Impact analysis workflows

Impact studies require converting Earth system model outputs into formats suitable for specialized impact models, such as crop yield models, hydrological models, and fisheries models. These conversions typically involve downscaling and bias correction, and sometimes other variable transformation or preprocessing steps (for example, the calculation of indices of climate extremes). Different research groups currently conduct this process independently, using ad-hoc methods, making results difficult to reproduce and compare.

Developing standardized, open-source workflows would accelerate impact research across the field while improving transparency and reproducibility. It would also help to ensure that impact modelers are using the widest possible set of the most up-to-date Earth system model runs, as new simulations emerge over time. Additionally, hosting the resulting, sometimes large, datasets publicly, in cloud-optimized formats reduces the technical and accessibility barriers that currently limit which researchers can conduct impact studies, broadening access to SAI modeling results and enabling a broader range of policy-relevant impact assessments. It also accelerates these research projects, since researchers can adapt one another's code without friction, and time spent downloading and managing local archives of data can be reduced.

This activity will support development of open-source, documented workflows that standardize the conversion of outputs from GeoMIP and other Earth System Model simulations of SAI into formats required by different impact models. These workflows will be developed collaboratively with researchers across different impact domains (e.g., crop models, human health), to ensure that the outputs match what is needed across a diverse set of research questions. This activity also includes conversion of any larger data outputs into cloud-optimized formats and public hosting of the data, for example by adding it to the Reflective Cloud Hub. This is a distributed research effort across multiple teams working on different types of impact.

Deliverables will include:

1. Cloud-hosted, analysis-ready GeoMIP data: with clear documentation of processing steps.
2. Standardized, versioned open-source code repositories: for each major impact domain.
3. Example notebooks and tutorials: demonstrating how to use the data and workflows.

By providing both the data infrastructure and the analytical tools, this activity will establish a foundation for rapid, reproducible impact research across multiple domains and enable researchers to focus on scientific questions rather than data engineering challenges.