Conflicting drivers of land carbon uptake variability reconciled by land-atmosphere coupling

Abstract

<p>Obtaining reliable estimates of the sensitivity of carbon fluxes to water availability, temperature and vapor pressure deficit is essential for constraining climate-carbon feedbacks in Earth system models. However, these variables often co-vary because of soil moisture – atmosphere feedbacks, especially in situations where they are most susceptible to strongly impact ecosystems (e.g. during droughts and heatwaves), leading to potentially conflicting results when sensitivities are assessed independently. In particular, there is conflicting evidence on the role of temperature versus water availability in explaining these variations at the global scale.</p><p>Here, we show that accounting for the effect of soil moisture – atmosphere coupling resolves much of this controversy. Using idealized climate model experiments, we find that variability in soil moisture accounts for 90% of the inter-annual variability in land carbon uptake, mainly through its impact on photosynthesis. Without SM variability, the inter-annual variability (IAV) of land carbon uptake is almost eliminated. We show that the effects of soil moisture can be decomposed into 1) a direct ecosystem response to soil water stress and 2) a dominant indirect response to extreme temperature and vapor pressure deficit triggered by land-atmosphere coupling and controlled by anomalous soil moisture conditions.  Importantly, these two mechanisms do not necessarily have the same spatial extent, and some regions can be more sensitive to indirect effects than to direct effects.</p><p>These two pathways explain why results from coupled climate models suggest a dominant role of soil moisture, while uncoupled simulations diagnose a strong temperature effect. These findings have strong implications for offline model sensitivity analyses as well as field scale manipulation experiments (i.e. rainfall exclusion studies) where the impact of drought on carbon exchange and vegetation activity is often studied by intervening solely on soil moisture content with little consideration of the physical feedbacks on temperature and air humidity occurring in natural conditions.</p>