Page updated 5 December 2005

Workshop Home

List of Participants

Newsroom

Background

Program and Events

Agenda

Abstracts

Presentations

Call for Contributed Presentations

Sponsors

Registration and Rates

General Information

Now available in PDF format: Abstract Book [7.4 Mb] (posted 10 November 2005)

Abstracts for Posters

General Climate Science (P-GC)

Sub-Theme 2: Model Evaluation & Strategy

P-GC2.1

Simulations of Committed Climate Change and Sea-Level Rise through 2400 AD

William Collins, NCAR, wcollins@ucar.edu

Gerald Meehl, NCAR

Tom Wigley, NCAR

Haiyan Teng, NCAR

The effects of sea-level rise in the next few centuries are of particular concern for coastal regions and many island nations. This talk summarizes simulations of the minimum, or committed, sea-level rise for the 21st through 23rd centuries based upon models analyzed for the IPCC AR4. The committed sea-level rise is the increase in sea-surface level associated with the historical changes in greenhouse gases to date and the associated trends in ocean temperature and ocean volume. The results, at least for the 21st century, are unaffected by future scenarios of emissions and emission controls. Both simple and complex models of the ocean atmosphere system suggest that global-average sea level will increase by approximately 10 cm over the next century. These results represent the minimum increase in sea level, since most of the complex models neglect the effects of melting continental ice sheets, including those of Greenland and the Antarctic ice sheets.

[Poster PDF]

P-GC2.2

Model Evaluation and Projections of Climate
for the Middle and Upper Atlantic Region of the United States

Raymond Najjar, The Pennsylvania State University, najjar@meteo.psu.edu

Steve Graham, The Pennsylvania State University

Steve Crawford, The Pennsylvania State University

Steve Greybush, The Pennsylvania State University

The Consortium for Atlantic Regional Assessment (CARA) aims to provide information to stakeholders through an interactive web site in order to help them make decisions that may be affected by climate change and land use in the Mid- and Upper-Atlantic Region of the United States. Model evaluations and projections are an essential component of the climate assessment process. In that spirit, we have analyzed the regional output of seven global climate models forced by two greenhouse gas emissions scenarios from 1900 to 2100. We evaluated the models in terms of their long-term means and variability (seasonal, interannual and decadal) using the temperature and precipitation observations at U.S. Historical Climate Network (HCN) stations. Because model and observed topographies differ substantially (due to the coarse resolution of the models), an altitude correction is made to the temperature output of the models, which reduces simulation errors by 20%, on average. When averaged over the models, the simulated mean temperature at the HCN stations for the current climate (1971-2000) is 0.12ºC higher than observed, with typical (i.e., RMS) errors of 1.4ºC. The mean precipitation at the HCN stations for the current climate has a model mean of 3.0 mm per day, which is equal to the observed, and typical errors are about 15%. Model seasonality in temperature was evaluated using the difference between the summer (June-August) and the winter (December-February). The model-mean seasonal difference is 14% larger than observed. Observed precipitation in the region is highest in summer and lower in winter. The model-mean summer-winter difference in precipitation is about 10% greater than the observed, but the phasing of precipitation is generally ahead of the observations, with four of the models putting highest precipitation in spring and five putting lowest precipitation in fall. Analysis of simulated interannual variability, decadal variability and future projections, which were not completed at the writing of this abstract, will also be presented.

[Poster PDF]

P-GC2.3

Development of Regional Probabilities of Climatic Change for Decision Making

Linda Mearns, NCAR, lindam@ucar.edu

Claudia Tebaldi, NCAR

David Yates, NCAR

Kathleen Miller, NCAR

It is known that in some resource sectors, such as water resources, probabilistic information about future climate, can be useful for decision making by e.g., water managers. Several techniques have now been developed to provide, specifically, regional probabilities of climate change. We will present an assessment of these methods, including those of Tebaldi et al. (2004, 2005), Greene et al. (2005), and Raisenen (2005). These three methods make use of the newest simulations of climate change by AOGCMs produced for the IPPC AR4. In this regard, the methods are all based on multi-model ensembles (multiple AOGCMs running the same climate experiments), and two of them take a Bayesian approach. The three methods produce different results for large regions, particularly in the width of the distributions. Even though these methods should be viewed primarily in the research mode, we present some examples of how such probability distributions could be used in a decision making context. Use of these research results by stake holders (e.g., water managers) can be useful for pedagogical purposes, and such iterations among climate scientists, statisticians, and users should result in refinements of methods in service to decision making.

