal analysis methods based on convolutional neural networks, long short-term memory components together with random forest, and support vector machine classification methods. This work has focused specifically on the case study of the detection of Karenia brevis algae ( K. brevis ) HAB events within the coastal waters of Florida (over 2850 events from 2003 to 2018; an order of magnitude larger than any previous machine learning detection study into HAB events). The development of multimodal spatiotemporal datacube data structures and associated novel machine learning methods give a unique architecture for the automatic detection of environmental events. Specifically, when applied to the detection of HAB events, it gives a maximum detection accuracy of 91% and a Kappa coefficient of 0.81 for the Florida data considered. A HAB forecast system was also developed where a temporal subset of each datacube was used to predict the presence of a HAB in the future. This system was not significantly less accurate than the detection system being able to predict with 86% accuracy up to 8 d in the future.
Highlights
A LGAL blooms are defined as high concentrations of phytoplankton
Our work describes the definition of a unique datacube data structure for supervised machine learning of spatiotemporal events; harmful algal bloom (HAB) events together with a novel machine learning architecture to provide optimal HAB event classification and prediction performance
Given that the temporal range investigated within the datacubes is small, we have investigated flattening the time series sequences and utilized simple high performance non-network classifiers as a last stage: Random forests (RF) [43]; support vector machine (SVM) [44]; and nontemporal, fully connected networks [multilayer perceptrons (MLPs)]
Summary
A LGAL blooms are defined as high concentrations of phytoplankton (algae) (for example, >50000 cells/L [5]). Harmful algal blooms (HABs) are problematic algal blooms causing toxicity and associated environmental impacts. Often termed “Red Tides,” HABs have been a significant worldwide research topic over three decades [1]–[7]. They continue to be of major concern, due to their considerable environmental and societal impact and a recent significant increase in frequency reported around the world [2]. Manuscript received April 16, 2020; revised May 16, 2020; accepted June 1, 2020. Date of publication June 10, 2020; date of current version June 23, 2020.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
