Critically evaluate whether your findings answered the original question(s). Check back in with the people who were involved in determining data needs for the application.
If you determine that the results were not sufficient to answer your original question(s), either reframe the analysis by re-visiting one or more of the above principles, or reframe the original question so that the available data can sufficiently answer it.
Finalize the documentation for your analysis once you determine that the results are sufficient. At a minimum, this should include the steps taken within your workflow, as well as limitations, assumptions, and decisions made at each step.
Consider a plan and timeline for updating, improving, or changing the analysis as new information or resources become available.
Determine the format in which the analysis and results should be presented and communicated with the intended audience(s) (e.g. reports, one-pagers, figures/maps, presentation formats).
Consider archiving the data needed to reproduce your final analysis. If space constraints prevent downloading or retaining the full dataset, consider saving a smaller derived or processed dataset to enable future changes to final calculations.
Important Considerations
Climate data analysis can be an iterative process. It often takes multiple evaluations and iterative data assessments to develop a robust analysis that produces results that sufficiently answer a question.
Sometimes the questions we ask of the data cannot be answered, because of the data’s inherent limitations. It is important to understand that a “non-result” can still be a valid answer to your original question and may provide meaningful information about the question and/or analysis (e.g. the uncertainty range is too large to distinguish a signal, or the statistical significance is too low). Do not cherry-pick results to answer the original question.
The goal of documenting your workflow should be to ensure that others can replicate or build upon your analysis in the future. Consider focusing on: transformations to the original dataset, spatial or temporal aggregations or disaggregations, coordinate system transformations, metric calculations, statistical tests applied, technical documentation of developed code, and data download format.
Consider sharing your results, data, and analysis via an open-source forum, such as an open-access code repository (e.g. GitHub), online blog post, or open-access journal (if appropriate). Be aware that many scientific journals and regulatory institutions strongly recommend — if not require — that code and data be open-access and citable via DOI.