Spatial patterns and recent temporal trends in global transpiration modelled using eco-evolutionary optimality

Transpiration from vegetation accounts for about two thirds of land evapotranspiration (ET), and exerts important effects on of global water, energy, and carbon cycles. Resistance-based ET partitioning models using remote sensing data are one of the main methods to estimate global land transpiration, overcoming the limitation by the sparse distribution and short observation periods of site-level measurements. However, the uncertainties of estimated transpiration for these models mainly come from the resistance parameterization based on specific empirical parameters across different plant functional types (PFT). A model based on eco-evolutionary optimization (P model) has recently been proposed to simulate stomatal conductance without the need of calibrated parameters. Here, we calculated global long-term (1982-2018) monthly transpiration with the Penman-Monteith (PM) equation using canopy conductance estimated by the P model (PM-P) and Ball-Berry-Leuning model (PMBBL). Using the observations of SAPFLUXNET and FLUXNET sites as reference, the performance of PM-P was comparable with that of PM-BBL and Global Land Evaporation Amsterdam model (GLEAM). Multi-year mean and trends in growing season transpiration estimated by GLEAM and the PM-P model revealed a similar spatial distribution globally. Both GLEAM and the PM-P model showed a widespread increasing trend of growing season transpiration over 72.06%similar to 80.38% of global land, especially for some main greening hotspots with >3.0 mm/ year. The good performance of the P model indicated that it could avoid the uncertainties emerging from the resistance parameterization with too many empirical parameters and had the potential to accurately estimate global transpiration.