CEO Geothara and Cornell Tech, Cornell University Los Angeles, California
Engineers carefully analyze hydrologic data before constructing dams that regulate rivers, the arteries that sustain terrestrial life. Non-stationarity detection, commonly known as trend detection, is an indispensable toolkit for detecting climate change risks and help with safe dam design. Yet, it remains one of the least understood technical concepts in the engineers’ vocabulary, leaving us quite unprepared to plan for climate change impacts on water systems. It is not an understatement to say that we still use a machete today (descriptive statistics and outdated non-stationarity methods) in practice, when we need a surgical knife. In this session, I will present a solution to this problem: a set of algorithms and data-based tools made for robust trend detection, based on my recent Ph.D. dissertation work (UCLA, 2022), where I enumerated 45 statistical concepts, 80 methods, and 7 technical trade-offs (chicken-and-egg problems) associated with picking the right non-stationarity analysis method. In this technical session, I offer an intro presentation of theory and necessary practical tools available for robust trend detection, followed by a visual animated demonstration of the tools to intuitively appreciate the importance of learning this complex, important subject.
