Remote sensing of harmful algal blooms: Data modeling and business aspects

Abstract

The present work accomplished a spherical investigation of remote sensing techniques for harmful algal bloom (HAB) detection. The areas covered by the research are: (1) environmental science aspects of the problem, (2) devising a pipeline and a model for HAB prediction based on artificial neural networks (ANN) from acquired satellite imagery, (3) a high-level design for the system capable of integrating data from various types of sensors and (4) a preliminary business analysis for a HAB detection service proposed herein.

The modeling activity resulted in a robust pre-processing pipeline able to transform acquired satellite products (set of images) to feature vectors for model training purposes. The feature vector components were carefully selected according to literature review findings. The resulting ANN model showed indications of overfitting. Further analysis of the dataset though revealed the limitation of the current research owed to the fact that no in-situ sensor data for HAB detection were available.

The service was conceptualized as an integrator of HAB related environmental data collected from various type of sensors. Issues concerning the technical requirements of the service were analyzed with the scope of enabling future system architecture activities. The combination of satellite, drone and sensor in-situ data streams led this analysis to the selection of a micro-services based architecture with well identified core application components.

The businesses analysis for the proposed HAB service yielded a well-defined target market (i.e. governmental environment agencies). Solid hypotheses for the customer profile are formed and based on that, a viable business model is framed. A cost analysis investigation resulted in a very competitive marginal operating cost that amounts to $0.42 per km2 for HAB monitoring using the proposed service. This cost figure indicates that the service can be marketed at very competitive prices allowing for high profit margins and economies of scale.