ncx-gitbot
commented
1 year ago NCX response: The business as usual model is a hierarchical statistical model that predicts one-year harvest risk and intensity based on FIA training data and a suite of covariates that include geographic, biological, economic, and sociological factors. Partial pooling across forest types ensures that the model is able to leverage the similarity and ubiquity of covariate relationships across the forests of the continental U.S. while still allowing for regionally specific differences. This comment describes a situation where the model's predicted outcome does not match reality which is a possible situation. That is why we propagate and account for uncertainty and require a deduction associated with the uncertainty of carbon stocks in the project and baseline scenarios.
Commenter Organization: American Forest Foundation
Commenter: Lynn Riley
2021 Deferred Harvest Methodology Section: No Section Indicated
Comment: Model Credibility and Recommended Addition of Ex-Post Baseline Accounting: This methodology outlines a dynamic performance baseline approach. It could be made stronger, however, with the addition of an ex-post baseline during verification. Our understanding is that the baseline model/data would be updated frequently (annually or sub-annually), and that each update is used to inform the BAU assessment for the next cohort of enrollees. However, this does not address the frequent criticism of modeled baselines for forest carbon projects, in that even a robust, dynamic performance baseline model that learns and improves with each cohort cannot with assurance predict what will take place, even just a year ahead of time. For example, if a project instance enrolls in a 1-year program with this methodology, and it’s BAU assessment said that it’s carbon at risk of harvest was 70% of its merchantable volume. Two months after its project activity period begins, and the closest mill announces its closure; an unexpected pest invades the local baseline population; a global pandemic significantly alters timber prices. That 70% may not have turned out to be true, even though it was based on the best possible understanding at the time and conditions of the instance’s enrollment—and “best possible understanding” in this case is not well enough laid out in the methodology, as there is not enough transparency in the baseline model as currently provided (for example, example “non-exhaustive” covariates are listed in table 11-2, leaving methodology users with no minimum requirements of covariates to inform a baseline model). Modeled dynamic performance baselines can be robust approaches if they are transparently reviewed, locally calibrated, and conservatively discounted for uncertainty (as per the VCS methodology requirements); however, due to lack of transparency around the data and model in this methodology, these requirements have not been met. This methodology would be more credible if, as they are learning and in this case, because they have the data available, they look back after T1 and see if what was modeled in the baseline to happen between T0 and T1 actually took place, rather than set the baseline between T0 and T1 prior to the activity period (and all that could transpire) having happened. We propose building in this ex-post approach to the methodology to strengthen its credibility.
Proposed Change: Incorporate an ex-post baseline approach with the proposed model, rather than an ex-ante modeled baseline.