CEE Seminar: Modern Precipitation Data and its Applications: Errors, Insights and Flood Frequency Analysis in a Changing World

Abstract: Robert Horton’s “Rule of Hydrologic Data” (EOS, 1931): “[Hydrologic analysis] is nearly always determined by the nature and extent of the data available. The best possible use should be made of all the available data... tempered according to the necessities of economy of labor.” Nonstationarity, and the wealth of new hydrologically relevant data, suggest that we can and should revisit Horton’s rule.Wright first shows recent work in quantifying and reducing errors in precipitation remote-sensing data using a modeling framework based on censored shifted Gamma distributions, an alternative formulation to the conventional Gamma distribution that can describe both precipitation occurrence and magnitude. He demonstrates that merging satellite- and numerically-based precipitation estimates using this framework can leverage the complementarity in these two data sources to reduce random errors beyond what can be achieved using standard error modeling. This merging can also be used to create probabilistic estimates of precipitation that are more accurate than either data source individually.

Wright also argues that modern rainfall data, whether from radar, satellites, numerical models or rain gage networks, give us new ways of exploring important and often overlooked spatiotemporal aspects of rainfall extremes and their impacts. Despite the obvious role that rainfall (and its complex variability in space and time) plays in floods, most flood frequency analyses do not make use of rainfall information at all. Trends in rainfall and flooding due to climate or land use change place further limits on existing techniques. Wright shows how we can couple modern rainfall data with a probabilistic framework known as stochastic storm transposition (SST) to perform rainfall and flood frequency analysis at multiple scales. SST “lengthens” the rainfall record by resampling observed storms and extracting space-time information from rainfall data. He has codified the approach in RainyDay©, a platform for quickly generating large numbers of realistic probabilistic extreme rainfall scenarios, and highlights some advantages of SST and the implications for understanding the nature of flood risk in a changing environment.

Bio: Wright holds bachelor's and master's degrees in civil and environmental engineering from the University of Michigan with a focus in hydrology and hydraulics. He served as a regional sanitation engineer with the Peace Corps in Bolivia from 2006-2008 and worked as a consulting hydropower engineer in Chile from 2008-2009 before earning his Ph.D. in environmental engineering and water resources from Princeton University, where he studied urban rainfall and flood hydrometeorology. He worked as a disaster risk management consultant at the World Bank from 2013-2014, focusing on flood and landslide risk management in Latin America and the Caribbean. Before joining the CEE faculty at the University of Wisconsin-Madison in 2016, Wright was a NASA postdoctoral program fellow in the Hydrological Sciences Lab at Goddard Space Flight Center in Greenbelt, Maryland. His research interests include hydrometeorology and hydroclimatology, the role of rainfall space-time variability in floods and other hydrologic and environmental phenomena, and applications of satellite and ground-based remote sensing.