On the seasonal relation of sun-induced chlorophyll fluorescence and transpiration in a temperate mixed forest

Novel strategies are required to reduce uncertainties in the assessment of ecosystem transpiration (T). A major problem in modelling T is related to the complexity of constraining canopy stomatal resistance (rsc), accounting for the main biological controls on T besides non biological controls such as aerodynamic resistances or energy constraints. The novel Earth observation signal sun-induced chlorophyll fluorescence (SIF) is the most direct measure of plant photosynthesis and offers new pathways to advance estimates of T. The potential of using SIF to study ecosystem T either empirically or in combination with complex mechanistic models has already been demonstrated in recent studies. The diversity of environmental drivers determining diurnal and seasonal dynamics in T and SIF independently requires additional investigation to guide further developments towards robust SIF-informed T retrievals. This study consequently aims to identify relevant biotic and abiotic environmental drivers affecting the capability of SIF to inform estimates of ecosystem T. We used observational data from a temperate mixed forest during the leaf-on period and a Penman-Monteith (PM) based modelling framework, and observed varying sensitivities of SIF-informed approaches for diurnal and seasonal T dynamics (i.e. r2 from 0.52 to 0.58 and rRMSD from 17 to 19%). We used the PM based modelling framework to investigate systematically the sensitivity of SIF to diurnal and seasonal variations in rsc when empirically and mechanistically embedded in the models. We used observations and the Soil-Vegetation-Atmosphere-Transfer model SCOPE to study the dependence of SIF and T on abiotic and biotic environmental drivers including net radiation, air temperature, relative humidity, wind speed, and leaf area index. We conclude on the potential of SIF to advance estimates of T and suggest preferring more sophisticated modelling frameworks constrained with SIF and other Earth observation data over the single use of SIF to assess reliably ecosystem T across scales.