Improving Spaceborne GNSS-R Algal Bloom Detection with Meteorological Data

Abstract

Algal bloom has become a serious environmental problem caused by the overgrowth of plankton in many waterbodies, and effective remote sensing methods for monitoring it are urgently needed. Global navigation satellite system-reflectometry (GNSS-R) has been developed rapidly in recent years, which offers a new perspective on algal bloom detection. When algal bloom emerges, the water surface will turn smoother, which can be detected by GNSS-R. In addition, meteorological parameters, such as temperature, wind speed and solar radiation, are generally regarded as the key factors in the formation of algal bloom. In this article, a new algal bloom detection method aided by machine learning and auxiliary meteorological data is established. This work employs the Cyclone GNSS (CYGNSS) data and the fifth generation European Reanalysis (ERA-5) data with the application of the random under sampling boost (RUSBoost) algorithm. Experiments were carried out for Taihu Lake, China, over the period of August 2018 to May 2022. During the evaluation stage, the test true positive rate (TPR) of 81.9%, true negative rate (TNR) of 82.9%, overall accuracy (OA) of 82.9% and the area under (receiver operating characteristic) curve (AUC) of 0.88 were achieved, with all the GNSS-R observables and meteorological factors being involved. Meanwhile, the contribution of each meteorological factor and the error sources were assessed, and the results indicate that temperature and solar radiation play a prominent role among other meteorological factors in this research. This work demonstrates the capability of CYGNSS as an effective tool for algal bloom detection and the inclusion of meteorological data for further enhanced performance.