Climate Change Workshop

General Information
Application
Schedule

Climate Change Workshop Survey link: Click here to take survey

General Information

The Statistical and Applied Mathematical Sciences Institute (SAMSI) announces a Workshop on Statistical and Mathematical Problems in Climate Change, scheduled for February 17-19, 2010, at the Radisson Hotel in Research Triangle Park, NC. This workshop is part of the year-long program on Space-time Analysis for Environmental Mapping, Epidemiology and Climate Change. The workshop will bring together climate scientists and modelers, geographers, statisticians and applied mathematicians to explore a range of research topics in the climate sciences.

The two and a half day workshop is structured into 5 sessions. Each session will consist of two or three invited talks, followed by research presentations from one or more of the SAMSI working groups within the session topic; discussion time is also included in each session. In addition, Bruno Sanso will give an invited talk and tutorial on the topic "Spatio-temporal modeling for oceanic and climatological variables". As in all SAMSI workshops the goal will be to favor cross-pollination of ideas in the scientific disciplines that the program and the workshop bring together.

The topics of the five sessions are: Paleoclimate Analysis, Extremes, Climate Models, Regional Climate Modeling and Downscaling and Uncertainty Characterization.

Paleoclimate Analysis: Past climates are an indispensable test bed for our theories and models of climate. The instrumental climate record covers only about 100 years. Climate information beyond the instrumental record has to be inferred from proxies such as tree rings and speleothems. This usually involves establishing statistical relationships between the proxies and target climate variables such as surface temperatures, and then using such statistical relationships to infer past climates given proxies. The statistical problem can be understood as the problem of analyzing combined proxy and instrumental data with missing values, where the instrumental data in the past are the sought missing values. Spatial and temporal correlations among the data are important, and the estimation problems are typically ill-posed or ill-conditioned. Recent advances in spatial statistics and applied mathematics (e.g., L1 methods) have the potential to improve our inferences about past climates. The field would benefit from increased interaction between statisticians and climate researchers--an interaction this workshop is aiming to foster.

Extremes: A very lively topic in climate research concerns the effect of climate change on extreme events, such as extreme temperature and rainfall events, extreme wind speed events including hurricanes, and droughts. There is much interest in quantifying trends in extreme events and determining to what extent those trends can be attributed to human causes. However, there have also been new methodological developments in extreme value theory, particularly to spatial and spatio-temporal extremes and associated stochastic processes, especially max-stable processes. This session will feature talks describing both the climatological and statistical aspects of this field, and will develop interactions between them.

Climate Models: This section will address questions in the area of numerical models simulating the climate system. The challenges deriving from the multiscale nature of the climate system and current research directions in stochastic modeling will be discussed. A range of model complexity will be considered, from simple to intermediate complexity to fully coupled climate models. Moreover, data-driven techniques for understanding climate trends, validating and improving climate models will be presented.

Regional Climate Modeling: Global Climate Models (GCM) are a fundamental tool in climate change science. Currently, most of them work by discretizing the surface of the Earth into grid boxes on the order of 100 km in the latitude/longitude dimensions. However, behavior at finer scales is critical in assessing the risks of climate change. Thus GCM results need to be downscaled using methods that can be grouped into three categories: (1) dynamic downscaling, by nesting a higher resolution Regional Climate Model in a GCM. (2) statistical downscaling, by estimating a statistical relation between fine scales and large scales on the basis of observational data, which is then applied to large scale GCM output; (3) statistical-dynamical downscaling that involves methods which combine features of (1) and (2). This section will focus on methods for assessing and improving the quality of downscaling procedures.

Uncertainty Quantification: Climate science relies on both scientific models and observational data, both involving uncertainties. Thus identifying, quantifying and managing these uncertainties is paramount.

        Observational uncertainties include traditional statistical variations, measurement errors, and representational errors and biases. Other issues arise from the high-dimensionality and complexity of modern, often remotely sensed, spatial-temporal datasets. Historical data ranging from earlier centuries to earlier millennia offer additional challenges. Finally, some data sets are often computed by combining observational measurements and physical models, and thus suffer from a complex combination of measurement and model errors.

        In numerical models of the physical system uncertainty arises from such things as: approximate physics; the coupling of components representing the individual subsystems (i.e., atmosphere, ocean, land and ice), operating at a range of spatial-temporal scales; parameterized or altogether unrepresented processes, imprecise forcings, and uncertain feedbacks; and coding and numerical errors.

        While there is awareness of the need for uncertainty quantification, determining the methods to explore and quantify these uncertainties is difficult, because of the high computational demands, the number and interdependence of the uncertainty sources and the challenges in constraining model behavior through observations, especially relevant when it comes to future changes. Further difficulties arise in representing uncertainties and communicating results in a fashion usable by consumers of climate science information such as policy makers and the public.

Organizers: Mark Berliner (Ohio State), Jim Zidek (British Columbia), Richard Smith (North Carolina-Chapel Hill), Bruno Sansó (UC Santa Cruz), Tapio Schneider (California Institute of Technology), Claudia Tebaldi (Climate Central), Ilya Timofeyev (Houston).

Application

REGISTRATION IS CLOSED. CAPACITY HAS BEEN REACHED

Please make reservations at the Radisson RTP as soon as possible. The SAMSI room block and rate ($109) is effective until February 1, 2010. After this date, there is no guarantee a room will be available.  If you have a change in plans, individual room reservations must be cancelled 72 hours prior to arrival. Check-in is at 3:00 PM; check-out is 12:00 noon.

Schedule

Wednesday, February 17, 2010
Radisson RTP

Thursday, February 18, 2010
Radisson RTP

Friday, February 19, 2010
Radisson RTP

Entire site © 2001-2010, Statistical and Applied Mathematical Sciences Institute. All Rights Reserved.