[Poster PDF]

P-GC2.4

Evaluation of Regional Climate Simulations for Water Resources and Air Quality Assessment

L. Ruby Leung, Pacific Northwest National Laboratory, ruby.leung@pnl.gov

Yun Qian, Pacific Northwest National Laboratory

William Gustafson, Pacific Northwest National Laboratory

We have developed and applied dynamical downscaling methods for generating regional scale information of climate variability and change. This presentation will focus on evaluation of regional climate simulations for the U.S. at time scales ranging from diurnal to seasonal and interannual,
emphasizing the hydrological cycle including different aspects of precipitation characteristics and extreme events. These include simulations driven by global reanalyses and global climate simulations. The downscaled and global simulations will be contrasted to examine the value added by dynamical downscaling. Projections of future climate changes produced using dynamical downscaling of different global climate simulations will be discussed in the context of their applications to assessing water resources and air quality impacts, and to elucidate uncertainties. Lastly, prospects for new developments in regional climate modeling and their relevance to addressing model bias and interdisciplinary issues important to the society will be discussed.

P-GC2.5

The North American Regional Climate Change Assessment Program (NARCCAP):
Multiple AOGCM and RCM Climate Scenario Projections for North America

William Gutowski, Iowa State University, gutowski@iastate.edu

Linda Mearns, NCAR

Raymond Arritt, Iowa State University

Sebastien Biner, OURANOS

George Boer, CCCma

Daniel Caya, OURANOS

Phil Duffy, LLNL

Michel Giguere, OURANOS

Filippo Giorgi, ICTP

Isaac Held, GFDL

Richard Jones, Hadley Centre

Rene Laprise, UQAM

Ruby Leung, PNNL

Ana Nunes, Scripps

Jeremy Pal, ICTP

Yun Qian, PNNL

John Roads, Scripps

Lisa Sloan, UC/Santa Cruz

Eugene Takle, Iowa State University

The North American Regional Climate Change Assessment Program is using an ensemble of global and regional climate models (GCMs and RCMs) to produce downscaled estimates of changes in water and energy cycles decades into the future. The program plans to include the RCMs MM5, HadRM3P, RegCM3, the Canadian Regional Climate Model (CRCM), the NCEP Regional Spectral Model (RSM), and the Weather Research Forecast (WRF) model. The models are being driven by NCEP reanalyses and will eventually use output from several GCMs: potentially the Hadley Centre's HadCM3, the NCAR CCSM, the Canadian CGCM3 and the GFDL AOGCM. The resulting climate model runs will form the basis for multiple high-resolution climate scenarios to be used in climate change impacts assessments in the U.S. and Canada. High-resolution global time slice experiments using the GFDL AGCM and the NCAR AGCM CAM3 will also be produced and compared with runs of the regional models. The collective analysis of output from multiple models provides a framework for projecting climate change and its uncertainty over multiple time scales.

Initial results show precipitation and temperature biases that are comparable to those seen in previous regional simulations, even though the
program is using a large domain covering most of North America. Perhaps most important, model output indicates that the collection of all models produces a more complete depiction of the uncertainty in estimating precipitation change than could be obtained from a single model.

