Contents 1. Introduction
Satellite remote sensing data is essential in advancing research and applications in Earth science domain. Data from the Earth observing platforms are produced at rates that challenge the scalability and efficiency of search systems. Search based on traditional faceted filtering falls short of addressing advanced use cases of searching data based on Earth science phenomena. An Earth science phenomenon is an observable occurrence of particular physical significance within Earth's dynamics, for example, a hurricane. An Earth science event is an instance of a phenomenon, for example, hurricane Katrina. Currently, systems such as the Earth Observatory Natural Event Tracker (EONET) [1] employ manual curation of Earth science events. Such an approach is time consuming and not scalable. Automated event detection can provide a scalable solution to augment current approaches. One way to allow Earth science phenomena-based search systems is to use machine learning (ML) techniques. ML is the field of study that involves development of algorithms to teach computers to learn from data. Automated detection of Earth science events is possible using data driven approach enabled by ML techniques, specifically, deep learning techniques. Deep learning consists of machine learning algorithms with multiple layers, where each layer progressively learns features. Advancements in deep learning techniques have produced state-of-the-art image classification results, mostly using convolutional neural networks (CNNs) [2]. However, very few of these results have been transitioned to production applications using ML lifecycle [3]. This is mainly because the scientists who develop the ML models are rarely equipped to deploy and scale the models in production. Towards that end, a phenomena portal is developed as a production system that utilizes deep learning-based real time Earth science event detection with search and analysis interface to explore cataloged Earth science events and complementary information.
