A risk-benefit model to simulate vegetation spring onset in response to multi-decadal climate variability: Theoretical basis and applications from the field to the Northern Hemisphere

- A risk-benefit model is developed for predicting vegetation spring onsets.
- The two-leaf model accurately predicts photosynthesis and evaportranspiration.
- The growing production-day model accurately predicts vegetation spring onsets.
- The model captures effects of multi-decadal climate variability on spring onsets.

Vegetation spring onset regulates canopy photosynthetic activities and subsequent ecosystem processes, thereby influencing the complex interactions between the biosphere and the atmosphere. Robust models that predict the timing of vegetation spring onsets are required to account for the ecosystem response and adaption to climate variability. Here, a risk-benefit model is proposed to account for the fundamental tradeoff underlying plant leafing strategies: earlier timing of leaf-out events leads to greater vegetative carbon gain but higher risks associated with hazard damages. The proposed model named the Growing Production-Day (GPD) model uses the cumulative productivity of a hypothetical reference vegetation cover as the overall benefit and predicts the events of vegetation spring onset when a certain threshold that vegetation invests to mitigate potential hazard damages is reached. The daily canopy photosynthesis of the hypothetical reference vegetation cover is simulated by a two-leaf canopy model, which considers sunlit and shaded leaves within a canopy separately and accounts for the biogeochemical processes of canopy radiative transfer, leaf photosynthesis, leaf conductance, leaf transpiration, and soil evaporation. When validated against measurements from available flux tower sites of deciduous broadleaf forests, the two-leaf canopy model accurately simulated daily canopy photosynthesis and evapotranspiration rates, indicated by significant correlations (R2 = 0.787 and 0.745 for gross primary production and latent heat, respectively) and low root-mean-square errors (RMSE = 2.25 gC m2 day−1 for gross primary production and 21.53 W m−2 for latent heat, respectively) between the observed and modeled values. Based on the two-leaf canopy model, the GPD model predicted the dates of spring onsets accurately for three studied biomes (RMSE = 9.10, 5.54, and 12.76 days for evergreen needleleaf forests, deciduous broadleaf forests, and grasslands, respectively) as derived from the flux tower data. In addition, the GPD model could simulate the long-term interannual variation of species-level leaf onset dates as obtained from in-situ observations, and capture the spatiotemporal patterns of multi-decadal variation of vegetation spring onsets across the Northern Hemisphere as derived from satellite data. Although the GPD model requires further refinements, it shows promises with respect to simulating vegetation spring onset in response to multi-decadal climate variability.