Addressing Epistemic Risk in Actionable Earth System Science

As the fields of Earth system and climate science emphasize the importance of producing actionable data to guide policymaking, community action, and funding priorities, risks produced by researchers’ choices during the research process become more consequential. Epistemic risk refers to the risk of limitations, assumptions, or biases in research findings that result from unavoidable methodological choices or trade-offs during the research process, potentially rendering research results inadequate for their intended purpose. When research results are shared with decision makers without sufficient understanding of the limitations or assumptions within the dataset, a second level of epistemic risk occurs: decision makers may use scientific products for a need they are not appropriate for, resulting in undesirable outcomes. In this essay, we identify the epistemic risks that arose in five community-based research projects conducted by the authors as part of a multi-year program to produce actionable Earth system data and tools at the U.S. National Science Foundation National Center for Atmospheric Research. Our cross-case analysis illustrates that actionable science not only makes epistemic risk more visible, it also creates the conditions under which this risk can be addressed. Through community-based and convergence research approaches, user communities can contribute to methodological decisions about which limitations carry more weight, which assumptions to accept, and the level of uncertainty to accommodate. We find that these approaches allow for improved, multidirectional communication about uncertainty and assumptions in the data, thus allowing researchers and data users to make more informed decisions.