Project Summary
This is a renewal proposal to continue our research on improving understanding and predictions of decadal-timescale climate variability in the Pacific Ocean sector in the context of greenhouse warming scenarios. This region is emphasized because of its importance to climate over North America, both directly via atmospheric links with North Pacific Ocean processes and indirectly via atmospheric and oceanic teleconnections from tropical Pacific variability. The outcome of this work will be an improved understanding of Pacific climate variability at decadal timescales and their expected changes and regional impacts under conditions of greenhouse warming. These insights are necessary to improve and understand the limits of long-range climate forecasts in the region.
The work makes heavy use of global coupled ocean/atmosphere general circulation models (O/A GCMs), such as PCM, supplemented by regional models, such as RSM and ROMS, and simplified theoretical models. It also includes data analysis for model validation and climate diagnostics. Our emphasis is on obtaining a detailed understanding of the relevant physics involved, rigorous statistical tests of proposed cause/effect relationships, and comparison with observations where available.
Predicting the climate for the next decades requires understanding both natural and anthropogenically forced climate variability. Such variability is important because it has major societal impacts, for example by causing floods or droughts. Our work has been and will be aimed at increasing our ability to predict low-frequency variability.
The work will be divided into three main topics. 1) Fundamental dynamics and predictability limits of decadal climate variability in the North and South Pacific Ocean, including the role of ocean-atmosphere feedbacks, the effects of subducted thermal anomalies on the tropics, and how midlatitude wind variability affects the Indonesian throughflow. 2) The process of making and evaluating climate predictions, including the effect of ocean heat content initial conditions, the effect of specifying SST versus surface heat flux in forced atmospheric runs, and improved code for model diagnoses. 3) Predictability of natural and anthropogenically forced subcontinental scale climate variability in the North American monsoon, U.S. west coast, and California current system. Much of this work will be performed by analyzing existing historical and anthropogenic forcing scenario runs of the Parallel Coupled Model (PCM), which makes good use of this already-available resource. We will use PCM for targeted numerical experiments as well. The results will span a range of geographical scales from interhemispheric to subcontinental, but will have a common focus on decadal timescales and the extraction of a predictable, forced signal from the noisy background of internal atmospheric variability.
1. Overview
The overall goal of the proposed research is to improve understanding and predictions of decadal-timescale climate variability in the Pacific Ocean sector. This region is emphasized because of its importance to climate over North America, both directly via atmospheric links with processes in the North Pacific and indirectly via atmospheric and oceanic teleconnections from tropical Pacific variability. The outcome of this work will be an improved understanding of Pacific climate variability at decadal timescales, and their expected changes and regional impacts under conditions of anthropogenic forcing. These insights are necessary to improve and understand the limits of climate forecasts in the region. Additionally, this work will result in new model techniques and code to help make and evaluate those forecasts. This work makes heavy use of global coupled ocean/atmosphere general circulation models (O/A GCMs), supplemented by regional and simplified theoretical models where appropriate. Our emphasis throughout is on obtaining a detailed understanding of the relevant physics involved, rigorous statistical tests of proposed cause/effect relationships, and comparison with observational diagnoses where available.
2. Introduction and Relevance
Predicting the climate for the next decades requires understanding both natural and anthropogenically forced climate variability. Such variability is important because it has major societal impacts, for example by causing floods or droughts. The work proposed here is aimed at increasing our ability to predict low-frequency variability. It is organized into three topics.
1) The fundamental dynamics of decadal climate variability in the Pacific Ocean, including predictability and the expected effects of anthropogenic forcing.
2) The process of making and evaluating climate predictions, including the effect of the initial conditions of ocean heat content, the effect of specifying SST versus specifying surface heat flux, and techniques for improved model diagnoses.
3) Predictability of natural and forced climate changes over western North America, including the coastal ocean and American southwest.
The bulk of this work directly addresses the first area of relevance to the Climate Change Prediction Program (CCPP), ``theoretical limits to global climate prediction over decadal to multi-century time frames with subcontinental and smaller scale spatial accuracy''. Our primary goal is to assess those limits by identifying the physical mechanisms of climate variability, which is necessary to understand the predictive limits. Both natural variability and the expected changes under anthropogenic forcing will be examined. It is anticipated that forced changes, at least in some mechanisms of natural variability, will provide the basis for decadal-timescale predictions of climate change.
