Hurricane Tracking Using Multi-GNSS-R and Deep Learning

Abstract

Hurricane Monitoring using GNSS-R is an emerging application in earth remote sensing. Practically, the hurricane is categorized by its maximum win speed which can be measured from the reflected GNSS signals. In order to track the hurricane movement efficiently, a large number of measurements is required. Alternatively, by using deep learning algorithms, useful information from much fewer measurements can be inferred. In this paper, a deep learning-based technique is proposed to track the core of a moving hurricane using limited measurements of a reflected GNSS signals. The proposed technique has achieved a very high accuracy of 96.6% using a CNN model.