Climate Change Research Section Narrative
The Climate Change Research Section (CCR) is part of the Climate and Global Dynamics
(CGD) Division at the National Center for Atmospheric Research (NCAR) in Boulder,
Simulating the Climate of the Late Permian
Jeffrey Kiehl (CCR) and Christine Shields (CCR) have been researching one of the more intriguing time periods in Earths history, the boundary between the Late Permian and Triassic periods at 251 Ma (1). This boundary marks the largest extinction recorded in Earths history, where across this boundary approximately 90% of marine and terrestrial species were lost. Associated with this event was an extended period of magma activity and extended ocean anoxia. What caused such a catastrophic change in life? A number of hypotheses have been proposed to explain various aspects of the extinction and climate of this time period. No one hypothesis can explain all of the paleo information for this period. Although a number of climate model simulations have been carried out for this period, no fully coupled climate model simulation has existed for this time period. Fully coupled climate models are required to accurately simulate the ocean circulation, since using specified ocean boundary conditions highly constrains the climate solution and does not allow for a coupling of atmospheric and land hydrological processes to the ocean, e.g., input of fresh water to the ocean.
Kiehl and Shields are carrying out the first fully coupled climate simulation for this time period. They are using paleogeography conditions and specified CO2 levels for this time period in the recently released CCSM3. The coupled simulation is currently out ~1800 years, where the length of the integration is determined by the long time scales of the deep ocean circulation.
The above figure shows the annual mean surface temperature (°C) from years 1700-1709. The western tropical Panthalassic ocean has a warm pool of water with sea surface temperatures (SSTs) reaching 33 °C, compared to the present day western Pacific warm pool temperatures of 30 °C. The warmest regions over land occur in the subtropical desert regions.
One metric of the climate of this time period comes from the geographic location of evaporite deposits (2). These occur in shallow water regions where the evaporation minus precipitation (E-P) is positive. The above figure shows the annual mean E-P from the CCSM3 simulation, with the location of evaporite deposits marked by red triangles. All but one of these observed evaporite sites matches the model simulation of the region of positive E-P.
An important observation of this time period is the indication that the oceans were anoxic for an extended period (~20 My) (3) of time across the boundary. Low oxygen conditions could explain the limited marine life at this time. One hypothesis suggests that, in a warm greenhouse climate with an associated low pole to equator thermal gradient, ocean mixing to depth would be limited due to strong stratification. Shown below is the meridional overturning circulation for the Late Permian and Present day simulations from the CCSM3. Whereas in the present day world there is strong downward circulation at the poles, this does not exist in the late Permian. The late Permian deep ocean circulation is composed of two weak cells in each hemisphere. The implication of this is that the mixing of tracers such as oxygen to depth is limited for the warm world of the Late Permian. This simulation supports the idea of deep ocean anoxic conditions for this time period.
The present simulation indicates that the strength of the overturning circulation at depth (~3000 m) continues to slow down, indicating an even more sluggish deep ocean circulation than shown in the above figure. This result indicates the need for substantial computational resources to explore the realm of deep time, where deep ocean time scales play a critical role in biogeochemical processes relevant to life on Earth.
Future efforts in exploring the climate of the late Permian include: exploring the role of atmospheric chemical changes at this time including enhanced methane levels, fully coupled dynamic vegetation simulations, and coupling of biogeochemical processes. As shown fully coupled climate models can now be applied to questions of climates in deep time. These types of models can be used to integrate observations and test scientific hypotheses. They can also serve as a catalyst for more synergistic collaborations on key climate issues.
Other Deep-time Applications
Matt Huber and Ryan Sriver (Purdue University), collaborating with Bette Otto-Bliesner (CCR), Esther Brady (CCR), Shields, and Zav Kothavala (CCR), are demonstrating the feasibility of CCSM3 for deep past paleoclimates and theoretical climate dynamics. Benchmark idealized configurations, including a sector world and two islands are being tested. Techniques for accelerating the model spinup are needed due to uncertainties in the ocean state and the long equilibration time of the deep ocean and are being developed and validated as part of this collaboration.
This figure shows idealized (top) and actual (bottom) "deep-time" continental distributions for a "sector world" (left) and "two islands" (right). It has been conjectured that the climate dynamics of these time intervals was largely a function of the continental geometries.
Huber, Rodrigo Caballero (University of Chicago), Otto-Bliesner, and Brady are analyzing CSM1.4 simulations for Present, Last Glacial Maximum, Eocene, and Cretaceous in terms of tropical and high-latitude modes of variability. These periods vary in terms of their paleogeography, topography, continental ice sheets, and atmospheric trace gas concentrations. In contrast to conventional wisdom, El Niño Southern Oscillation (ENSO) is a robust feature when a Pacific-like Ocean is present and gains preeminence in warm climates (Eocene and Cretaceous) when the equator-to-pole temperature gradient is reduced. There are records of ENSO going back to the Eocene.
Climate of the last 150,000 Years
Otto-Bliesner (CCR), collaborating with Jonathan Overpeck (University Arizona), Shawn Marshall (University of Calgary), and Giff Miller (University of Colorado), studied the sensitivity of the Arctic climate and the Greenland Ice Sheet to summer warming at the beginning of the Last Interglacial (130,000 years before present) associated with late spring-early summer insolation anomalies associated with Milankovitch orbital forcing. This study used CCSM2 simulations in combination with a Greenland Ice Sheet model and terrestrial, marine, and ice core proxy records. CCSM2 seasonally warms the Arctic by up to 5°C, and forced with CCSM2 predicted surface temperature and precipitation changes, the Greenland Ice Sheet shrinks to half its present volume in 3500 years. Vegetation and ice sheet feedbacks still need to be incorporated and should indicate enhanced sensitivity.
This figure shows the extent and thickness of the Greenland Ice Sheet after 5000 years when forced with CCSM predicted surface temperatures and precipitation for the Last Interglacial (LIG). Inset shows the time evolution of Greenland ice volume for present-day (blue line), the LIG (black line), and a future scenario of 4xCO2 (red line) when forced with CCSM predictions.
