Loess Deposition And Paleoclimate

Loess is a deposit of wind-blown silt that blankets large areas of the continents. It is characteristically creamy-brown in color and calcareous, consisting predominantly of quartz feldspars and micas (Pye, 1984, 1987). Geographically, loess is extensive in the North American Great Plains, south-central Europe, Ukraine, central Asia, China, and Argentina (Fig. 7.1). In North America, loess deposits are related to formerly extensive outwash deposits from the Laurentide ice sheet and to the floodplains of large, braided rivers. Similarly, in Europe loess deposits are common between the former Alpine and Scandinavian Ice Sheets though the thickest loess sections, in the Czech Republic, Slovakia, and Austria, may also have had a more local origin related to formerly extensive braided rivers at a time when vegetation cover was greatly reduced (Kukla, 1975b). In North America, loess deposits are up to 50 m thick in Alaska (Beget et al., 1991) and sections along the Mississippi River valley may reach 20 m in thickness (Forman et al., 1995; Oches et al., 1996). Elsewhere, loess is related to desert conditions, especially the formerly extensive deserts of central Asia. In north central China (southeast of the Gobi Desert) loess deposits are extremely thick, up to 300 m in places, and completely cover the underlying topography, forming an extensive loess plateau (Liu et al., 1985). It is there that the most comprehensive pa-leoclimatic studies of loess deposits have been carried out (Kukla, 1987a; Kukla and An, 1989). Loess has accumulated on the Loess Plateau for -2.5 Ma and during episodes of warmer and wetter conditions weathering of the loess led to soil forma-

Magnetic Susceptibility Austria Loess
FIGURE 7.1 Location of principal loess deposits in the world (Pye, 1984).

tion. Today the alternating sequence of loess units and intervening paleosols (Fig. 7.2) form the most complete terrestrial records of Quaternary paleoclimatic conditions to be found on the continents (An et al., 1990; Ding et al., 1993).

7.2.1 Chronology of Loess-Paleosol Sequences

Early studies of loess-paleosol sections took a simple relative dating approach, counting the first well-developed soil below the surface (Holocene) soil as equivalent to the last interglacial, and older soils representing earlier interglacial episodes. Thus, in China the conventional terminology is based on counting back from the Holocene soil (S0) to the last loess episode (L,) then to the preceding interglacial soil (Sj) and underlying loess (L2) and so on down the section. In this way, as many as 37 soils and associated loess episodes have been recognized at the Baoji type-section (on the southwestern margin of the Loess Plateau); of these, 32 (Sx—S32) are at least as well developed as the Holocene soil, and document the alternation between glacial and interglacial conditions throughout the Quaternary (Rutter et al., 1991a). In Europe, loess deposits are mostly restricted to river terraces and are, therefore, generally less continuous, leading to confusion over the chronology of loess-paleosol sequences and how they are related from one region to another. Recent aminostratigraphic studies (on snails embedded in the loess) have clarified the stratigraphic relationships between many different locations, improving the correlations between sections (Oches and McCoy, 1995b, 1995c). Thermoluminescence dating has proven useful in determining the chronology of loess deposition during the last glacial cycle but

section of loess with interbedded soil units. In this section, the upper loess unit (L,) and Holocene soil (S0) have been eroded away. Soils developed during more humid periods when monsoon rainfall was higher, but loess continued to accumulate in winter months albeit at a slower rate. During major loess deposition episodes, the summer monsoon rainfall rarely penetrated this far into China (photograph by R.S. Bradley).

section of loess with interbedded soil units. In this section, the upper loess unit (L,) and Holocene soil (S0) have been eroded away. Soils developed during more humid periods when monsoon rainfall was higher, but loess continued to accumulate in winter months albeit at a slower rate. During major loess deposition episodes, the summer monsoon rainfall rarely penetrated this far into China (photograph by R.S. Bradley).

dates on older deposits remain controversial (Wintle, 1990; Forman, 1991). For the longer Quaternary record, paleomagnetic studies have provided the fundamental chronology (Heller and Liu, 1984; Rolph et al, 1989; Rutter et al., 1990; Thistle-wood and Sun, 1991) (Fig. 7.3). Once the basic reversal record has been determined some studies then simply interpolate between chron/subchron boundaries to obtain a chronology of the intervening loess and paleosol units (Rutter et al., 1991a). However, this does not realistically take into account the changing loess accumulation rates from full glacial to interglacial conditions (when loess slowly accumulated as soils developed). A more realistic approach assumes a higher loess deposition rate in glacial times (Liu et al., 1993 and Ding et al., 1994 assumed loess accumulated at 1.5 to 2 times that during interglacial soil-forming intervals). This provides a first-

Thickness(m) 0 -i

Lithology

Polarity

150-

Lithology

Polarity

150-

Inclination

Inclination

i / s

\ /

x

*s

>

A

's>.

