The approach to carbon removal quantification for this multi-pathway system is three-fold:
Model the system – specific to both carbon removal activities conducted and the broader Earth system.
Quantify the intervention – inclusive of a publicly available protocol for quantification, in-situ instrumentation, and record keeping.
Audit the quantification – based on carbon accounting and life cycle assessment best practices, and the ISO 14064-2 standard for quantifying and reporting greenhouse gas emissions. Along with an assessment of total carbon impact, comprehensive third-party auditing also includes an independent review of environmental impact assessments prior to deployments, relevant project documentation post-deployment, and ongoing socioeconomic impact reporting.
Transparent quantification processes, carbon and emissions accounting following industry best practices, and independent external auditing are critical to ensuring the quality and credibility of any carbon removal approach. The inherent challenges in measuring and monitoring phenomena occurring over the vast expanse and depth of the open ocean require a focused and disciplined approach to quantification that is based on computational modeling, direct in-situ observation, and laboratory ground-truthing.
This approach is informed by best practices from the field of climate research. All Earth-scale systems require a layered approach to generate uncertainty-bounded estimates of the impacts of a perturbation over a large geographic area. For diffuse systems, direct measurements are not inherently better than modeling, as there reaches a point of diminishing returns between the cost of direct measurement and the additional insight it provides. In-situ direct measurement, remote sensing, and modeling are complementary, rather than competing, approaches to building a rigorous quantification system.
This quantification approach is built on and informed by the ‘best available science’, an established practice in natural resource management that ensures an activity evolves to match the best currently available understanding of Earth systems.
Governing Principles for Quantification
Quantification will be model-driven and validated by direct measurement and rigorous laboratory and field-based testing. The design, assimilation, tuning, and validation of the modeling framework will incorporate a variety of unique observational data from experiments designed to answer critical questions, as well as the latest scientific knowledge.
Quantification is uncertainty-bounded. This means that quantification computations must include not only the expected value of carbon removed but also the accumulated uncertainty which bounds this expectation, accounting for the uncertainty of the measured inputs for each specific variable and the modeling framework itself (i.e., model stability). The goal of this effort is to quantify the total carbon removed through an intervention and demonstrate that the uncertainty associated with that quantity is bounded within a given error. Effectively characterizing this uncertainty ensures that conservative discounting is appropriately applied, increasing confidence in the quantified outcome.
For the final estimation of removed carbon, programmed conservatism is applied, ensuring high certainty (95% or above) that at least the estimated amount of carbon was removed. In practice, this means that the estimate of carbon removed will be bound by the uncertainty (either lower or upper) in each input variable that yields the most conservative (i.e., lowest) quantity of total carbon removed. In cases of conflicting information with respect to model assumptions, such as standard emission factors, the more conservative assumption will be utilized.
The combination of these principles for conservative, model-driven quantification with bounded uncertainty provides a high degree of confidence in the total net carbon that has been removed from the system through a given intervention.
Figure 5: Quantification modeling system. An iterative system based on the best available science that integrates computational testing, empirical testing, and in-situ open-ocean data collection.
Modeling is at the core of this quantification approach. The ability to effectively model the system is achieved through:
Computational testing: applied to characterize model stability as distinct from variation in the input data. This analysis is incorporated into overall uncertainty.
Empirical testing: testing of models against large data sets generated in both laboratory and monitored open ocean settings, and characterizing model uncertainty against these observed data. Empirical testing is composed of two primary approaches:
Regulated: controlled variation of input parameters against a measured response. Where possible, regulated empirical testing will be a crucial component of model design. The key advantage of regulated testing is an ability to vary controlling parameters over the range expected in the offshore environment. In contrast to an uncontrolled observational environment, where controlling parameters are oftentimes correlated, properly designed regulated experiments can isolate causal relationships between the controlling responding parameters in the models.
Observational: monitored but uncontrolled variation of input parameters against a measured response. The observational testing will help validate and refine models built on experiments in regulated environments. For the model components that cannot be developed in regulated experiments, the observational testing is crucial.
Direct measurement: in-situ data collected from an instrumented subset of the open-ocean project establishes the correlation between modeled and observed behaviors and informs the quantification of environmental impact. This correlation analysis supports extrapolation from subset to project total; the stronger the correlation, the lower the overall uncertainty of the extrapolation.
Models must be based on acceptable industry standards, developed in accordance with peer-reviewed academic best practices, and ground-truthed with empirical data. If and where models are developed or adapted to be fit for purpose in quantification, it is critical that detailed descriptions of models used and how they support replicable quantification are documented and provided alongside project outcomes to enable external validation and review, and build trust in the underlying results.