Erica L. McCormick

Contrary to the long-standing assumption that plants access water only in soils, we found that woody plants across the continental U.S. commonly use rock moisture: water stored in the cracks and fissures of weathered bedrock. Because direct measurements of bedrock water are rare, we used plants as sensors, inferring bedrock water use from remotely sensed water balances.

See the project page (opens in a new tab) for maps, data, and code.

Vegetation response to drought is studied far more than response to extreme storms, even though their global impacts may be comparable. Using machine learning with data from 54 FLUXNET sites, I quantified changes in photosynthesis after extreme storms and found that the response is explained better by soil texture, climate, topography, and vegetation density than by storm intensity.

Code and data (opens in a new tab) on Zenodo.

Nearly all evapotranspiration (ET) datasets build in assumptions about soil moisture and plant behavior, which makes them hard to use for learning about the subsurface. I implemented a remotely sensed ET dataset based on the surface flux equilibrium approach, which uses only atmospheric temperature and humidity, and showed with triple collocation that it performs comparably to commonly used ET products.

The daily ET dataset (opens in a new tab) is on Zenodo.