\

II

■r'

.Jr'

■f

"3

y

eî:*

FIGURE 7.3 Magnetic stratigraphy of the loess-paleosol section at Baoji, Loess Plateau, north central China. Major paleosols (S) and loess units (L) are numbered from the top down (Holocene soil = S„). Changes in polarity are clearly indicated by the inclination record on the right (B/M = Brunhes-Matuyama boundary; J = Jaramillo subchron; O = Olduvai subchron; M/G = Matuyama-Gauss boundary). Loess began to accumulate over a red clay deposit around the time of the Gauss-Matuyama transition (Liu et al., 1993).

order approximation that reveals the strong relationship between the chronology of loess-paleosols in China and the marine isotope record (Fig. 7.4). Loess units correspond to even-numbered marine isotope stages (i.e., times of continental ice buildup) and the soil units correspond to interglacial (odd-numbered isotope stages). Weak soils appear to be related to interstadial events.

-2 -4

Baoji

I ', 1 1 ' ^^ 1

^--5

7

■■■■

Loess Specmap Correlation

FIGURE 7.4 Composite marine oxygen isotope record (left) with stages indicated, and its proposed correlation with the loess-paleosol sequence at Baoji, Loess Plateau, north central China. Paleosol units (S0—S32) are indicated by dark bands, though not all soils are equally well developed. Magnetic stratigraphy of the records is indicated (Rutter et al„ 1990).

Loess Section

l5 u

Lis l,6

L24 L25 L26

L27 L2s l29

FIGURE 7.4 Composite marine oxygen isotope record (left) with stages indicated, and its proposed correlation with the loess-paleosol sequence at Baoji, Loess Plateau, north central China. Paleosol units (S0—S32) are indicated by dark bands, though not all soils are equally well developed. Magnetic stratigraphy of the records is indicated (Rutter et al„ 1990).

This general correspondence suggests that a more accurate chronology of loess-paleosol sequences might be determined by tuning some diagnostic parameter to orbital forcing, just as the SPECMAP group did to resolve the chronology of the marine isotope record (Section 6.3.3). Using this approach, Ding et al. (1994) adjusted the record of grain-size variations at Baoji (a proxy of winter monsoon strength in north central China) to optimize power at frequencies corresponding to the principal orbital periodicities (Berger and Loutre, 1991). Following Imbrie et al. (1984), they used changes in orbital eccentricity, obliquity (lagged 8000 yr), and precession (lagged 5000 yr) as the combined target to which the grain size record was tuned (Fig. 7.5). Confidence in the resulting timescale is provided by a good match with predicted ages of K/Ar-dated paleomagnetic boundaries in the loess-paleosol sequence. Spectral analysis of the Baoji timescale, so derived, reveals some important changes over the course of the Quaternary. For the last 600 ka the grain-size record was dominated by the 100 ka eccentricity period, which accounts for 46% of the variance in this interval of time. However, from 0.8-1.6 Ma the record has a much stronger 41 ka obliquity cycle (30% of total variance) and from 1.6-2.5 Ma a more complex spectrum is apparent, with power concentrated mainly at 55 ka and 400 ka periodicities (35% of total variance). This shift towards a stronger obliquity signal around 1.6 Ma is not seen in the marine isotope record, suggesting some regional response to forcing, whereas the mid-Pleistocene change to a strong -100 ka period is also clear in the marine isotope record and appears to be of global significance (Ding et al., 1994).

7.2.2 Paleodimatic Significance of Loess-paleosol Sequences

Several approaches to interpreting the alternation of loess deposits and paleosols have been made. At the simplest level, modern analogs of similar soils and loess accumulation areas provide an approximation to climatic conditions prevalent at different times in the past. Thus, Ding et al. (1992) suggest that the oldest thick loess units (L33 and L32) represent mean annual temperatures 12 °C lower than today and precipitation levels less than 25% of modern values, based on the modern climate of locations thought to be analogous to conditions prevailing when the loess accumulated. Maher et al. (1994) and Maher and Thompson (1995) focused on magnetic susceptibility as a proxy of rainfall. They argue that susceptibility is higher in soils due to the in situ formation of ultrafine (<0.02 pm) ferromagnetic (maghemite) grains by inorganic precipitation, aided by the presence of magneto-tactic (iron-reducing) bacteria (Maher and Thompson, 1992; Heller et al., 1993; Verosub et al., 1993; Liu et al., 1994). Alternations of wetting and drying cycles favor this process so that susceptibility profiles can be considered a proxy for past rainfall variations. By establishing the relationship between modern rainfall amounts and susceptibility measurements on young soils throughout the Loess Plateau, they were able to calibrate the long record of susceptibility at Xifeng (spanning the past 1.1 M years) in terms of paleoprecipitation (Fig. 7.6). On this basis, rainfall has varied by a factor of -2 over this interval (from -400 to -750 mm) and for 80% of the last 1.1 Ma, rainfall has been less than today at Xifeng. Liu et al. (1995) took a similar approach in reconstructing rainfall at Xifeng, but only for the

41 ka filter 100 ka filter

0 0

Responses

  • Ruairi
    How thick are the loess deposits in Gobi Desert?
    4 months ago

Post a comment