Modelling the effect of climate, dust deposition and ice sheets on paleosol development on the Chinese Loess Plateau

The Chinese loess-paleosols sequences are important archives of the Quaternary paleoclimate and paleoenvironmental information. Loess was deposited during glacials, with a rate of high dust addition. Paleosols underwent strong pedogenesis during interglacials in response to strong summer monsoon and low dust influx rate. Most previous studies of paleosols in the CLP focused on paleoclimate-paleosol linkages by analyzing bulk paleosol characteristics, however much remains unknown about their relationship in terms of relative contributions of paleoclimatic conditions (e.g. precipitation, temperature, evapotranspiration, vegetation and dust addition). Soil formation is controlled by various environmental factors (e.g. dust input, climatic conditions) over an interglacial period; thereby, deducing such interpretations is complex. This motivates us to quantitatively examine relative contributions of precipitation, temperature, evapotranspiration, vegetation and dust addition on paleosol development in the CLP. As far as climate is concerned, ice sheets play an important role in regulating glacial-interglacial cycles during the Quaternary period and have been found to be closely linked with paleosol development in the CLP. However, how ice sheet-induced climate change has affected paleosol development is still unclear. We use SoilGen2-LOVECLIM process-based soil-climate models to simulate paleosol development using time series of climate data (e.g. precipitation, temperature, dust input) for a given interglacial. A sensitivity analysis was done to quantitatively assess the relative contributions of these factors on paleosol development. The sensitivity experiment includes simulations for each climatic factor alone, combined climatic factors (e.g. precipitation and evapotranspiration) and a reference simulation (Pre-Industrial climate). The results were analyzed by calculating the mean absolute error between each of the sensitivity simulation and the actual simulation (com