Sediments deposited on the sea floor, on lake bottoms, as well as on land as accumulations of windblown silt (loess) and as ancient soils (paleosols) formed in the loess possess many different properties that record past changes in climate. Geologically recoverable information on paleoclimate includes changes in global temperature as it may affect global ice volume and thus temperature of sea water, net precipitation, storm intensity (wind speed) and frequency, as well as wind direction. Such information further contributes to interpretations about changes in oceanic and atmospheric circulation patterns. As such, the understanding of past climatic changes can provide baselines of natural climatic fluctuations as related to those that may be induced or influenced by human activities.
Among the properties that provide proxies for paleoclimate changes in marine, lacustrine, and loess-paleosol sequences are the magnetic properties that indicate the concentration, composition, and magnetic grain sizes (domain states) of magnetic minerals. Sediment-magnetic studies that employ these properties play a multifaceted role in global-change investigations [ Banerjee, 1994]. First, high-resolution and high-sensitivity magnetic records may be used to test theories of the ultimate causes of climate change, such as the orbital-insolation theory of Milankovitch [see review paper by Berger, 1988]. Second, magnetic records can delineate glacial and interglacial intervals represented by sediment sequences. Third, magnetic records have the potential to provide information on the rate of climate change and on the different temporal responses of the inorganic and biological climate proxies. Finally, magnetic records are potentially important for comparisons between oceanic and terrestrial Quaternary climate cycles. Such comparisons will improve our understanding of the processes and timing of climatic forcing and will help evaluate possibly different responses in marine and continental settings.
Magnetic-property studies can be applied widely to paleoclimate investigations for several reasons. First, magnetic minerals are found in nearly all natural materials. Second, many climatic processes that influence weathering, erosion, and transport have profound effects on the concentration, size, and mineralogy of the magnetic minerals found in natural materials. Finally, certain magnetic-mineral assemblages are diagnostic for particular source areas and (or) transport mechanisms. In the system approach to environmental magnetism in general, the links among sediment sources, weathering, transport processes, and depositional processes must be understood to interpret properly the magnetic records from lakes, the sea, and land [ Thompson and Oldfield, 1986]. Such understanding is essential for studies of paleoclimate, because many factors other than climate can influence the input of magnetic minerals to these settings [ Thompson and Oldfield, 1986; Verosub and Roberts, in press]. Examples include tectonism, seismic events, volcanism, and changes in source areas. Other variations can be produced by post-depositional alteration, including the destruction of detrital magnetite or the production of certain magnetic minerals. Changes can also result from biogenic production of magnetic minerals and from anthropogenic sources of magnetite, such as coal combustion.
Three major requirements are critical to the successful application of magnetic-property studies to paleoclimate investigations. First, the spatial dimensions of the physical-ecological system under study must be correctly parameterized. For example, the appropriate spatial dimension for a study of sediments from a small lake is usually the lake and its surrounding watershed, whereas a global perspective is necessary for a study of marine sediments due to the long-range influences of currents, wind systems, and ice sheets. Second, an accurate chronology of sediment accumulation is critical both for dating events and determining rates of processes. A combination of independent dating techniques (i.e., radiometric, isotopic, bio- and magnetostratigraphic) is usually needed to obtain an accurate chronology. Third, because smoothing of the environmental record is inversely related to the sediment accumulation rate, it is critical to study lacustrine and marine sediments with an accumulation rate that is appropriate for high-fidelity recording of the frequency range of the paleoclimatic behavior of the system. For example, Mitchell [1976] described the relative variance spectrum of climate variability from 100,000 years (100 kyr) to 0.1 years as having a decrease in power with decreasing period that is produced by the integration of stochastic processes on many time scales. Significant concentrations of variance above the continuum are observed only at the Milankovitch periods (100 kyr, 41 kyr, and about 20 kyr) and at a sub-Milankovitch period of 2,500 years [ Mitchell, 1976]. Recent work also indicates significant variance at a period of 10,000 to 12,000 years [ Park et al., 1993; Hagelberg et al., 1994]. Linear forcing mechanisms due to variations in Earthþs orbital tilt [obliquity] and orbital precession have been identified, respectively, for the climatic variance observed at 41 kyr and around 20 kyr [e.g., Imbrie et al., 1992], whereas non-linear forcing due to variations in eccentricity (that defines the ellipticity of the orbit) is often invoked to explain the climatic variance observed at 100 kyr [e.g., Imbrie et al., 1993]. Non-linear climatic responses at periods equal to precessional harmonics can be invoked to explain the sub-Milankovitch variability [e.g., Hagelberg et al., 1994]. Recent work on the recording fidelity of magnetic secular variation (SV) [ Lund and Keigwin, 1994] illustrates that sediment accumulation rate controls the fidelity of the recording process. For example, SV cycles with periods of about 500 years or longer are accurately recorded in marine sediments with sedimentation rates of about 30 cm/kyr, whereas accurate recording is limited to periods greater than perhaps 1,000-2,000 years at sedimentation rates of about 10 cm/kyr [ Lund and Keigwin, 1994]. A similar concept is applicable to the accurate recording of paleoclimatic variations by magnetic minerals in lacustrine and marine settings. Thus, only very rapidly deposited sediments (greater than 30 cm/kyr) should be studied to characterize accurately climatic variations with a period of a few hundreds of years, whereas more slowly deposited sediments can accurately record longer period variations. Varved sediments, however, offer outstanding future opportunites for higher resolution magnetic-property studies applied to paleoclimate problems.