Otto-Bliesner and Bob Tomas (CCR) are studying the mechanisms underlying decreased aridity in northern Africa in the Holocene. CCSM2 simulations for 8500 yr BP show that Milankovitch orbital variations in solar radiation for this past period enhance the North African monsoon with a more northern position of the Intertropical Convergence Zone (ITCZ), a stronger Tropical Easterly Jet, a displacement north of the African Easterly Jet, and stronger low-level westerlies at low latitudes. These dynamical changes over North Africa are analogous to conditions observed at present during wet years in the Sahel and sub-Saharan region.
This figure shows the time-average June-July-August zonal wind (m/s) over North Africa predicted by CCSM for present (left) and 8500 years before present (right). The CCSM simulates important features of present summer circulations, including equatorial westerlies near the surface and two easterly jets, the African Easterly Jet at 600-700 mb and the Tropical Easterly Jet at 150 mb.
As part of Paleoclimate Modeling Intercomparison Project Two (PMIP-2), CCSM3 at T42x1 resolution has been integrated for Preindustrial, Mid-Holocene, and Last Glacial Maximum forcings. Otto-Bliesner, Brady, Tomas, Sam Levis (TSS), Kothavala, and Gabriel Clauzet (Universidade de Sao Paulo, Brazil) are analyzing these simulations as part of the PMIP-2 project. Results will first appear in a special issue of Journal of Climate devoted to CCSM3. The results for the Last Glacial Maximum show a global surface cooling of 4.5°C compared to Preindustrial conditions and expanded sea ice in both hemispheres, a much colder and saltier deep ocean, and a shallower and weaker North Atlantic overturning cell.
This figures shows the Atlantic Ocean SST for the Last Glacial Maximum (LGM) as predicted by CCSM3 (red line) and estimated by the Climate Mapping, Analysis, and Prediction (CLIMAP) (dashed line), Mix et al. (pink circles), and Glacial Atlantic Ocean Mapping (GLAMAP) (blue circles) proxy reconstructions.
The slab ocean version of CCSM3 provides a useful tool to assess climate sensitivity for a variety of forcings. Otto-Bliesner, Brady, and Kothavala are comparing the response of this model to past and future forcings.
Otto-Bliesner participated with a group of researchers from 22 U.S. and international universities and laboratories to trace the spatio-temporal pattern of peak Holocene warmth (Holocene thermal maximum, HTM) across the Western Hemisphere of the Arctic. This collaboration is part of the Paleoenvironmental Arctic Sciences (PARCS) research initiative sponsored by the National Science Foundation (NSF). At the 16 terrestrial sites where quantitative estimates have been obtained, Holocene Thermal Maximum (HTM) temperatures (primarily summer estimates) were on average 1.6±0.8°C warmer than present (approximate average of the 20th century). The warming, though, was time-transgressive over the western Arctic with warming in Alaska and northwest Canada coincident with precession-driven summer insolation anomaly, but delayed warming by about 4000 years in northeast Canada linked to the lingering Laurentide Ice Sheet.
PMIP-1 compared the response of 18 Atmospheric General Circulation Models (AGCMs) to mid-Holocene (6000 years before present) solar forcing changes associated with Milankovitch orbital cycles. As a precursor to PMIP-2, seven modeling groups (NCAR, Hadley Centre, Laboratoire des Sciences du Climat et de l'Environment (LSCE),Meteorological Research Institute (MRI), Royal Netherlands Meteorological Institute (KNMI), Max-Planck, University of Wisconsin) performed AOGCM experiments for this same time period, albeit with varying specification of atmospheric trace gases. Otto-Bliesner collaborated with model intercomparisons of the AOGCMs. The expansion of the area influenced by the Afro-Asian summer monsoon during the mid-Holocene is one of the most striking features shown by paleoenvironmental data and was poorly simulated by the PMIP-1 AGCMs. Results from the Atmospheric Ocean General Circulation Models (AOGCMs) show that ocean feedbacks enhance the African monsoon and shift the belt of maximum precipitation further north. The models also show a northward shift of the northern limit of the boreal forest in response to simulated summer warming in the high-northern latitudes, and a northward expansion of temperate forests in North America in response to simulated winter warming. These northward shifts are supported by paleovegetation data.
Fortunat Joos and Stefan Gerber (University of Bern), Colin Prentice (University of Bristol), Otto-Bliesner, and Paul Valdes (University of Bristol) used time slice simulations of the last 21,000 years from CSM1.4 and the Hadley Centre Unified Model to drive the carbon component of the Bern Carbon Cycle Climate (Bern CC) model. The Holocene ice core record of a 20 ppm rise is matched by the model. Calcite compensation, in response to earlier terrestrial uptake, terrestrial carbon uptake and release, SST changes, and coral reef buildup, explain this increase. The greening of the Sahara, peat buildup, and land use probably influenced atmospheric CO2 by only a few ppm.
This figure shows the cumulative changes in terrestrial carbon inventory predicted by the BERN CC model when forced by CSM (global changes indicated by dotted line) and Hadley Centre Unified Model (latitude bands indicated by shaded areas).
Zhengyu Liu (University of Wisconsin), Otto-Bliesner, John Kutzbach (University of Wisconsin), L. Li (Chinese Academy of Sciences), and Shields compared coupled climate simulations of the NCAR CSM1.4 and the University of Wisconsin Fast Ocean-Atmosphere Model (FOAM) on the evolution of global monsoons in the Holocene. The evolution of the six major summer monsoons, Asian, North African, North American, Australian, South American, and South African, are investigated for the insolation forcing of 3000, 6000, 9000, and 11,000 years before present. In the Northern Hemisphere, the models show an enhancement of the monsoons in the early Holocene and a gradual weakening toward the present, with ocean feedbacks modifying their forced response.
This figure shows monthly evolution of land precipitation anomalies from present predicted by CSM1.4 for 3500 (solid), 6000 (dot), 8500 (dash-dot), and 11,000 (dash) years before present for the Northern Hemisphere monsoons.
Clauzet, Ilana Wainer (Universidade de Sao Paulo, Brazil), Alban Lazar (Universidade de Sao Paulo, Brazil), Brady, and Otto-Bliesner examined the CSM1.4 simulation of the South Atlantic at the Last Glacial Maximum. Changes of mass transport within this basin are associated shifts in circulation features.
