Towards a consistent quantification of ecosystem transpiration and its uncertainty from the SAPFLUXNET database

Transpiration from forests and woodlands is the main component of terrestrial evapotranspiration. Ecosystem-scale transpiration estimates are needed to inform models and remote sensing products so that they can improve their quantification of the magnitudes, spatiotemporal patterns, and environmental sensitivity of transpiration at regional to global scales. Tree-level sap flow measurements can be used to estimate ecosystem transpiration in forests and woodlands and these data are now globally available in the SAPFLUXNET database (Poyatos et al. 2020). However, observational errors, sampling assumptions, and missing data propagate uncertainties in the upscaling process to the ecosystem level. Here we quantify ecosystem transpiration and its uncertainty, from hourly to annual scales, across SAPFLUXNET sites using two different approaches. In a first approach, we estimated hourly sap flow per unit basal area at the species level, which was then aggregated to the stand level using species-specific basal areas available in SAPFLUXNET metadata. In this approach, uncertainty was quantified from the observed variability in tree-level sap flow within a species. In a second approach, we used empirical relationships between tree diameter and sap flow to obtain stand-level transpiration and propagated the uncertainty in this relationship to the stand-level values. For both approaches, sap flow data obtained with uncalibrated heat dissipation probes were also adjusted using a recently published correction based on sap flow calibrations. The different upscaling methods, implemented in R code, will allow reproducible upscaling and uncertainty quantification from SAPFLUXNET datasets, paving the way towards a better understanding of ecosystem transpiration and its controls across the globe.

Poyatos, R., Granda, V., Flo, V., […], Steppe, K., Mencuccini, M., Martínez-Vilalta, J. (2020). Global transpiration data from sap flow measurements: the SAPFLUXNET database, Earth System Science Data Discussions, 1–57, .