The work examining how model initialization and surface forcing affect model accuracy is relevant to the third CCPP interest area, the development of improved mathematical and numerical techniques for more accurate predictions of global climate. We will also develop and disseminate software for improved statistical evaluation of changes in model variables. This work directly addresses the fourth CCPP interest area, the development of improved diagnostic methods and tools for GCMs.
Section 3 of this proposal describes the understanding we have gained of climate variability in the Pacific Ocean by our previous work, much of it accomplished with DOE funding. Section 4 describes our research goals and outlines the tasks necessary to realize them. Section 5 describes personnel and collaborations, and Section 6, budget justification.
3. Previous Accomplishments
Our accomplishments under previous long-term DOE support provide the basis for much of the proposed work. This previous support is reflected in XXX publications with full or partial DOE support that have either appeared in or are currently submitted to the peer-reviewed literature (see reference list).
Our new results, which have been obtained in the first 2.5 yrs of current DOE funding, fall broadly into two topics: the physics and predictability of decadal climate variability in the Pacific Ocean sector, and the regional impacts of climate changes over the North Pacific, western North America and its coastal ocean. We now briefly describe the results of these refereed publications.
3.1 Dynamics and Predictability of Decadal Climate Variability Over the Pacific
The `null hypothesis' of climate variability is that low-frequency climate is essentially red noise. Pierce (2001) describes in detail the expected response of the ocean to chaotic atmospheric forcing. The expected response to this white-noise forcing includes strongly enhanced power in the decadal frequency band relative to higher frequencies, pronounced changes in basin-wide climate that resemble regime shifts, preferred patterns of spatial variability, and a depth-dependent profile that includes variability with a standard deviation of 0.2--0.4 C over the top 50-100 m. Weak spectral peaks are also possible, given ocean dynamics. Detecting coupled ocean-atmosphere modes of variability in the real climate system is difficult against the spectral and spatial structure of this ``null-hypothesis'' of how the ocean and atmosphere interact, especially given the impossibility of experimentally decoupling the ocean from the atmosphere. Turning to coupled ocean-atmosphere models to address this question, a method for identifying coupled modes by using models of increasing physical complexity is described by Pierce. He found that a coupled ocean-atmosphere mode accounts for enhanced variability with a time scale of ~20 years/cycle in the Kuroshio extension region of the model's North Pacific.
Schneider, Miller and Pierce (2002) studied the 20-30 year peak in that coupled model's oceanic streamfunction and SST fields. They determined the peak was due to gyre-adjustment processes and a positive ocean-atmosphere feedback, much like what was suggested originally by Latif and Barnett (1994) in their analysis of this coupled model. However, no delayed negative feedback loop was identifiable. Decadal changes in wind-stress curl force ocean Rossby waves that force changes in Kuroshio-Oyashio Extension (KOE) SST that vent heat fluxes to the local atmosphere. While Schneider and Miller (2001) showed that these SST changes are predictable out to 2-3 years (much longer than ENSO predictive timescales), the atmospheric response appears to be local in the coupled model.
The sensitivity of the atmosphere to changes in the KOE region is therefore fundamentally important in setting up coupled decadal modes in the North Pacific gyre mode context. Yulaeva, Schneider and Pierce (2001) examined this sensitivity by forcing a coupled model with mixed-layer heat flux anomalies, which mimic the effects of Rossby waves arriving from the east. The full coupled system is allowed to respond to this, unlike the inconsistent AMIP framework where SST anomalies (with infinite heat capacity) are used as forcing. Wintertime model responses to mixed layer heat budget perturbations in the KOE and in the tropical central Pacific show statistically significant anomalies of 500-mb geopotential height (Z500) in the midlatitudes. The response of Z500 to forcing in the KOE region resembles the mixture of western Pacific and Pacific-North American patterns, the first two modes of the internal variability of the atmosphere. The response in the wind stress field alters Ekman pumping in such a way that the expected change of the oceanic gyre, as measured by the Sverdrup transport, would counteract the prescribed forcing in the Kuroshio extension region, thus causing a negative feedback. This response is consistent with the hypothesis that quasi-oscillatory decadal climate variations in the North Pacific result from midlatitude ocean-atmosphere interaction.
