Frontiers in the Detection and Attribution of Climate Change (12w5037)
Amy Braverman (Jet Propulsion Laboratory, California Institute of Technology)
Paul Kushner (University of Toronto)
Richard Smith (University of North Carolina Chapel Hill)
Daithi Stone (Lawrence Berkeley National Laboratory)
Claudia Tebaldi (NCAR)
Michael Wehner (Lawrence Berkeley Lab-Scientific Computing Group)
The workshop will focus on developing and improving the statistical methodology of detection and attribution of climate change (D&A). Thus, we will seek ways in which human influence on Earth's climate, and the attendant impacts of climate change on human systems, can be more precisely quantified in light of the latest developments in applied statistics, climate modelling, and Earth observations. We will concentrate our exploration of this topic via lectures, poster presentations, and discussions on three fields we have chosen: physical climatology (including atmosphere/ocean science, cryospheric science, and hydrology), agricultural science, and epidemiology. These fields heavily rely on applied statistics but use distinct statistical approaches, including treatment of uncertainty and confounding factors. Further, all three fields are deeply concerned with the impacts of anthropogenic climate change but use climate inputs (climate models and earth observations) in distinct ways.
We will combine expertise in statistical climatology, epidemiology, and agricultural statistics with the aim of initiating the development of new D&A methodologies based on rigorous statistical approaches and using a hierarchy of models, including process based, statistical, and computational models. We also aim to include enough expertise in uncertainty quantification of existing and planned Earth Observation systems to allow observational uncertainty analysis to be integrated into this effort.
The invited participants and time allotted in the workshop will reflect our desire for cross fertilization between the three fields. For instance, one session we envision will treat event attribution from the perspective of climatology, epidemiology, and agriculture: In each field, how is an "event" defined and detected, how is its cause inferred quantitatively from available data, how is confidence in this inference measured, and how might newly available Earth observations improve the methodologies? Such a session will be lead by one or two chairs and will incorporate lectures, directed discussion, and undirected discussion. An ideal situation would be one in which a lecturer provides core concepts for half of his or her talk, and then initiates a moderated discussion based on "blue skies" ideas that would involve people from within and outside his or her field. Parallel breakouts might also be employed towards the end of the week, and poster presentations will be used for additional presentations.