Climate change is expected to alter the frequency, amplitude and timing of hydrometeorlogical extreme events. In particular heat waves, droughts or cold spells directly affect ecosystem functioning like gross primary production (GPP) or ecosystem respiration (R\({}_{\rm eco}\)). Past studies have adressed this issue based on individual sites or investigating singular large scale events across sites. Here we aim at a first global cross ecosystem analysis to generalize site level findings and to identify drivers and trends in the impacts on the terrestrial carbon cycle.

We detect hydrometeorological extremes at FLUXNET sites using consistently downscaled meteorology over 30 years. We quantify the impact of such extremes on GPP, R\({}_{\rm eco}\) and on the net carbon exchange fluxes (NEE) for univariate (i.e. heat waves) and combined extreme events (e.g. high temperature extremes coocurring with water scarcity). In particular we investigate how these impacts differs according to the events duration or timing within the GS. In addition we investigate differing impacts caused by differences in the affected ecosystem like ecosystem type or average climate conditions.

Globally, we find that heat extremes alone slightly increase R\({}_{\rm eco}\) but no impact on GPP, resulting in slight reduction of the carbon sink capacity. Co-occurring heat and water scarcity generally leads to strong reductions of GPP, and weaker reduction of R\({}_{\rm eco}\) and as a consequence decreases the net carbon sink potential of the ecosystems during the growing season. However, for both types of extremes (i.e. heat waves and heat waves plus drought) we find that these impacts systematically increase with duration at all sites confirming earlier findings that the dose is a critical aspect to consider.

However, the spread of impacts between sites is large and similar types of extreme events are partly causing opposing (i.e. positive vs. negative) impacts on the same processes. Plant functional types (PFTs) only explain parts of these differences. Future research mus elucidate if other, so far unconsidered factors such as functional diversity may explain the differences in the impacts - especially for events of short duration.