The influence of tropical oceanic processes was investigated in two studies. Pierce (2002) examined the interaction of decadal variability with ENSO's effects over North America. During individual El Nino and La Nina episodes, atmospheric circulation anomalies over North America are characteristically different for different phases of the NPO. Two physical mechanisms could account for this observed link between North Pacific SSTs and ENSO's effects over North America. 1) NPO SSTs could force changes in the overlying atmosphere that modulate ENSO's effects. 2) The atmosphere could undergo internal variability that both modulates ENSO's effects and imprints a characteristic pattern of North Pacific SSTs. The first mechanism suggests methods for increasing the skill of seasonal climate predictions by incorporating the state of the North Pacific, using simple persistence of SSTs if nothing else. The second mechanism implies that North Pacific SSTs add no information that could be used to improve seasonal climate predictions of ENSO's effects. Analysis of a 300-year run of a coupled ocean-atmosphere model shows that it exhibits links between NPO and ENSO similar to those observed. It is found that specifying NPO SSTs does not force these links. This suggests that the association found between NPO SSTs and ENSO's effects is primarily because both are responding to the same internal atmospheric variability. In such a case, incorporating accurate predictions of NPO SSTs into ENSO prediction schemes would have little ability to improve forecasts of ENSO's effects.
Schneider (2004) studied the ocean-atmosphere response to the surfacing of temperature anomalies from the tropical oceanic thermocline. Using a coupled general circulation model, he showed how density compensating temperature and salinity (spiciness) anomalies emerging in the upwelling region of the equatorial Pacific modulate tropical climate. The coupled response is qualitatively consistent with a coupled decadal climate mode that results from a positive feedback between equatorial emergence of spiciness anomalies and the equatorial pycnocline and Southern Hemisphere responses, and a delayed, negative feedback due to Northern Hemisphere subduction. However, feedbacks are weak, and, at best, slightly enhance a decadal modulation of the tropics due to spiciness anomalies generated by stochastic atmospheric forcing.
Pierce (2004b) developed new techniques for how large-scale climate models are validated by going beyond a comparison of the model's mean and variability to observations. New applications are placing more demands on such models, which can be addressed by examining the models' distributions of daily quantities such as temperature and precipitation. An example of the systematic differences between model and observations that this technique uncovers can be seen along the coast of northern California and southern Oregon (Figure 1). Observations show positive values, which is the result of multiple day 'heat spells' skewing the distribution in the positive direction. In the model, this behavior is incorrectly limited to the southern California. One inference is that this particular model should probably not be used to examine heat spells along the U.S. west coast. Extending this kind of analysis to other variables, and selecting certain periods of interest (e.g., El Nino vs. La Nina years), can help describe the ability of the model to address particular questions, in particular regions.
Miller, Chai, Chiba, Moisan and Neilson (2004a) reviewed the various theories of decadal-scale climate variations in the Pacific Ocean, including the mechanisms by which the physical ocean-atmosphere system and the oceanic ecosystem interact. These mechanisms include physical forcing of the ecosystem by changes in solar fluxes, ocean temperature, horizontal current advection, vertical mixing and upwelling, freshwater fluxes, and sea ice. These also include oceanic ecosystem forcing of the climate by attenuation of solar energy by phytoplankton absorption and atmospheric aerosol production by phytoplankton DMS fluxes. Miller, Gabric, Moisan, Chai, Neilson, Pierce and Di Lorenzo (2004) examined these physical-biological ocean-atmosphere interactions in the context of the global warming scenario. A more complete understanding of the complicated feedback processes controlling decadal variability, ocean ecosystems, and biogeochemical cycling requires a concerted and organized long-term observational and modeling effort.
3.2 Regional Impacts of Climate Changes Over the North Pacific and Western North America