CEE Alumni Spotlight: Physics-guided Probabilistic Modeling of Extreme Precipitation Under Climate Change
Earth systems models are crucial to understanding future climate conditions, an important tool for a planet facing climate change. However, current modeling methods often provide competing predictions, and thus make it difficult to discern future changes in extreme precipitation at the local and regional level. Accurate predictions would provide policymakers and industry leaders with important knowledge for protecting their societies and businesses. For example, understanding future extreme precipitation events in a region would be invaluable to deciding where to develop land for agricultural purposes, or the risks of placing a new housing development in a previously low-risk floodplain.
In a newly released paper, published in Scientific Reports, Northeastern civil environmental engineering alumni Drs. Udit Bhatia and Evan Kodra examine new methods for improving earth systems models. “At a global scale, severe precipitation intensifies as the atmosphere warms, manifesting in more destructive and frequent flooding. Risk managers in a wide range of sectors need reliable projections for how extreme precipitation will change at the regional and local scales that are most relevant to them,” said Dr. Kodra. They propose blending existing earth systems models using a physics-guided understanding of the relationship between crucial factors such as temperature and atmospheric moisture capacity. By combining the predictions of various earth systems models, they can receive a more accurate picture of future precipitation events at the local and regional level.
Improved earth systems modeling for extreme precipitation
“One barrier to using climate models is that there are many of them, their projections vary substantially, and it is not clear which of them are trustworthy,” explained Kodra. “This research provides a physics-informed approach for blending multiple climate models and producing credible uncertainty bounds for how severe precipitation will change at regional scales. Those bounds can serve as reliable, probabilistic best- and worst-case outcomes, which are crucial for risk management decisions in sectors ranging from financial services to water resources and infrastructure design.”
“Given the large number of climate model ensembles available from previous generation of Climate Model Inter-comparison Project Five (commonly known as CMIP5) and new model outputs that are being made available through CMIP6, proposed physics informed approach to blend these outputs to produce credible estimates of extremes could aid water resource managers and infrastructure stakeholders in risk assessment and management,” said Bhatia.
About the Authors
Udit Bhatia received his doctorate from Northeastern University and is currently an assistant professor of civil engineering at the Indian Institute of Technology, Gandhinagar. Evan Kodra received his doctorate at Northeastern in 2014 and is currently the CEO of startup risQ. Their doctoral advisor Northeastern Professor Auroop Ganguly, Snigdhansu Chatterjee of University of Minnesota, and Stone Chen of risQ contributed to the report. The majority of the work covered in this paper was completed by Bhatia and Kodra during their doctoral studies in Professor Ganguly’s Sustainability and Data Sciences Laboratory.