According to research presented this month at the American Geophysical Union annual meeting, combining satellite technology with machine learning may help scientists to better track and prepare for climate-related natural calamities.
Rising global temperatures over the past few decades have led to an increase in the intensity and frequency of many natural occurrences, including hurricanes, snowstorms, floods, and wildfires.
Although these catastrophes cannot be stopped by humans, the rapidly expanding number of satellites that orbit the Earth from space provides a greater opportunity to track their development, according to C.K. Shum, co-author of the study and a professor at The Ohio State University’s Byrd Polar Research Center. According to him, empowering locals to make knowledgeable decisions may increase the efficiency of local catastrophe response and management.
According to Shum, “predicting the future is a relatively challenging problem, but by utilising remote sensing and machine learning, our research seeks to help establish a system that will be able to monitor these climate-induced dangers in a way that would enable a timely and informed catastrophe response.”
Geodesy, the science of measuring a planet’s size, shape, and orientation in space, is used in Shum’s research to examine phenomena connected to anthropogenic climate change.
Researchers tested whether a combination of remote sensing and deep machine learning analytics could accurately monitor abrupt weather phenomena, including floods, droughts, and storm surges in specific regions of the world using geodetic data obtained from multiple space agency satellites.
These techniques were employed in one experiment to test whether radar signals from the Global Navigation Satellite System (GNSS) of Earth, which were reflected over the ocean and received by GNSS receivers at towns offshore in the Gulf of Mexico, could be used to track hurricane evolution by gauging rising sea levels after landfall. The group looked at how seven hurricanes, including Hurricanes Hana and Delta, altered coastal sea levels between 2020 and 2021 before they made landfall in the Gulf of Mexico. They observed these intricate variations and discovered a link between increased sea levels and stronger storm surges.
The Gravity Recovery and Climate Experiment (GRACE) mission and its follow-on, GRACE Follow-On, were launched by NASA and the German Aerospace Center. Over the past 20 years, both satellites have been used to track changes in Earth’s mass, but they have only been able to see the planet from a height of about 400 miles. However, Shum’s team was able to lower this resolution to roughly 15 miles using deep machine learning analytics, significantly enhancing society’s capacity to monitor natural dangers.
Utilizing deep machine learning requires conditioning the algorithm to continuously learn from different data inputs in order to get the desired result, according to Shum. When two Category 4 Atlantic hurricanes, Hurricane Harvey in August 2017 and Hurricane Laura in August 2020, made landfall over Texas and Louisiana, respectively, satellites were able to quantify the path and evolution of the storm surges they caused.
According to Shum, accurate measurements of these natural disasters may one day help with hurricane forecasting. Shum would prefer, however, that in the near future, nations and organisations make their satellite data more easily accessible to researchers, as deep machine learning projects frequently require substantial volumes of diverse data to support effective forecasting.
Shum stated that processing enormous amounts of precise data requires time and effort for many of these novel satellite approaches. “If researchers have more resources at their disposal, we may be able to develop technologies that will help people be better prepared to adapt, as well as enable disaster management organisations to improve their response to severe and frequent climate-induced natural catastrophes.”
The project’s co-authors included Kazim Atman of Queen Mary University of London, Orhan Akyilmaz and Metehan Uz of Istanbul Technical University, Yu Zhang, Yuanyuan Jia, Yihang Ding, and Junyi Guo from Ohio State. The National Science Foundation, the National Aeronautics and Space Administration, the Scientific and Technological Research Council of Turkey (TÜBTAK), the United States Agency for International Development (USAID), and the National Science Foundation all provided funding for this research.