Carrie Morrill (CCR) is creating a database of high-resolution paleoclimate records for studying abrupt climate change during the last deglacial and the Holocene. These records have been selected from the large number of published paleoclimate records for their robust age models and well-supported climatic interpretations. Thus far, over 120 records have been compiled. Morrill is using these records to understand the timing, spatial patterns, and possible causes of several past abrupt climate changes. These include events in the mid-Holocene (~4000-6000 years ago) and early Holocene (~8200 years ago). Abrupt climate changes are identified in these records using a variety of non-parametric tests to identify discontinuities in the paleoclimate time-series. In collaboration with Claudia Tebaldi (Research Application Program), Morrill is further developing and refining these methods.
For the mid-Holocene, Morrill has found several periods of preferred change centered on ~4200 and ~5600 years ago. Abrupt changes at these times occurred over decades to centuries, and were associated with the transition from the Holocene climatic optimum to the Neoglacial during which temperature decreased in the middle to high latitudes and precipitation decreased in the tropics. The event at ~4200 years ago was also associated with archaeological evidence for the collapse of several ancient civilizations, including the Akkadian in Mesopotamia and the Indus in India and Pakistan.
Morrill, in collaboration with Bob Jacobsen (William and Mary), is constructing global maps of climate anomalies during the 8.2K event. These will be used for comparison with output from experiments planned for CCSM3 in collaboration with others in CCR and Oceanography. These experiments will examine the climatic impact of freshwater pulses to the North Atlantic that are similar to the glacial meltwater flood that caused the 8.2K abrupt climate change. The last synthesis of paleoclimate records spanning this event was completed in 1997 and, since then, there have been dozens of new records relevant to this subject published. Morrill and Jacobsen have identified several patterns that will be of use in model-data comparison. These include a temperature decrease of 1-2 ºC over Europe, regional drying in the tropics, and little evidence for change in the Southern Hemisphere.
Morrill, Otto-Bliesner, Caspar Ammann (CCR), and Brady are working with Dwight Owens and Dolores Kiessling of COMET to create a web-based learning module for undergraduates. The module explains how changes in the earths orbit affect insolation seasonally and latitudinally and, in turn, how this drives glacial-interglacial cycles. The module contains a number of interactive exercises and animations. These can be used by instructors for classroom demonstrations and homework or lab exercises.
Morrill contributed a review of the mechanisms of abrupt climate change for the NOAA Paleoclimatology Programs Paleo Perspective on Abrupt Climate Change website www.ncdc.noaa.gov/paleo/abrupt/index.html. This website reviews the state of our current knowledge about the causes and effects of past abrupt changes.
This figure shows the location of paleoclimate proxy records that have been compiled from the published literature to study abrupt climate changes during the Holocene.
Natural Climate Variability During the Period of Annually Resolved Records
Ammann and Tomas, supported by the Weather and Climate Assessment Science Initiative of Linda Mearns (Environmental and Societal Impacts Program), use high-resolution paleoclimate proxy records in an attempt to extend the limited (~150 years) instrumental record. Prior centuries and millennia have exhibited significant natural climate variations that had a tremendous impact on the environment and societies around the world. Such records offers an ideal and highly relevant test bed for scientific evaluation of both empirical proxy-based research, as well as for testing the sensitivity and validity of global climate models normally applied for future climate simulations. Each of the approaches provides specific insights into the functioning of the climate system, but each also has its clear weaknesses. It is the goal of the research to work at the intersection of state-of-the-art climate models and the proxy-community in order to contribute to a better understanding of the range and causes of natural climate variability and to clarify natures role in climate today and possibly in the future.
Climate Modeling Using NCAR-CSM 1.4 and NCAR-CCSM 3
After implementation of natural external forcing factors into the standard coupled model (explosive volcanism and solar irradiance changes), Ammann and collaborators (in particular Fortunat Joos) performed numerous historical transient simulations and short sensitivity studies of the recent past. A core element is the detection and attribution of climate changes to external forcing. Ammann studies ways to detect this impact in collaboration with statisticians and their students (Philippe Naveau, University of Colorado; H-S. Oh, University of Alberta, Canada). The group developed multi-resolution techniques (based on discrete wavelet transforms and empirical mode decomposition methods) and state-space models (using modified Kalman filtering techniques taking skewed distributions into account) that are adapted to the typical signals of smooth solar irradiance changes and volcanic spikes, respectively. The detection work focused first on single time series, but has been expanded to large gridded datasets from model output. A clear solar influence has been found that is stronger than previously assumed. Because the same temporal fingerprint is also present in proxy reconstructions, the conclusions were drawn that the sun does affect climate on the century time scale (see Figure below). However, the same simulations also show that smaller rather than larger solar irradiance changes in the past are more consistent with the proxy based climate records, and independent of the possible magnitude of past solar changes, the late 20th century warming cannot be attributed to solar irradiance changes.
This figure shows wavelet decomposition of forcing (top: blue) and climate response as seen in CSM 1.4 Millennium simulation and a set of proxy reconstructions. The time scale is focusing on the dominant ~200-year deVries cycle.
The analysis is not restricted to large-scale Energy Balance questions, but the AO-GCMs are tested for specific response in the dynamics to the imposed radiative forcings. The proxy record is hinting at the possibility that the response of several climate modes might potentially be more systematic than previously thought. In such a case, and if the models should be able to reproduce a corresponding signature, this would tremendously enhance the predictive skills of climate models on regional scales. Collaborative work is looking into modes such as the North Atlantic Oscillation (NAO), ENSO and the Pacific Decadal Oscillation (PDO). Additional analysis is evaluating the monsoon response to the radiative forcing.
Ammann worked with Andrew Conley (CMS) on a 10-member ensemble simulation of the Mt. Pinatubo eruption (June 1991) with the coupled NCAR CCSM 3 model. Radiative fluxes and surface climate changes are studied. The volcanic aerosol is now a standard implementation in CCSM 3.
Proxy Climate Reconstructions: Methods, Uncertainty, and Optimization
Proxy-based climate reconstructions depend on a short calibration period using the assumption that the link to the climate parameter and often also to teleconnected modes of variability is stable over time. Models provide means to test these assumptions. Ammann and Eugene Wahl (Alfred University) evaluated the reliability of various climate reconstruction techniques using the fully forced coupled climate model output. Through this process, they help to optimize reconstruction efforts and determine crucial spatial gaps in the available proxy records.
