Temporal dynamics of terrestrial carbon dioxide removal

Virtually all future scenarios in the IPCC AR6 keeping climate change well below 2C include carbon dioxide removal (CDR), often leading to large transformations of global land surface and land use. Re-/afforestation (AR) and bioenergy with carbon capture and storage (BECCS) are the two most prominent CDR measures in those scenarios. The temporal evolution of carbon uptake and storage is very different between bioenergy plants, which are annually harvested to (ideally) permanent storage, and forests, which sequester carbon for decades on site but can be affected by disturbances. Additionally, while AR dominates current CDR deployment as tree seedlings and saplings can be planted right away, BECCS requires further processing and storage infrastructure leading to longer establishment time scales. Thus, BECCS covers only a tiny fraction of existing and announced amounts of CDR. Hence, depending on whether CDR is intended to support rapid, deep reductions of net emissions in the near term (as in the Nationally Determined Contributions of parties to the Paris Agreement) or to counterbalance residual emissions or even reach net negative emissions in the longer term, either AR or BECCS could be more effective. This will also vary across world regions. We compare the temporal dynamics of carbon storage efficiency between AR and BECCS with three state-of-the-art terrestrial biosphere models (JSBACH, LPJmL, LPJ-GUESS). We use a global, highly stylized setup where a fixed share per pixel of current agricultural land is replaced by forests or bioenergy plants, respectively. We analyze the effectiveness of the two CDR methods over time and in different world regions depending on the temporal CDR target. Furthermore, we quantify how the temporal dynamics are affected by the chosen start year of CDR (2015, 2030, 2050), background climate and CO2 concentrations (SSP1-2.6, SSP3-7.0), natural disturbances and assumptions on management and plant parametrizations in the underlying vegetation models. We specifically consider temporal dynamics on current agricultural areas adjacent to biodiversity hotspots, since these could also become relevant for achieving ecosystem restoration targets. There, CDR through restoration of naturally occurring forests or grasslands with support from local communities can bring synergies for multiple ecosystem services, while premature deployment of AR in non-forest areas or crop-based BECCS would likely decrease biodiversity.