[Poster PDF]

P-GC2.6

Observed and Modeled Climate Variability over the United States
Associated With Major Teleconnection Patterns

Katharine Hayhoe, Department of Atmospheric Sciences, University of Illinois at Urbana-Champaign, Urbana, IL; Department of Geosciences, Texas Tech University, Lubbock, TX

Anne Hertel, Department of Atmospheric Sciences, University of Illinois at Urbana-Champaign, Urbana, IL

Donald Wuebbles, Department of Atmospheric Sciences, University of Illinois at Urbana-Champaign, Urbana, IL

Atmospheric circulation patterns and their connections to surface conditions are among the primary influences on climate and variability over the U.S. As climate is expected to continue to change in the future in response to emissions from human activities, the need for informed decision making based on the latest understanding of climate and how it is changing is likely to become even more urgent.

Here we first assess historical observed surface temperature and precipitation variability associated with six major atmospheric teleconnection
patterns that affect surface conditions across the continental U.S., including the El Niño-Southern Oscillation and the North Atlantic Oscillation. We next evaluate the ability of four atmosphere-ocean general circulation models commonly used to produce simulations of future climate (PCM, HadCM3, CCSM and GFDL) to reproduce historic atmospheric circulation patterns and variability, as defined by the standard indices based on pressure, geopotential heights, winds, and sea surface temperatures. Model ability to simulate the teleconnection patterns in terms of the temporal frequency of index itself as well as the sign and magnitude of their influence on regional-scale temperature and precipitation has important implications for the ability of these models to simulate climate and climate change over these regions. Finally, we assess projected changes major teleconnection patterns under conditions of future change as simulated for the SRES A2 (mid-high) and B1 (lower) scenarios to demonstrate how these are likely to continue to influence climate over the continental U.S. in the coming century.

P-GC2.7

Providing Climate Information Across a Continuum of Time Scales

Arthur M. Greene, International Research Institute for Climate Prediction, The Earth Institute at Columbia University, Palisades, NY, amg@iri.columbia.edu

Lisa Goddard, International Research Institute for Climate Prediction, The Earth Institute at Columbia University, Palisades, NY

Walter Baethgen, International Research Institute for Climate Prediction, The Earth Institute at Columbia University, Palisades, NY

Climate-related decisions may be made with time horizons ranging from weeks to decades; decision-makers thus require useful predictive information on a correspondingly wide range of time scales. In the seasonal-to-interannual (SI) domain, ENSO drives climate variability in many parts of the world, providing a basis for prediction. At the other end of the spectrum, on centennial time scales, coupled atmosphere-ocean general circulation models (AOGCMs) may have something useful to say as well. However, given the intermediate time horizons on which many decisions are made, typically years, out to one or two decades, neither ENSO-based predictions nor the long-range climate projections provided by AOGCMs may provide appropriate guidance. The present work addresses this gap in available forecast products and tools.

Secular changes in climate are part of the "big picture" that decision-makers must assimilate as they look forward in time. Such changes are not experienced separately, however, but constitute only one component within the full spectrum of climate variability. Providers of climate information must therefore take variability across a range of time scales into account, these time scales in turn informing a range of forecast products. These products may derive from different kinds of predictions or even different prediction tools.

It is the goal of the International Research Institute to provide climate information in support of decision processes on a continuum of time scales, with a focus on developing countries. In this regard, collaborative work in Brazil, regarding water resources and economic development, is discussed. Here, a "layering" approach is adopted, in which secular variation is taken as a context for the interpretation of SI forecasts. Decisions can then be informed by a fully dynamic view of climate variability. The underlying philosophy considers multi-model climate change outlooks and probabilistic decadal predictions as a background against which response to SI forecasts may be considered, in the context of long-term planning.

[Poster PDF]

P-GC2.8

Needed: A National Policy in Regard to Climate Modeling

David Douglass, University of Rochester, douglass@pas.rochester.edu

In most fields of science, when a theory or model disagrees with observation the theory or model is considered to be wrong and is not taken seriously. This is not the case in the field of climatology. Consider the following example: Douglass et al. [1] published a paper: Altitude dependence of atmospheric temperature trends: Climate models versus observation in which they stated: "... all state-of the-art general circulation models predict a positive
temperature trend that is greater for the troposphere than the surface. ... the temperature trends from observational data sets ... show decreasing values." This devastating result has been ignored along with other such observations.