This figure shows a test of single point stationarity: Artificial Central Chile Tree-Ring series of the past millennium based on subsampling of fully forced CSM 1.4 simulation. White noise was added to the precipitation series to match real world tree-ring to El Niño correlation.
On the forcing side, Ammann is working with a group at Rutgers University (under guidance of Alan Robock) on a new reconstruction of volcanic aerosol based on the largest collection of ice core data over the past millennia. The global ice coring community actively supports the work. In parallel, Ammann is collaborating on statistical modeling of temporal forcing distributions (primary solar frequencies, volcanic forcing as extreme value series). Such detailed characterizations of the forcing (see Figure below) provide a quantitative base for signal detection in noisy climate reconstructions. Solar forcing is approached through collaborations with proxy and modeling partners to establish a process understanding and detection framework to evaluate which dynamical pathways are most consistent in explaining climatic signals extracted from the proxy reconstructions.
Tropical climate variability response to forcing: Previously regarded as most independent mode, ENSO variations over the recent past seem to contain significant externally forced response behavior: cooling in E equatorial Pacific during increase solar irradiance, but higher occurrence of warming during low-solar irradiance as well as after large volcanic eruptions.
Europe: Evaluation of dominating circulation regimes isolated in historical reconstructions and model. NAO and blocking cases with comparable behavior in control and low-solar case. No blocking was found in high-solar situation because of a too strong polar vortex and volcanic signal in circulation over Europe.
E-Asia (China): three selected time windows around significant volcanic or solar forcing periods as well as the drought episode that brought the Ming Dynasty to its knees. General circulation model (GCM) (NCAR-CSM and Goddard Institute for Space Studies (GISS)) analysis and Regional Modeling using GCM boundary conditions with MM5.
Centennial scale ocean anomalies: A strong ~125-year oscillation in the N-Atlantic ocean was found. Since no similar oscillations are known in the real world (except of about half the frequency), significant effort was put in to understand this variation. The impact is far reaching in atmosphere and ocean, including the thermohaline overturning.
Future Climate Change:
The Climate Change and Prediction (CCP) group consisting of Warren Washington (CCR), Gerald Meehl (CCR), Lawrence Buja (CCR), Aixue Hu (CCR), Vincent Wayland (CCR), Haiyan Teng (CCR), Gary Strand (CCR), David Lawrence (CCR), and Julie Arblaster (Bureau of Meteorology Research Centre) is one of the primary components of the joint NSF/DOE climate modeling research program. With a strong focus on climate change over decades to centuries, the CCP group continues to carry out centuries long climate simulations and ensembles addressing all five of the key global change issues identified by the Climate Change Science Program (CCSP) and the U.S. Global Change Research Program (USGCRP):
In June of 2004, the CCSM version 3 (CCSM3) became available for users. Since that time we have carried out over 5000 years of high-resolution climate change simulations with the new CCSM3 that have included the major forcings on the climate system such as ozone, land cover, solar variability, volcanoes, sulfate aerosols, and greenhouse gases for the Intergovernmental Panel on Climate Change (IPCC) simulations with climate forcings for the 20th, 21st, and 22nd centuries. In addition, climate simulations with the Parallel Climate Model (PCM) continue, adding to the already existing large archive of simulations. It is expected that the CCSM3 and PCM will provide the largest set of IPCC simulations of 20th, 21st and 22nd century climate for the IPCC Assessment Review 4 (AR4) of any group in the world. We have worked with the Program for Climate Model Diagnosis and Intercomparison (PCMDI) (see below) and used the DOE Earth System Grid (ESG) capability to make the data readily available to the broader climate research community.
Collaborations with Other Organizations
The CCP climate research program supports the DOE comprehensive climate modeling research program through climate model diagnosis and prediction of climate change from increasing greenhouse gas concentrations and other forcings of the climate system. This program relies on distributed research collaborations spanning NCAR, the Naval Postgraduate School (NPS), the Los Alamos National Laboratory (LANL), the Geophysical Fluid Dynamics Laboratory (GFDL), National Energy Research Scientific Computing Center (NERSC), Oak Ridge National Laboratory (ORNL), Argonne National Laboratory, PCMDI and several universities. This project will complement the National Oceanic and Atmospheric Administration's (NOAA) climate modeling program emphasizing decade-to-century climate prediction. The program is also tightly linked to DOE's component of the federal High Performance Computing and Communications Program through the High Performance Computing Research Centers at LANL and ORNL. The latter also has the Center for Computational Sciences and the National Leadership Computing Facility.
In 1998 at the National Workshop on Advanced Scientific Computation, the plans of DOE and the NSF were discussed. The DOE, NSF, and NASA have agreed to continue to collaborate on climate model development ,with emphasis on redesigning of the climate model components and the method of coupling components for newer generation supercomputers. This national effort has been transformed by each agency into computing initiatives that have led to increased computer capability. During the last few years, the NCAR DOE-supported research effort played a key role in the DOE Accelerated Climate Prediction Initiative (ACPI) Demonstration Project. This project has successfully concluded and the results were published this year. Our involvement in the DOE Climate Change Prediction Program (CCPP) Avant Garde project has resulted in the next generation flux coupler and improved aspects of the atmospheric model. The next generation flux coupler is now in the fully operational phase. The federal government has developed the CCSP which stresses reducing uncertainties. Special emphasis will be on improving climate models, as well as improved understandings of aerosols, regional climate change, and the impacts of climate change.
The coupled atmosphere, ocean, land/vegetation, and sea ice PCM has been in production at a number of centers for several years and is carrying out ensembles of climate change simulations including those for the IPCC AR4. The PCM continues to be used for climate change simulations with various climate forcings such as land cover and carbon aerosol forcings, and at the same time there has been a transition to the newly released CCSM version 3. The PCM is not undergoing any new development. However, because of the large number of existing PCM ensemble simulations, the PCM will continue to be used for climate sensitivity analysis. For example, two recent papers by Ben Santer of PCMDI along with DOE-funded collaborators Washington, Meehl ,and Arblaster have been published in Science. Both used PCM simulations to show that an increase in tropopause height and a warming of the troposphere have been mainly due to increases of greenhouse gases. These papers have generated a great deal of attention, and they indicate an additional measure of global warming other than surface air temperature. Another example is a recent paper in Science by Meehl and collaborator Tebaldi from RAP/ESIG. They analyzed PCM multi-member ensemble simulations of 20th and 21st century climate to show that heat waves will become more intense, more frequent and longer lasting in the 21st century due to an increase of greenhouse gases in the model. To support future PCM work such as this, the PCM is currently being ported to the Cray-X1 and Linux platforms.
