This talk is meant to provide a broad survey of observed climate variability. As the opening talk of the colloquium, however, some basic concepts and definitions will first be presented. This includes the definition of climate versus weather, an overview of the different components of the climate system, external versus internal mechanisms of climate variability, the predominantly irregular character of climate variability, and an introduction to different sources of climate information. Historical atmospheric and oceanic observations fall into the latter category, and this talk will highlight the strengths and limitations of several key instrumental records. Finally, evidence for decadal and longer time scale variability in the atmosphere and oceans will be presented, with an emphasis on recent changes and their contribution to warming trends in global mean temperature.
In the atmosphere, phenomena and events are loosely divided into the realms of weather and climate. Weather consists of large, short-term (hourly and daily) fluctuations that arise from internal atmospheric instabilities. The evolution of weather is governed by nonlinear, chaotic dynamics and, thus, weather events are not predictable deterministically beyond a week.Climate can be thought of as the prevailing weather and includes both the mean and the range of variations. It involves atmospheric interactions with other parts of the climate system, such as the ocean. The complex interactions between, as well as within, the different components of the climate system make climate prediction difficult.
The intimate link between weather and climate provides a basis for understanding how weather events might change under a changing climate (Fig 1). Since most atmospheric variables follow a normal frequency distribution, a change in the mean and/or variance will likely produce an amplified change in the occurrence of extreme weather events.
Although the common definition of climate refers to the prevailing or average weather, the definition of the climate system must include the relevant portions of the broader geophysical system that increasingly interact with the atmosphere as the time period considered increases. The climate system includes the atmosphere, the ocean, the cyrosphere, the land and its biomass.
The atmosphere is the fastest changing component of the global system and the most responsive. Because of the nonlinear nature of atmospheric interactions, the time and space scales over which it varies are not directly related to the time and space scales over which it is forced. In the past, the atmosphere alone was considered to be the prominently dynamic and changing component, while the other components were thought to be more static.
The ocean is the other fluid component of the climate system. It covers more than 70% of the Earth's surface and, through its high heat capacity, its motions, and its ecosystems, the ocean plays a central role in shaping the Earth's climate and its variability. Interactions between the ocean and atmosphere are very complex and occur on many different times scales, from days in the upper ocean layers to changes on the order of hundreds or thousands of years in the deep ocean.
The cyrosphere includes seasonal snow cover, sea ice, the ice sheets of Greenland and the Antarctic, and mountain glaciers.It represents the largest reservoir of fresh water. Its importance for the climate system comes mainly from its high reflectivity (albedo) for solar radiation and its low thermal conductivity, meaning it acts as an insulator for the underlying land and waters, preventing them from losing heat to the atmosphere. The continental snow and sea ice vary seasonally and interannually, and they cause large variations in energy exchange between the surface and the atmosphere and upper-ocean mixing.
The land component of the climate system includes the slowly changing extent, position and orography of continents and ocean bottom topography and basin configurations.These can be considered as static components of the climate system for the time scales of interest to this colloquium. However, lakes, rivers, streams, soil moisture and vegetation vary on more rapid time scales and must be taken into account.
Many of the following lectures will deal specifically with mechanisms of climate variability. Due to the complexity and interactive nature of the global system, the climate system is often separated into internal and external components. By and large, the separation is made on the basis of the characteristic time scale of change of individual components of the system.Changes in climate resulting from changes in orbital geometry or solar output are examples of externally produced climate variability, while variations intrinsic to the atmosphere or ocean, or variations that arise from the coupling of the atmosphere and ocean are examples of internally produced variability.
The evolution of the climate system assumes a variety of forms. Predominantly, climate variability is irregular:apart from the diurnal cycle and the annual cycle, there are few proven periodicities or even quasi-periodic signals, especially below the stratosphere. The characteristic irregularity arises from the large number of physical processes within and the interactions between the different components of the climate system. Some idealized examples of climate variability are illustrated in Fig 2.
Several sources of data are available for studies of decadal and longer time scale climate variability. Model data are our only tool for estimating future climate and unraveling cause and effect relationships. Proxy records from ice cores, corals, tree rings, sediments, and historical documents are very useful, but the extraction of climate information requires understanding of how signals are incorporated into the records and of competing influences. Here, the focus is on instrumental records as well as analysis products.
The longest instrumental records are on the order of 100 years.As such, the records are too short to adequately define decadal and longer-term climate variability.Moreover, many data sets are not complete, with extended periods when observations are not reported or stations come and go.It is also important to realize that most observations are taken to support daily weather forecasting, not for the purpose of monitoring climate.Consideration of biases arising from instrument exposure or changed location is not important when monitoring large day-to-day weather variability; however, it is critical when monitoring the much smaller changes associated with climate variability and climate change.
The result is poor quality and a lack of homogeneity so that much original data cannot be considered to be of research quality. Discontinuities can overwhelm long-term (decadal) changes.Often the adjustments that must be applied to a climate record are of such magnitude that it is difficult to be confident that the resulting, adjusted time series adequately reflects actual climate fluctuations. Presently, few observation systems and practices maintain the accuracy, precision, and absolute calibration required over time for climate and global change studies.Each data set, moreover, requires its own unique set of adjustments.
Progress has been made, however, through continual evaluations and comparisons of instrumental data sets. One way to gain confidence is through the physical consistency among independent variables.For instance, one might look for consistency between surface winds, surface pressure, and sea surface temperature (SST) in an analysis of variability over the oceans.
