Modeling Evapotranspiration at Global Scale

Evapotranspiration (ET) modeling is central to resolve water and energy balances and linking the terrestrial water and carbon cycles. An important challenge remains how to partition the ET flux into transpiration (T) and soil evaporation (E). Estimates from literature and predictions based on land surface models range between 10-90 % depending on model assumptions. ET and its partitioning depend on the soil hydraulic properties shaping the redistribution of precipitation and the adaption of plants to transport water from the roots to the stomata. To improve predictions of ET at global scale, we invest on three elements as shown in Figure 1: (a) modeling ET as function of critical water content in the surface layer and the root zone, (b) applying the most recent global maps of soil hydraulic properties, and (c) improving maps of root length distribution.

Evapotranspiration (ET) at global scale — Figure: Basic elements of research on evapotranspiration (ET) at global scale. (a) Root water extraction and soil evaporation are both limited at a critical water content, which can be deduced from soil (and plant) hydraulic properties and is implemented in a modified soil capacitor model. (b) Various maps on soil hydraulic properties are based on rules (pedotransfer functions) deduced from samples collected from arable lands, neglecting the variability of soil formation processes. To avoid these shortcomings, we implement recent maps of Gupta et al. (2021) that uses data collected from a wide range of climates and land use and considers effects of environmental covariates on soil properties. (c) A new model for maximum root depth was built by linking measured root depths values with environmental conditions using machine learning (random forest).

ET for each pixel of the global map is simulated by a two-layered ‘soil capacitor’ consisting of a surface layer and the root zone. For that purpose, we expand the surface evaporation capacitor model of Or and Lehmann (2019) by considering concurrent root water uptake. The original capacitor model simulates surface evaporation (focusing on stage-1) and internal redistribution following rainfall events. The thickness of the evaporation-active soil layer is defined by an intrinsic soil property termed the evaporation characteristic length (deduced from soil water characteristics and hydraulic conductivity functions). The modified model considers water extraction by plant roots (Carminati and Javaux 2020) from the capacitor layer as well as water redistributed into deeper layers (sheltered from soil evaporation but accessible for root water uptake). Depending on the amount of water leaking below the capacitor depth, vegetation can take up this natural storage at rates limited by the hydraulic properties of the rhizosphere.

Contact: Peter Lehmann (), Andrea Carminati (), Fabian Wankmüller ()