Both the NSF/CCSM Climate Change Working Group (CCWG) and the DOE CCPP are particularly interested in regional aspects of climate change. We have seen that by increasing the horizontal resolution with the uncoupled version of the Community Atmosphere Model (CAM) to T85 (developed by the Climate Modeling Section (CMS) and the CCSM Atmosphere Model Working Group), there has been substantial improvement in the regional aspects of precipitation especially in mountainous areas. This is particularly important for the western North American regions. Also, there is improvement in the surface winds in the Arctic Ocean region, which in turn improve the sea ice distribution. Even though there have been substantial improvements, there are some model biases that are not dependent on resolution. Working on such biases will be a high priority for CCSM in the next year and this Cooperative Agreement will contribute to the research involved in addressing them.
Since the CCSM3 release in the spring of 2004, we have worked with many CCSM scientists and software engineers preparing the climate forcing datasets for the twentieth century. We also worked with others to develop the future climate forcing scenario datasets for the twenty-first century. As noted above, it is expected CCSM will be one of the major contributors to the upcoming IPCC AR4. We also have performed the same scenario simulations with the PCM, which will allow for an intercomparison with same forcing between two different models. We are in the process of completing the simulations by November 2004.
The PCM continues to be used by researchers for additional simulations run on many different parallel supercomputers to further leverage existing climate simulations with greenhouse gases, sulfate aerosols, ozone, land cover, carbon aerosol, solar, and volcanic forcings. Most of the simulations for each forcing have an ensemble of four to five members. There are more than 100 simulations of a century or longer with a 1500 year control simulation for 1870 conditions. We believe this is the largest set of freely available climate forcing experiments ever completed by any modeling group, and is a significant achievement as this represents an unprecedented dataset to study climate variability and climate change. The data from the ensemble simulations are being archived at PCMDI and NCAR with total size of approximately 1.6 million files and 143 Tbytes. The data are freely available to researchers via the Internet by the use of the DOE/ SCIDAC sponsored Earth System Grid or by contacting PCMDI. We have already added over 100 Tbytes from the CCSM3 IPCC simulations to the archive.
Description of IPCC Simulation Production Process
This production effort represents a multi-institutional distributed effort by scientists and software engineers at NCAR, ORNL, NERSC, and the Lawrence Berkeley National Laboratory (LBNL).
This figure shows IPCC CO2 concentrations. In addition to CO2, the simulations are run with solar, volcanoes, sulfate aerosols, ozone, and other greenhouse gases for the period 1870-2000. A spread of CO2 concentrations for future forcing scenarios are used from years 2000-2200 that also include sulfate aerosols and ozone.
The IPCC scenarios that are being run can be categorized into two groups: a set of "commitment scenarios" and a number of "science scenarios" that explore alternative future emissions outcomes. The IPCC SRES science scenarios that were recommended to the global coupled climate modeling groups by the IPCC WG1 include SRES A2, SRES A1B, SRES B1. The A2, A1B, and B1 scenarios consist of five-member ensembles run from the year 2000 to the year 2100. Then, commitment scenarios are run with the concentrations of all atmospheric constituents in A1B and B1 held constant at year 2100 values, and the model continues to the year 2200 with five member ensembles. One of the ensemble members from each is continued for an additional 100 years to the year 2300 with constant year 2100 concentrations. There is also a 20th century commitment scenario that freezes concentrations at levels observed in the year 2000, and the model is run to the year 2100. For the first time, these runs are being carried out with the high-resolution T85 (roughly 140 km latitude-longitude resolution) version of the new CCSM3.
This figures shows the CCSM3 IPCC Run Plan: The IPCC scenario runs are being run at NSF/NCAR (blue), DOE/ORNL (green), and DOE/NERSC (red). Parallel runs on the Earth Simulator (pink) were made with the scientific assistance of the NCAR program.
The PCM and the CCSM3 IPCC runs are being performed at the NSF/NCAR, DOE/ORNL and DOE/NERSC supercomputing sites. (We recently hired Haiyan Teng through supplemental funds to assist in running the experiments and diagnosing the results. An additional set of parallel runs, using the same version of CCSM3 with forcing datasets prepared by the NCAR program, was run on the Japanese Earth Simulator (ES). Focused model development support from the ORNL and LANL science and software engineering teams, significant dedicated computing resources at NCAR and ORNL, a large computing allocation at NERSC, and 24 hour/7 day run monitoring/ support by NCAR staff have kept the IPCC runs on track for completion in time to meet IPCC data submission and paper publication deadlines Most recently, NERSC consultants assisted the NCAR team to run the NERSC A2 5 member ensemble as one single 1020 processor massively-parallel run.
This figure shows PCM IPCC simulations for the AR4: Globally averaged surface air temperature anomalies from the BAU, Stab, A1F1, B1, A1B, B2, and A2 scenarios are shown along with simulation in which the 20th century greenhouse gases are frozen at year 2000 values (20thC freeze), as well as simulations where A1B and B1 concentrations are frozen at year 2100 values and the model is run to the year 2200. The shading shows the range of ensemble members and the heavy line is the mean of the ensemble. Note for the 20th century freeze simulation surface temperature warms by 0.3-0.4 0 C.
The data products from the PCM and the CCSM3 IPCC simulations will be used for climate change impact studies and as boundary conditions to drive regional-scale models. To support DOE and NSF regional modeling efforts, we are outputting two additional subdaily data streams for each of the CCSM IPCC scenarios. By design, the raw and post processed data products from the different IPCC scenarios are being kept at the DOE and NSF sites where they were computed. This large distributed dataset is being freely served to the U.S. climate research and education community via the DOE Earth System Grid (ESG).