There is no accountability.

A new paradigm is needed.

Climate modeling began with individuals [Lorenz on a Royal McBee LGP-300] and has progressed to large groups [more than 20 worldwide] with super-computers. The magnitude of the funding and the uncritical publicity surrounding certain extreme predictions have catapulted climate science into a position of national importance to such an extent that policy makers are keenly interested in the results. These models represent a national resource that has attained a status comparable to that of a mission pursued at national facilities [telescopes, nuclear accelerators] and should be recognized as such. Future use of these climate models requires a new paradigm. At present, scientists:

• Build the models and also
• Determine policy in the use of the models results.

To insure accountability and reduce conflicts of interest these two functions should be separated as they are in other most other scientific fields of national interest (nuclear physics, astronomy, etc.). The model builders can not possibly conceive of or understand the full implications of the use of the models. The models and the use of the model results should be recognized as separate and independent.

Recommendation

The new paradigm calls for the creation of a committee or commission, a continuing body with rotating membership with no more than half from the climate community. A possible mission statement could include:

• Establishing national guidelines
• A mechanism to insure honesty and scientific integrity
• Oversee and make policy such as
  • 30% of each models time must go to outside scientists with a mechanism to ensure success in their use of the models
• Strategy in the use of all model results
• Setting up subcommittees with power, such as an "experiment selection committee"

¹Douglass, Pearson and Singer. GEOPHYS. RES. LET., VOL. 31, L13208, doi:10.1029/2004GL020103

P-GC2.9

A Closer Look at Sea Surface Temperature (SST) Trends

S. Fred Singer, University of Virginia and SEPP, singer@sepp.org

We examine here the fundamental physics of SS heating by short-wave (SW) and long-wave (LW) radiation. Solar SW radiation penetrates to some considerable depth, depending on wavelength and turbidity. With an average ocean albedo of 0.09, most of the visible part of the solar spectrum heats the euphotic zone, and through wave action and eddy mixing communicates this energy downward, heating the "mixed layer," conventionally taken as the upper 100 meters. But LW (atmospheric) radiation, typically around 10 microns, cannot penetrate into water beyond a "skin" of about 10-micron thickness. As a result, the enhanced (anthropogenic) greenhouse effect from an increase in GH gases may make only a minor contribution to SST. Thus SST should not warm appreciably in response to the anthropogenic GH effect.

The absorbed IR energy goes partly into radiation and partly into sensible and latent heating of the atmosphere. The processes involved cannot be reliably calculated. In order to discover what fraction of downwelling IR contributes to SST, one must carry out a measurement program in which the downwelling IR flux varies. We describe such an experiment.

But SST records of the past 25 years (referring to temperature of the mixed layer) do show an increase, comparable to that of land data. This temperature rise has conventionally been ascribed to GH warming.

To account for the obvious disparity, we examine more closely the types of measurements that make up the SST. They consist principally of temperature data from engine-cooling water measured at ship inlets (typically around 10 meter depth and below the euphotic zone) and—since 1980—an increasing amount of data from drifter buoys in the (warmer) euphotic zone.

We hypothesize that the reported SST warming is largely an artifact of the increasing percentage of (higher-temperature) buoy data. We also suggest various tests for falsifying the hypothesis.

Supporting the hypothesis is the reported disparity between surface and atmospheric temperature trends in the tropics [Douglass, Pearson, Singer, GRL 2004]. We suggest that the SST rise observed prior to 1940 is real but not caused by GH warming.

[Poster PDF]

P-GC2.10

Issues for Use of Climate Models to Inform Policymakers,
Assess Impacts, and Develop Adaptive Strategies

Robert Livezey, NOAA/NWS Climate Services, Silver Spring, MD, Robert.E.Livezey@noaa.gov

Take-Home Messages:

1. Impact assessment and scenario development must approach climate model output far more critically, conducting expert and thorough historical record validation of all critical aspects of the problem as a first mandatory step.

   Otherwise the assessments or scenarios may be worthless or, worse, misleading.

