When
Where
Available in person and via zoom (see email for link)
Abstract
To quantify transport of water and solutes in soils and the deeper vadose zone, it is essential to know soil hydraulic properties which are relations among water content, hydraulic conductivity, and matric potential. A common lament is that these properties are cumbersome to measure and do not readily exist in reliable high-resolution maps for national and global applications.
To circumvent the experimental difficulties with soil hydraulic properties, soil physicists have developed pedotransfer functions (PTFs) which are statistical methods to estimate soil hydraulic properties from widely available soils data such as soil texture and bulk density. A wide variety of approaches exist, and part of the talk will cover the author’s modest success of using (simple!) AI approaches for PTF development, starting in 1993 and ultimately resulting in the popular Rosetta model. Along the way, the author will identify past challenges with “training” AI PTFs that are presently identified as “hallucinations” and “black-box models”.
In the past decade, several national and global maps have been made that contain PTF input data at increasingly finer resolution (presently up to 30 meters!). In principle, this allows us to also generate global maps of soil hydraulic properties at the same resolution, which would greatly advance near-surface environmental modeling and assessment. The question is, however, how “good” are these maps? Secondary challenges are: how can we improve such maps and distribute the tens of terabytes of data, while also conveying the location-dependent reliability of the estimates?
The work presented is a result of informal collaboration with Professor Yonggen Zhang’s group in Tianjin University, China, and is part of the research of current main campus and online graduate students: Tianyi Huang, Jennifer Swepston, and Jennifer Kirchmeier. Parts of this talk were presented in May 2025, at the Federal University in Rio de Janeiro at the celebration of the 80th birthday of Dr. Rien van Genuchten who inspired and encouraged much of the initial work on soil hydraulic properties and AI PTF development.
Bio
Marcel Schaap hails from the carbonate-rich sandy soils that support the tulip-growing part of the Netherlands. He received his undergraduate and graduate degree from the University of Amsterdam and, in 1996, took a postdoc and later an assistant research professor position with Rien van Genuchten at the USDA-ARS Salinity lab and UC-Riverside. In 2006, he started climbing the tenure ladder to full Professor of Environmental Physics in the department of Environmental Science with a joint appointment in the Department of Hydrology and Atmospheric Science. When not distracted by administrative duties and instruction, his research focuses on the transport of mass and energy in the vadose zone.