This figure shows CCSM3 DOE/NSF IPCC scenario runs as of 4 October 2004: Observed forcings (solar, volcanoes, greenhouse gases, sulfate aerosols, carbon aerosols, and ozone) are used during the historical period, from years 1870-2000. A variety of future forcing scenarios (20th Century freeze, B1, A1B, and A2) are used from years 2000-2200 to simulate the most likely range of future climates. Two of the commitment scenarios (A1B and B1) will be executed out to year 2300.
The results of this project will be submitted as the U.S./DOE contribution for the IPCC AR4. The integrated picture formed by these scenarios will be the basis for long-term energy and resource use policies. By the use of higher resolution studies, we are providing data to the climate impacts community that will be more useful than previous studies. Performing simulations in the 2003-2005 timeframe is a high priority for the DOE and the national effort.
Land-Atmosphere Interaction and the Water Cycle
Global climate models are often employed to investigate whether and how soil moisture anomalies affect weather and climate in the current and in a potentially different future climate. A recent model intercomparison by Lawrence demonstrated that the degree of land-atmosphere interaction varies widely between current state-of-the-art AGCMs. Models with different inherent coupling strengths can lead to vastly different conclusions about climate sensitivity. Land-atmosphere coupling strength, or the extent to which a precipitation-induced soil moisture anomaly influences the overlying atmosphere and thereby the evolution of weather and the generation of precipitation, is reasonably strong in the (CAM2/CLM2) but is very weak in the Hadley Centre modeling system (HadAM3/MOSES2). The notable differences in behavior between these two models may enable the precise mechanisms that control land-atmosphere interactions to be ascertained. Key aspects of the indirect soil moisture-precipitation feedback have been evaluated and compared in both models. While the study is not yet complete, preliminary results indicate that it is relatively unlikely that differences in the land surface schemes can explain the large differences in coupling strength. Instead, it appears that the simulation of the diurnal cycle, particularly the simulation of boundary layer moist static energy and the relationship between moist static energy and convection, are likely the largest factors contributing the considerable differences in coupling strength. In CAM2/CLM2, moist static energy builds differently during the day over moist and dry soils, which leads to heavier precipitation over wet soils. In contrast in HadAM3/MOSES2, soil moisture exerts virtually no control on moist static energy in the boundary layer and henceforth convection is essentially unaffected by the soil moisture characteristics. Research aimed to identify how differences in the boundary layer and convection schemes between the two climate models affects the diurnal buildup of moist static energy is ongoing.
Climate Variability Associated with Atlantic Thermohaline Circulation
Aiguo Dai (CAS) has analyzed a 1200-yr unforced control run and climate change simulations under projected greenhouse gas forcing during the next two centuries for changes in Atlantic Ocean circulations as simulated by the PCM. The strength of the Atlantic thermohaline circulation (THC) shows large variability (amplitudes » 1.5-3.0 Sv) at 15-40 year time scales, with a sharp peak of power around 24 years in the control run. Associated with the THC oscillation, there are large variations in the North Atlantic SST, sea surface salinity (SSS), sea-ice fraction, and net surface water and energy fluxes, all of which lag the THC by 2-3 years. However, the net effect of the SST and SSS variations on upper ocean density in the midlatitude North Atlantic leads the THC by ~6 years or 90o in phase. This density vs. THC phase lag provides a direct mechanism for the 24-year cycle to oscillate through density's effect on North Atlantic deep water formation. The PCM-simulated SST and sea-ice variations resemble observed spatial patterns associated with the NAO. This suggests that the THC multi-decadal variability is coupled with the NAO in the model. Forced by projected CO2 and other trace gases for the next two centuries, the PCM shows significant weakening (by ~12%) of the THC during the 21st century, even after removing the un-forced THC drift during the period. In the 22nd century, the THC continues to weaken (by additional ~10%) if atmospheric CO2 keeps rising, but stabilizes if the trace gases level off.
Southern Annular Mode Trends
Arblaster and Meehl have investigated the trend in the Southern Annular Mode during recent decades that has involved an intensification of the polar vortex. The source of this trend is a matter of scientific debate with stratospheric ozone losses, greenhouses gas increases, and natural variability all possible contenders. Since it is difficult to separate the contribution of various forcings to the observed trend, the PCM is utilized here to isolate the responses of the climate system to single forcings. Ensembles of 20th Century simulations forced with the observed time series of greenhouse gases, tropospheric and stratospheric ozone, sulfate aerosols, volcanic aerosols, solar variability, and various combinations of these are used to examine the annular mode trends in comparison to observations. We find that ozone changes are the biggest contributor to the observed summertime intensification of the southern polar vortex in the second half of the 20th century, with increases of greenhouse gases also a necessary factor to reproduce the observed trends. Although stratospheric ozone losses are expected to stabilize and eventually recover to pre-industrial levels over the course of the 21st Century, these results suggest that the observed trends in atmospheric circulation will continue into the future.
Development of the Earth System Grid (ESG)
During the last year, the ESG was taken into full production, serving as the primary online data distribution mechanism for the DOE/NCAR PCM and CCSM runs. The extensive PCM climate experiment collection, the new CCSM3 control runs, as well as the initial results from the IPCC runs with CCSM3, are now being served to the US and international scientific community via the ESG. Gary Strand was instrumental in moving the ESG from a pilot project to a production facility, working with the ESG development team, building the initial metadata catalogs, and supervising the large-scale production of CCSM and PCM ESG metadata catalogs.
Effects of Land Cover Changes as Part of the IPCC Assessment
Johannes Feddema (University of Kansas) used the PCM to assess the sensitivity of a fully coupled climate model to changes in land cover. By comparing three simulations representing present day land cover from different data sources, we conclude that there is significant model sensitivity to land cover characterization, with an observed average global temperature range of 0.21K between the simulations. These results show that careful consideration needs to be made of the land cover boundary conditions used in land cover change experiments.
