In the Fall of 2019, the Information Quality Cluster (IQC) published a white paper entitled “Understanding the Various Perspectives of Earth Science Observational Data Uncertainty” (Moroni et al., 2019; DOI: 10.6084/m9.figshare.10271450). This paper provides a diversely sampled exposition of both prolific and unique policies and practices, applicable in an international context of heterogeneous policies and working groups, made toward quantifying, characterizing, communicating and making use of uncertainty information throughout the comprehensive, cross-disciplinary Earth science data landscape from the following four perspectives: Mathematical, Programmatic, User, and Observational. These perspectives affect policies and practices in a diverse international context, which in turn influence how uncertainty is quantified, characterized, communicated and utilized. The IQC is now in a scoping exercise to produce a follow-on paper that is intended to provide a set of recommendations and best practices regarding uncertainty information. It is our hope that we can consider and examine additional areas of opportunity with regard to the cross-domain and cross-disciplinary aspects of Earth science data. For instance, the existing white paper covers uncertainty information from the perspective of satellite-based remote sensing well, but does not adequately address the in situ or airborne (i.e., sub-orbital) perspective. The 2020 Winter ESIP meeting provided an IQC session to explore these additional perspectives from Uncertainty Quantification (UQ) and Uncertainty Characterization (UC) work done with Argo float data (presented by Mikael Kuusela) and challenges from the lens of the upcoming S-MODE EV-S mission (presented by Fred Bingham). Participants in this 2020 Summer session are highly encouraged to:
1) Attend the “Part 1” session to acquaint themselves with the scope of material and perspectives that will be discussed more thoroughly in this “Part 2” session. This “Part 2” session will be a “working” session, starting with an introductory panel discussion of the topics discussed in “Part 1”, followed by a mini breakout into teams that each focus on aspects of UQ/UC to more thoroughly define the scope of new perspectives and best practices, concluding with summary presentations describing the findings of each team.
2) Dive right in to help provide inputs that will be used to establish a working draft for a follow-on white paper using the requisite prior knowledge of the issues presented in Part 1, as well as any additional knowledge you may have regarding UQ/UC.
With the inputs acquired from this session, the IQC will then coordinate with those who are interested in taking part as co-authors or contributors to establish a more comprehensive set of perspectives of UQ/UC information within the realm of Earth science, culminating with a foundational set of recommendations and best practices that are representative of those perspectives.
Agenda:
4:00-4:10 Session Part 2: Introduce the Working Session
4:10-4:15 Zoom Breakout Sign-up/Assignments
4:15-4:50 Zoom Deep-Dive Breakouts:
- Zoom Breakout 1: UQ Perspectives on Modeling and Data Assimilation
- Zoom Breakout 2: UQ Perspectives on Sub-Orbital and In Situ
- Zoom Breakout 3: Best Practices for producing UQ/UC Information
- Lead: Hampapuram "Rama" Ramapriyan
- Zoom Breakout 4: Best Practices for interpreting/using UQ/UC Information
4:50-5:20 Plenary Deep-Dive Summary Debrief
5:20-5:30 Summarize 3 Takeaways and Closing
View Recording
View Session Notes
View Presentations: See attached.
Takeaways- Need to look into new use cases, particularly capturing modeling/assimilation and end user perspectives on implications of uncertainty on multiple interpretations of scientific assessments.
- Recognition of the Emerging need to resolve the difference in spatio-temporal scales between in situ, airborne, satellite and modeled/assimilated data; common ways of resampling for data fusion applications, and understanding the uncertainty associated with data resampling (i.e., representation error).
- Need to have discipline-independent, open-source tools to support UQ workflows for algorithm developers who are doing the UQ estimation, which would allow for more consistent representation of UQ in data files; having the UQ/UC information consistently represented with a common vocabulary in data files addresses users’ needs to use UQ/UC information in a manner that fits their purpose.