Recent studies have shown that loess records in China display similar patterns of climate variability as those observed in polar ice cores and deep sea sediments. While great effort has been made to correlate loess with other records, a major problem is outstanding: the accuracy of the loess chronology. This is a crucial issue in the loess-based palaeoclimatology on both the glacial-interglacial timescale and the millennial-scale. It also has implications for the understanding of the landform evolution in the loess regions. Here we discuss the accuracy for two methods which have been widely used in dating loess sequences: palaeomagnetism and luminescence.

The Matuyama/Brunhes Boundary (MBB) has been an important time marker in loess stratigraphy. While there is a striking similarity in the overall pattern between the magnetic susceptibility of the loess from the Loess Plateau in China and the oxygen isotope record of the deep sea sediments, the MBB is found at very different stratigraphic positions: its occurrence in a loess unit, which represents a glacial climate, is in contrast to its placement in the marine record in Oxygen Isotope Stage 19, an interglacial stage. This has given rise to significant uncertainty in the chronology of the loess-based palaeoclimate records and has impeded the stratigraphic correlation between continental and marine sequences. We have recently suggested that the measured positions of the geomagnetic polarity boundaries in loess are misleading due to the displaced ('delayed') acquisition of magnetic remanence. The 'lock-in' depth in loess could range between a few tens and over 300 cm depending on the process of remanence acquisition in loess, which is linked to the development of the loess structure. The apparent 'time lead' is of a few thousand to over 30,000 years. This means that the age for the level where the MBB is measured, i.e. mid-L8, is older than 795 ky. The 25,000-40,000-year difference from the previously published age estimate for the same level needs to be explained before we make palaeoclimatic interpretation of the loess records. We suggest that the chronology of the Chinese aeolian sequence (loess and red clay) should be re-evaluated with respect to the true age of the measured palaeomagnetic boundaries. Our remanence acquisition model predicts that similar displacement is likely to affect records of the reversals or excursions in loess and red clay of all ages and in most of the cases the age shift will be over 10,000 years.

To date palaeoclimate events within the last 50,000 years, luminescence method offers a most promising means. However, previous studies mostly employ thermoluminescence techniques which probe signals from mixed minerals. This presents an intrinsiclimitation on the accuracy of the published TL dates. We have recently tested a series of new protocols of dating loess by photon-stimulated luminescence. For well-bleached samples, we are able to obtain paleodose estimates with uncertainty as low as 2%. While such a high precision is readily obtainable, the accuracy of the method is more difficult to test due to the lack of suitable material for applications of other independent, more reliable dating methods.