20th and 21st Century Climate
Collaborative work by Meehl, Dai, Washington, Tom Wigley (CAS), Arblaster, and Ping Liu (International Pacific Research Center, University of Hawaii) examined the influences of various anthropogenic forcings over the 20th and 21st centuries in the PCM. These experiments show that the response of globally averaged temperature in the model to individual forcings is mostly additive, and that only by the addition of anthropogenic forcing mainly from greenhouse gases can the model produce the large warming observed in the second half of the 20th century. They also studied the effects of increasing CO2 emissions on the magnitude of future global warming and the effects of global warming on the extent of the Sahara desert. The latter study showed that the PCM, along with several other models in the Coupled Model Intercomparison Project (CMIP), reproduced a decreasing rainfall trend during the 20th century in the Sahara region as observed, and in future climate the Sahara becomes smaller, moves north and west, and continues to dry. Meehl and Washington collaborated with Ammann and others to document the effects of a new volcanic forcing dataset now being used in the PCM and CCSM3 20th century simulations.
Meehl and Arblaster completed a study on the processes that affect changes in precipitation in the Indian monsoon. They used the PCM to document increases in mean monsoon rainfall and enhanced interannual variability with increased CO2. Regional increases in Indian Ocean SSTs played a role in these changes, but influences from the tropical Pacific were even more important for future increases of interannual variability in the monsoon.
Ben Santer (LLNL) collaborated with a team of scientists, including DOE funded NCAR scientists Meehl, Washington, and Arblaster, and used the PCM to show that trends in the lower tropospheric temperatures from satellite data were consistent with a global coupled model that included known anthropogenic forcings for the period of the Microwave Sounding Unit (MSU) data (including aerosols from Mt. Pinatubo). The model and observed data analyses indicate that tropopause height is a valid new index to monitor global warming. They also addressed the issue of large El Niño events and their influence on trends in short data records such as the MSU, where the 1997-98 El Niño event at the end of the record induced a warming trend that was absent prior to that event. They showed that the models with El Niño amplitudes comparable to observations can produce trends nearly as large due to sampling of 20-year records similar to the observations.
There is ongoing collaborative work at NCAR by Meehl with Tebaldi and Doug Nychka (GSP) to analyze changes in variability and extremes in ensembles of future climate projections. This work is using statistical analyses of extremes, as well as threshold methods to study changes in weather and climate extremes in the PCM. They have shown that the change in base state atmospheric circulation, due to the increase of greenhouse gases in the PCM, affects the pattern of future changes in frost days (nighttime minimum temperatures below freezing) and heat waves. For example, frost days decrease most over western North America due to an anomalous ridge of high pressure over that region that is caused by changes in tropical SSTs and convective heat sources. Heat waves become more intense over the western and southern U.S. in the 21st century due to positive 500 hPa height anomalies caused by the increase of greenhouse gases. These circulation anomalies are again related to changes in tropical SSTs and convective heating anomalies.
Meehl with Washington, Wigley, Arblaster, and Dai have analyzed the solar forcing ensemble runs with the PCM for 20th century climate compared to the ensemble runs without solar forcing. Results show that the observed early century warming is only simulated when solar forcing is included. A main difference in climate system response in the solar runs compared to the runs without solar forcing is that tropical precipitation regimes are more intense due to early century solar forcing compared to late century greenhouse gas forcing, suggesting for the first time that solar forcing could affect the climate system in some regions through coupled ocean-atmosphere interaction and consequent precipitation processes.
Dai, Hu, Meehl, Washington, and Strand analyzed the ensemble simulations of the 20th and 21st century climate by the PCM under greenhouse gas and sulfate aerosol forcings. Over the midlatitude North Atlantic Ocean, the model produces a moderate surface cooling (1-2°C, mostly in winter) over the 21st century. This cooling is accompanied by changes in atmospheric lapse rates over the region (i.e., larger warming in the free troposphere than at the surface), which stabilizes the surface ocean. The resultant reduction in local oceanic convection contributes to a 20% slowdown in THC. Dai and collaborators also documented results from the ACPI pilot project, where the PCM was used to force a regional model, as well as streamflow models, to examine the effects of global warming on water resources in California.
Arblaster and Meehl collaborated with David Karoly (University of Oklahoma) and a team of scientists to use a set of simple indices in a detection/attribution study for North America. The PCM was one of a set of models used in this study to show that humans have indeed influenced the climate over North America.
Meehl collaborated with Harry van Loon and Ralph Milliff (Colorado Research Associates) and addressed onset of warm events in the Southern Oscillation in the early 1990s. Two aspects of the onset of a warm event in the Southern Oscillation are 1) the subtropical South Pacific High is weakened in early southern winter which weakens the trade winds and reduces upwelling along the equator, and 2) in the months that follow, and particularly during the mature phase of the warm event during southern summer, the negative SST anomalies in the equatorial eastern Pacific cold-water tongue are displaced by positive anomalies. Based on these criteria, Meehl and collaborators identified two warm event onsets that occurred during the early 1990s, in 1991 and 1994. This result refutes earlier work that the early 1990s may have been one long ENSO event and has implications for understanding and interpreting anthropogenic influences on ENSO in climate change simulations.
van Loon also collaborated with Meehl and Arblaster to follow up on the analysis of solar forcing in the PCM, but to look for similar signals in satellite and reanalysis data. Results show that comparable processes are at work in the observations and model in that greater solar forcing enhances climatological patterns of precipitation in the tropics, resulting in higher amplitude precipitation over the monsoons and ocean convergence zones.
Meehl and Arblaster worked with Johannes Loschnigg (University of Hawaii) and a team of scientists to study recent variations in Indian Ocean circulation related to interannual variability associated with the Tropospheric Biennial Oscillation (TBO). Results showed that the Indian Ocean plays an active role in monsoon and tropical Pacific interannual variability that is important to understand for interpreting the processes that contribute to 20th century climate variability.
Meehl worked with Duane Waliser (State University of New York) and a team of scientists to document intraseasonal variability in global coupled climate models including PCM. They showed that such variability is generally weak in current models, but that ocean coupling seems to improve the simulations of intraseasonal variability somewhat.
