Seminar: Lu Tian

How do we know when ecosystems start to experience water stress? The critical soil moisture (SM) threshold (θ_crit), the SM level at which evapotranspiration shifts from being energy-limited to water-limited, is widely used to diagnose water stress across ecosystems.

Traditionally, θ_crit is estimated through empirical local regression of the SM–evaporative fraction relationship at individual sites. However, θ_crit derived in this way depends strongly on the range of conditions sampled at an individual site and may not reflect the underlying ecosystem threshold. This problem is most obvious at the dry and wet ends. In very dry conditions, the energy-limited regime is rarely observed, while in persistently wet conditions, water stress events are too infrequent to constrain the transition, which can potentially lead to artificially high θcrit estimates.

To address this, we use a hybrid physics-informed approach to jointly infer θ_crit across ecosystems using information from diverse hydroclimatic conditions. Do the resulting patterns suggest that θcrit is systematically overestimated in humid regions? Can plant functional traits independently support these findings? I will present our answers and discuss their implications for diagnosing ecosystem water limitation.