Assessment of biochar longevity and carbon dioxide removal (CDR) durability using modelling: implications for agricultural carbon management and global climate strategies

Intensified human activities have led to a dramatic growth of emissions of greenhouse gases (GHG), especially carbon dioxide (CO₂), causing significant global warming. To mitigate climate change, the Intergovernmental Panel on Climate Change (IPCC) has set a target of net zero emissions by 2050 and keeping global warming below 2℃ by 2100. However, to achieve this target, it is not enough to solely rely on reducing CO₂ emissions, but it also requires other ways for neutralization of emissions such as carbon dioxide removal (CDR). CDR technologies or approaches aim to remove and durably store atmospheric CO₂, and are considered to have significant potential for achieving net-zero or even negative emissions. In the current technological system dominated by land-based CDR, biochar technology is one of the most noteworthy. Biochar, a solid by-product produced from the pyrolysis of biomass residues under oxygen-limited conditions, features a highly porous structure and is rich in stable aromatic carbon. Biochar is not only a material for carbon storage, but can also enhance the sequestration of largest terrestrial pool of organic carbon—soil organic carbon (SOC) when applied as a soil amendment, with many additional advantages such as improving nutrient availability and promoting crop growth. As such, biochar application to soil has been considered as a cost-effective and competitive CDR option with multiple benefits. However, the durability of biochar for CDR is determined by the time that biochar can persist or biochar absolute longevity in soil. The accuracy of biochar longevity prediction in soil can directly affect assessment of biochar CDR durability and contribution potential, and thus influence the confidence of market investment and development of CDR strategy. This thesis aims to analyse and reduce the uncertainty in predicting biochar longevity in order to enable a more robust evaluation of biochar CDR durability. Due to the highly physico-chemical stability, biochar can persist in soil for centuries to millennia. It is impractical to determine the absolute longevity exactly by monitoring natural decomposition process from experiments. Currently, the most widely used method for estimating biochar longevity involves developing kinetic models based on short-term decomposition data from incubation experiments. However, this approach introduces significant uncertainty and often results in substantial discrepancies compared to black carbon—a natural analogue of biochar, due to the need to extrapolate over much longer timescales. Assessing the possibility of reducing the uncertainty by improving the extrapolation from kinetic models was the first objective of this thesis. Since two carbon components (pools) with different decomposition rates are detected in most experimental observations, a dual exponential model is often adopted to describe the short-term decomposition kinetics. This model was applied to fit decomposition data from 29 biochar incubation experiments to derive kinetic parameters. By analysing the distributions, correlations and predictabilities of these kinetic parameters, the possibility of inferring the decomposition kinetics of more stable pools to improve model extrapolation and reduce uncertainty was evaluated. The results showed that the kinetic parameters exhibited high variability, low correlations, and limited predictability, indicating a constrained potential to inform on more stable carbon pools or to reduce the uncertainty of kinetic models derived from short-term incubation data. Field experiments share similar limitations, but they can potentially be used to calibrate kinetic models and improve their responses to environmental conditions—at least theoretically in the short term under field settings. However, such calibration depends on a robust kinetic foundation, which in turn requires substantial statistical power from tightly controlled and standardised experimental systems. The second objective aimed to assess whether explainable decomposition kinetics can be derived under complex field conditions to enable the parameterisation of environmental factors and the calibration of kinetic models. By assuming biochar carbon to behave as a permanent, inert pool within the Rothamsted Carbon Model (RothC, Version 26.3), deviations between model-predicted and measured total organic carbon (TOC) following biochar application were analysed using data from 17 published field experiments lasting over three years. If the deviations of TOC stocks between field measurements and RothC model simulations can be interpreted from a kinetic perspective, the measurement data may be used to calibrate and enhance kinetic models for more accurate predictions of longevity and CDR durability. However, the results revealed highly visible and variable deviations caused by model-related and data-related uncertainties across 17 field experiments, highlighting the limitations of these field experiments in calibrating kinetic model and parameterising environmental responses to biochar decomposition. According to these limitations of current modelling methods, the third objective of this thesis was to explore an alternative approach for longevity modelling. Biochar carbon pools of varying stability were re-defined based on the aromaticity represented by the hydrogen-to-carbon (H:C) molar ratio. Using existing H₂O₂ oxidation data, complete kinetic curves of biochar decomposition under accelerated oxidation were generated, and subsequently calibrated to the natural timescale by the surface oxygen-to-carbon (O:C) molar ratio, which represents the ageing degree, with reference to known ages of black carbon. Based on the kinetic characteristics of each pool estimated, a new model for predicting absolute longevity was developed, described by a triple exponential function. Compared to other studies, this new kinetic model provides a different longevity prediction. The accuracy of new model prediction is highly dependent on the reliability of pool partitioning and kinetic calibration. Therefore, the effectiveness of using the H:C and O:C ratios to characterise biochar kinetics needs further validation with a broader range of biochars and black carbon samples. At the regional level, differences in predicted biochar longevity can inevitably affect the assessment of its CDR contribution potential, but longevity is not the only influential factor. Therefore, it is essential to quantify the extent to which uncertainty in longevity predictions affects the assessment of biochar CDR contribution potential on a global scale. This was the fourth objective of this thesis. Based on a global dataset of cereal crop residues production and usage, the availability of residues for biochar production and potential yield of biochar were estimated. Under a generalized simulation framework, biochar CDR contribution potential under different scenarios were assessed using different kinetic models integrated into RothC. The results showed that the CDR contribution potential of biochar estimated by different kinetic models were similar in most regions within a short timescale (75 years), but the discrepancies between predictions from different models may become significant in more regions with the timescale extending and climate changing. In conclusion, this thesis explored an alternative approach to modelling biochar longevity in response to the demonstrated limitations of existing methods, due to the constrained potential for reducing the extrapolation uncertainty in kinetic model developed from incubation experiments and low efficacy of current field experiments in calibrating environmental responses. The new model offers an alternative reference for carbon markets and policy makers in predicting biochar longevity and assessing CDR durability and contribution potential, thereby increasing market confidence and enhancing the credibility and practical utility of biochar within the broader CDR landscape. Further improvements are needed, particularly in the following areas: identification of the molecular characteristics of decomposed components in incubation experiments; robustness analysis of pool partitioning and kinetic calibration indicators; definition of standard decomposition environments and parameterisation of environmental effects; establishment of systematic benchmark sites for field experiments; and the development of more comprehensive datasets for regional CDR assessments.