Statistics in climate research: The importance of stochastic modelling and uncertainty quantification
Thordis Thorarinsdottir, Ph.D.
Date: February 21, 2019Time: 1:30 pm
Place: VG 0.111 (Verfügungsgebäude, Platz der Göttinger Sieben 7)
Talk in cooperation with RTG 1644 "Scaling Problems in Statistics"
High-resolution unbiased information on future climate change is commonly required for local impact assessment and adaptation decision-making. While climate models are our primary source of knowledge about future climate change, global and regional climate models are generally biased and their resolution is often lower than desired. Furthermore, not all climate variables are included in the climate model simulations. When producing climate projections at fine grid resolutions or individual locations, it is imperative to include stochastic components to model local variability not accounted for by the lower resolution model. This involves developing space-time models that are consistent with observational data over historical periods while also being computationally feasible. In subsequent decision problems it is imperative to propagate the uncertainty in the climate projections and other model components through the decision-making framework even if the resulting answer should be given by a single number. We discuss these aspects and show examples of daily mean temperature projections for Norway and sea level adaptation decision-making in Bergen.
Joint work with Qifen Yuan and Peter Guttorp.