2. Model validation needs greater research attention, both to meet user needs under (1) and to sensitize modelers to deficiencies.

   Currently, model validation is grossly inadequate.

3. More attention needs to be paid to the development of credible mesoscale (to avoid downscaling compromises) global coupled models that correctly treat the full spectrum of variability.

   Downscaling (whether statistical or with nested models) is inherently flawed.

Discussion:

Global climate change will not be uniform either geographically or seasonally. Associated changes will not only be in temperature means but in its variability, both intraseasonal and interannual, in the means and variability of precipitation, winds, etc. and in the risks of high impact weather types and events. It is rare that a policymaker, adaptive planner, or resource manager concerned with climate sensitive issues or sectors should not have a critical interest in this granularity of change, certainly at a minimum the seasonal and geographic distribution of changes in the mean temperature and precipitation; changes in the mean annual global surface temperature are of no practical interest to anyone by themselves.

Unfortunately, state-of-the-art climate models that credibly product the historical trace of the global mean annual temperature also credibly reproduce at best the geographic distribution of mean temperature trends in some, but not all, seasons. Major deficiencies remain in historical simulations of some aspects of mean temperature changes and all aspects of mean precipitation changes, while little or no attention has been paid to validation of variability, extremes, etc. and their changes. The ability of these models to credibly treat much of the latter depends on their ability to reproduce the form, seasonality, and variance of the phenomenon that constitute the dominant controls on weather systems and their variability. These phenomena minimally include ENSO, the MJO, and the NAO. Since no climate model has been shown to collectively and correctly treat all three it is unlikely that any model can currently and credibly address variability or risks of high impact events. Because of these and other deficiencies, downscaling (whether statistical or dynamical) based on perfect model approaches cannot be a productive approach to scenario development in almost all instances. Correction approaches are problematic as well because of data limitations and the inherent non-stationarity of climate change.

[Poster PDF]

P-GC2.11

Towards Interoperability of Global Geospatial Data Sets: Discrete Global Grids

Michael Freeston, Institute for Computational Earth System Science, University of California, Santa Barbara, CA 93106, freeston@alexandria.ucsb.edu

This research addresses a fundamental information infrastructure issue as it relates to climate modeling: the interoperability of geospatial data sets on a global scale, and their analysis using a globally distributed computational network. The focus of the research is on advances in the computational support for hexagonal discrete global grids (DGGs), and the rationale for their adoption.

It might be asked why there is a need to impose a grid of any kind on the Earth's surface, if the application merely needs to record the location (lat/long) of a global set of geospatial objects. At the application modeling level there may well be no such need, but it is a basic requirement at the computational modeling level. It comes down to the efficiency of sorting, searching and dynamic data organization. Ultimately, in order to achieve the twin objectives of logarithmic access time and a guaranteed minimum overall memory occupancy, data must be stored in indexed buckets (which in practice are usually disk blocks). And efficient indexing requires that there is a direct mapping from a geospatial reference (lat/long) to a specific bucket. But all these requirements cannot be simultaneously satisfied with a one-to-one mapping between indexed buckets and a disjoint set of lat/long grid partitions (most familiarly, the grid of squares defined by equal increments of latitude and longitude on a Mercator projection).

Applications involving modeling of dynamic flows over the Earth's surface or upper atmosphere impose the further requirement that grid partitions are of equal area and as far as possible have the same shape. And their adjacency properties with directly neighboring partitions should be the same for all of them. These requirements cannot be satisfied by any rectangular grid, and demand direct computation on the surface of the spheroid.

Over the past ten years this has led a small group of mathematicians and geographers to develop promising new spherical data models based on DGGs. But they have run into performance problems because of the limitations of the indexing techniques currently available. Yet at the same time, within the spatial database research community, there have been significant advances in multidimensional indexing techniques with improved performance, and with properties which appear to be ideally suited to the support of such DGGs. This poster presents the basic concepts involved, and emphasize the advantages of a canonical global grid structure in facilitating computations between global data sets.

[Poster PDF]