A good example is the historical SST record.Probably no single climate element has been studied more that SST and marine air temperature (MAT) over the oceans, and near-surface air temperature over land.These data sets suffer from two key deficiencies:poor and changing spatial coverage, and time-varying biases.
Over the oceans, there are many areas with few data, especially in the early part of the record. The most poorly sampled regions are the tropical and southern oceans. Even in recent decades, there are few ship-based measurements of SST south of 45?S. In recent decades, satellite data provide global coverage, but they too suffer from time-varying biases and other problems that must be dealt with before they can be utilized effectively.
Perhaps the most striking evidence of heterogeneity in the historical SST record is a sudden increase in global SSTs around 1941.This is consistent with a fairly abrupt transition from the use of uninsulated or partially insulated buckets to the use of engine inlets to measure SST.Just how widespread and sudden this transition was is not easy to determine, and the magnitude of the jump depends regionally on the climatic conditions and specifically on the details of measurements made by different ships.Generally, the bias is believed to be on the order of 0.5°C, although the quality of engine intake temperatures varies considerably. Other sources of noise, such as errors in making individual measurements, and incomplete sampling of the diurnal cycle, within-month variance, and strong spatial gradients all contribute to uncertainty in monthly mean SSTs. The inherent noise level in an individual SST observation is about 1°C over the tropics and 1.2-1.4°C over the extratropical oceans. Since the standard error of the mean decreases by a factor of ~ÖN (where N is the number of SST observations in a 2° grid square), the overall noise in monthly mean SSTs ranges from less than 0.1°C over the North Atlantic, ~0.1°C over the north Pacific, tropical Indian and Atlantic Oceans, ~0.2-0.3°C over the tropical and south Pacific Oceans, and more than 0.5°C south of about 35°S.
Subsurface marine data are critical for the documentation of the mean and variability of the ocean circulation. However, relative to the SST archive, the data are much more limited and, except for a few regions, are available over only the past four decades. The subsurface marine data include ocean station data, conductivity-temperature-depth (CTD) profiles, and bathythermograph measurements (MBT and XBT). Since XBTs are inexpensive, disposable, and can be released from merchant ships, they are more abundant and provide most of the subsurface measurements today.
A final instrumental data set to consider today is that of upper-air observations.They began in the 1940s; however, large numbers of observations only became available starting with International Geophysical Year (IGY) in 1957/58. There are fewer stations than for the surface network, with most stations over North America and Europe, and very few south of about 20°N.
Many sources of error for upper-air data are similar to those for surface observations, including random errors and time-varying biases resulting from changes in instrumentation and operating or processing procedures. These problems, moreover, are exacerbated by inadequate station history metadata throughout much of the world. Also, a major problem is the lack of continuous time series.Related to this is that monthly means, which are frequently used in climate research, are often based on less than a full month of data, and the missing days are often consecutive and not random.The result is considerable uncertainty, especially in estimates of temperature change over time.
The availability of global analysis products has resulted in tremendous advances in our understanding of how the atmosphere works.These data sets are produced by operational centers for weather forecasting purposes.They represent a statistical combination of short-term forecasts and observations. The result is gridded, global fields with no missing data, and many derived fields (surface fluxes, vertical motion, diabatic heating, etc.) are produced. A major limitation for climate research, however, is that improvements in models and data assimilation methods introduce large changes over time that obfuscate real climate signals.The goal of producing the best analysis of the atmosphere at any moment in time is not necessarily compatible with the goal of producing the best product for monitoring climate.
Retrospective analysis of all available data with a constant, state-of-the-art assimilation system is known as reanalysis.Several major efforts have been made to date, and these have the advantage of including more observations and eliminating spurious climate jumps due to changes in models and parameterizations.However, these efforts have also been a learning process, and several problems limit the utility of reanalysis products for studies of decadal climate variability.For instance, the reanalysis products are only as good as the input database, and changes in observing systems over time produce artificial changes in the perceived climate.One example is the introduction of satellite data in the 1970s. So while these data have significantly improved our understanding of the evolution of the atmosphere, they too must be used with caution.
The 20th century was one of a long-term warming trend.Global surface temperatures at the end of the century were about 0.7°C warmer that they were around 1900.Moreover, the rate of warming over the past 20 years is unprecedented. Over the Northern Hemisphere during winter and spring, the pattern of change has been one of warming over the continents and cooling over the oceans, much like the surface temperature change pattern expected as a result of increased greenhouse gas forcing. One of the reasons it has been difficult to ascribe this observed pattern to changes in anthropogenic forcing, however, is because it is strongly related to decade-long changes in the atmospheric and oceanic circulation.
At the Earth's surface, the changes in circulation are reflected by lower-than-average sea level pressure over the middle and high latitudes of the North Pacific and North Atlantic Oceans, as well as over much of the Arctic, and higher-than-average sea level pressure over the subtropical Atlantic. These changes in circulation over the past 20 years are related to natural modes of atmospheric variability, in particular the Pacific-North American (PNA) teleconnection pattern and the North Atlantic Oscillation (NAO).
The PNA is the dominant mode of atmospheric circulation variability over the western hemisphere, and it is often viewed as the extratropical arm of the El Niño/Southern Oscillation (ENSO) phenomenon in the tropical Pacific.The NAO dominates climate variability from the east coast of the United States downstream through Eurasia, as well as over much of the Arctic. The rest of today?s talk will define these patterns and examine their variability over the historical record, as well as the links between them.Later talks will focus much more on some of the possible mechanisms of decadal climate variability.