Meehl collaborated with Washington and Arblaster to analyze four recent NCAR global coupled models (PCM, CSM, PCTM, and CCSM) to show that the relevant feedbacks for climate system response to increasing CO2, (ice/albedo, water vapor, and clouds) are managed by the atmospheric model. The ocean, sea ice, and land surface play secondary roles for the globally averaged response. Two models with identical atmospheres but different ocean and sea ice components (PCM and PCTM) have the most similar response to increasing CO2, followed closely by the CSM with comparable atmosphere and different ocean and sea ice. The model that is most dissimilar, the CCSM, has a different response from either of the other three, and in particular is different from PCTM in spite of very similar ocean and sea ice but different atmospheric model components. The role of ocean heat uptake is also crucial to understand the response of the respective models, and depends on the THC in the Atlantic Ocean.
Porting of the PCM on to the CRAY and Linux Computer System
Although the PCM is not being developed further there are still several science groups that are interested in performing simulations. The reason for the continued use is that we have an extraordinary archive of previous simulations with various climate forcings. For example we are still exploring different solar forcing, carbon aerosol, and land cover changes for researchers at several universities. Since the model is still in use, Vince Wayland is making the model available on the CRAY and Linux machines. The status is that the PCM has ported to the Cray X1 this past year; however, there are still software problems to be solved. Preliminary performance results showed good scaling on 4 to 16 processors without any Cray specific additional optimization (vectorization). Additional work will be done to enable the use of up to 64 processors. Some Cray specific optimization may also then be undertaken. The PCM is currently being ported to a Linux cluster named Bangkok in the NCAR CGD environment. This cluster consists of Dell workstations containing Intel chips. As soon as the NCAR SCD IBM Linux cluster, Lightning, is made available to users, the PCM will be ported there also. These ports should provide a general framework for additional Linux ports as they are required.
Examination of Future Abrupt Climate Change
Hu, with Meehl and Weiqing Han (at Program in Ocean and Atmospheric Sciences, University of Colorado), worked on detecting the THC changes from ocean properties using NCARs CCSM2 model, since significant changes of the THC are likely to cause abrupt climate change. Here we intend to find a simple measure to detect changes in THC through examining several factors proposed to control the THC variations using a coupled climate model. These factors are Equatorial-South Atlantic upper ocean temperature, Southern Ocean freshening, inter-basin sea surface salinity contrast, and meridional steric height gradient. Three experiments are analyzed: a present-day control run, a freshwater hosing run, and a 1% CO2 run. Results show that, if freshwater flux is the primary cause, all examined factors can predict the THC changes. If both thermal and haline forcings are involved, only the Atlantic meridional steric height gradient gives a consistent measure of the THC variations. A new result presented here is that the inter-basin sea surface temperature contrast between North Atlantic and North Pacific is found to be an indicator of THC changes.
Understanding Recent Past North Atlantic Ocean Circulation
Hu worked on understanding the changes in North Atlantic Ocean and the relationship with the meridional overturning circulation in the last 50 years of the 20th century using the PCM. Observations show a freshening and cooling sub-polar North Atlantic in the last 50 years of the 20th century. This trend is related to the changes on the NAO pattern. In PCM historical runs, the observed trend in North Atlantic is realistically simulated. However, the model didnt capture the observed NAO trend. By analyzing the changes of the oceanic meridional overturning circulation and the surface heat and freshwater fluxes into the northern North Atlantic in our model, we found that it is the slowing of the meridional overturning circulation that induces a reduced meridional freshwater divergence responsible for the freshening, and a reduced meridional heat convergence is responsible for the cooling. Those surface fluxes and the sea ice volume flux into this region work towards reducing the effect induced by a slower meridional overturning circulation.
Hu also worked on analyzing the THC, sea level, and heat content changes in 20th and 21st centuries from PCM IPCC runs. Preliminary results show that the THC weakens about 1 Sv in the last 25 years of the 20th century. And it stays at roughly the same strength in the 21st century for the 20th century freeze scenario, and by the end of the 21st century; it weakens by 3 to 5 Sv for B1 scenario, 5 to 6 Sv for A1B scenario, and 6 to 8 Sv for A2 scenario. The simulated changes of upper 300-meter ocean heat content in the last 50 years of the 20th century are equivalent to an increase of upper 300-meter ocean temperature by about 0.2 oC, agreeing well with the observations (Levitus et al, 2000). For the projected scenarios, the upper 300-meter ocean temperature will additionally increase 0.2 oC, 0.45oC, 0.9oC, and 1.05 oC for 20th century freeze, B1, A1B, and A2 scenarios, respectively. The upper 3 km ocean heat content changes also agree well with the observations, with about a 0.05oC increase in upper 3 km ocean temperature in the last 50 years of the 20th century. The global mean sea level change in the 20th century is about 4 cm, roughly 25% of the changes suggested by observations. The projected sea level change by the end of the 21st century in comparison with year 1999 is 6 cm, 13 cm, 17.5, and 19 cm for 20th century freeze, B1, A1B, and A2 scenarios. Changes in the THC due to increased CO2 are important in future climate regimes. Using the PCM, regional responses of the THC in the North Atlantic to increased CO2 and the underlying physical processes are studied here. The Atlantic THC shows a 20-year cycle in the control run, qualitatively consistent with observations. Compared with the control run, the simulated maximum of the Atlantic THC weakens by about 5 Sv or 14% in an ensemble of transient experiments with 1% CO2 increase per year at the time of CO2 doubling. The weakening of the THC is accompanied by reduced poleward heat transport in mid-latitude North Atlantic. Analyses show that oceanic deep convective activity strengthens significantly in the Greenland-Iceland-Norwegian (GIN) Seas, but weakens in the Labrador Sea and the south of Denmark Strait Region (SDSR). The strengthening of deep convective activity in the GIN Seas is mainly caused by an increased salty North Atlantic inflow, reduced sea ice volume fluxes from the Arctic into this region, all of which lead to a saltier (denser) upper ocean. The weakening of deep convective activity in the SDSR is induced by a reduced sea ice flux into this region and a reduced heat loss to the atmosphere, which leads to a warmer (lighter) upper ocean. On the other hand, the weakening in the Labrador Sea is mainly attributed to increased precipitation that freshens the surface ocean. These regional changes produce the overall weakening of the THC in the Labrador Sea and SDSR, and more vigorous ocean overturning in the GIN Seas. The northward heat transport south of 60oN is reduced with increased CO2, but increased north of 60oN due to the increased North Atlantic water across this latitude.
This figure shows Atlantic maximum meridional overturning circulation evolution from 1870 to 2099.