Probabilistic assessment of land-based carbon dioxide removal and biospheric feedbacks under overshoot pathways

Limiting global warming to 1.5 °C is increasingly likely to involve temporary temperature overshoot followed by large-scale deployment of carbon dioxide removal (CDR). However, the effectiveness and reversibility of overshoot pathways remain uncertain due to climate–biosphere feedbacks and disturbance processes that may undermine net-negative emissions.

Here we present a probabilistic assessment of land-based CDR under overshoot scenarios using the reduced-complexity Earth system model OSCAR, extended with two new modules: OSCAR-Crop and OSCAR-Fire. OSCAR-Crop emulates climate–crop yield interactions for major food and bioenergy crops using Monte Carlo ensembles trained on complex crop model intercomparisons and field experiments, enabling efficient exploration of uncertainty in biomass availability for BECCS. OSCAR-Fire represents wildfire occurrence and post-fire carbon dynamics as functions of climate, vegetation, and human drivers, capturing both immediate emissions and delayed carbon losses as well as post-disturbance recovery.

We apply the fully coupled OSCAR framework to peak-and-decline pathways that temporarily exceed 1.5 °C before returning to lower warming levels through net-negative emissions. Results highlight substantial regional and probabilistic uncertainty in achievable carbon removal, driven by climate impacts on crop productivity, wildfire-induced carbon losses, and feedbacks between warming, land carbon sinks, and disturbance regimes. Our findings indicate that large-scale CDR deployment in overshoot pathways is constrained not only by socio-economic feasibility but also by nonlinear Earth system responses that may limit reversibility and increase climate risks.