ISSB and IFRS S2: Why Validation and Traceability Matter in Climate Analyses

A climate analysis is not reliable simply because it presents detailed maps, graphs, or projections through the end of the century. To support decision-making and public communication, it is necessary to understand where the data comes from, how it was processed, how well it performs, and what limitations remain.

This concern takes on greater significance in light of the standards set by the International Sustainability Standards Board (ISSB), a board of the IFRS Foundation responsible for financial disclosure standards related to sustainability. In this context, methodological transparency and high-quality evidence help ensure that the information can be reviewed by managers, investors, and professionals responsible for auditing the studies.

What do IFRS S1 and IFRS S2 establish?

IFRS S1 sets out general requirements for disclosing information about sustainability-related risks and opportunities that may affect an organization’s prospects. IFRS S2 specifically addresses climate-related issues, including physical risks, transition risks, and opportunities.

IFRS S2 requires the use of scenario analysis to assess the resilience of the strategy and business model. It also requires information on the scenarios used, their sources, the time horizons considered, the key assumptions, and the relevant uncertainties.

The standard does not prescribe a single method for climate modeling or a mandatory list of statistical tests. However, presenting well-founded information requires evaluating the appropriateness of the inputs and explaining the choices made. It is at this point that technical validation contributes to the quality of the analysis.

Validation means evaluating performance, not guaranteeing the future

Climate projections describe futures that depend on specific scenarios and methods. It is not possible today to verify whether a condition projected for decades from now will actually occur.

Among other aspects, it is possible to evaluate the ability of models and processing procedures to represent historical climate characteristics, in comparison with an appropriate reference standard.

This assessment must be consistent with the study’s purpose. A model that satisfactorily represents average temperature may have limitations when it comes to the frequency of extremes. Similarly, reproducing annual precipitation does not guarantee an accurate representation of the duration of dry periods or the intensity of rainfall.

The question is not just “Was the data validated?”, but “Which characteristics were evaluated, based on what criteria, and with what results?”

What should a technical evaluation make explicit?

Useful documentation for a review must identify the reference dataset, its coverage, resolution, limitations, and the time period covered. It must also explain the procedures for quality control, regionalization, and bias correction.

Depending on the application, the assessment may include systematic errors, differences in distributions, seasonality, percentiles, the frequency of extremes, and the persistence of dry or hot spells. There is no single metric capable of demonstrating the adequacy of all these aspects.

Whenever possible, evaluating performance over a period different from the one used to fit the model helps distinguish between the quality of the fit and the model’s ability to apply to data not used in the calibration. If the evaluation takes place during the same period as the fitting, this fact should be disclosed, without presenting it as an independent validation.

These procedures are technical best practices that support the study’s rationale; they do not constitute a universal statistical protocol required by IFRS S2.

Climate regionalization does not preclude verification

The application of a climate regionalization method aims to improve the match between historical simulations and a reference. However, the name of the method alone does not indicate the quality of the results.

It is important to examine performance before and after processing, the physical consistency of the data, and the properties relevant to the intended application. When an index combines temperature, humidity, wind, or radiation, for example, the consistency among variables also warrants attention.

Similarly, presenting a result for a coordinate does not make it an observation measured at that point. The spatial representativeness and the limitations of the dataset used must be taken into account in the interpretation.

Auditability: allowing the path to the result to be examined

Validation assesses performance. Traceability identifies the origin and transformations of the data. Auditability depends on evidence and records that allow these procedures to be examined.

To this end, it is important to document models and versions, scenarios, coordinates, time periods, units, index definitions, processing parameters, and criteria for composing the multimodel set. Validation metrics should be accompanied by interpretations and limitations, rather than simply presented as isolated tables.

Records must allow a conclusion to be linked to the data and calculations that support it. The reproducibility of results also depends on authorized access to the necessary inputs and procedures.

Being auditable does not mean that an organization has already been audited, certified, or approved by an institution. Nor does a climate report, on its own, ensure compliance with IFRS S2: disclosure also depends on an assessment of the organization’s exposure, vulnerability, financial impacts, governance, and strategic response.

Scenarios and uncertainties also require justification

Using various models and scenarios helps to explore differences among possible futures, but does not eliminate the need to justify their selection. IFRS S2 requires transparency regarding the appropriateness of the scenarios and the uncertainties relevant to the resilience assessment.

Ranges between models should not be automatically interpreted as probabilities of future outcomes. It is necessary to explain what they represent and which sources of uncertainty they do not account for.

How Astra Contributes

Astra Science & Data organizes its products to support this need for a solid foundation: ad hoc climate analyses using different scenarios and models, results accompanied by uncertainty estimates, methodological documentation, and validation metrics included in the technical report. The structured files provide the data for the client’s analyses.

This framework provides guidance for identifying physical threats in studies guided by IFRS S2, without replacing the organization’s risk assessment or full disclosure. The goal is to provide a climate-related foundation whose origins, performance, and limitations can be examined—not just numbers to fill a report.

Previous
Previous

Next
Next

Why analyze climate risk before making decisions?