A cave δ18O based 1800-year reconstruction of sediment load and streamflow: The Yellow River source area

Located on the northeast edge of the Qinghai-Tibet Plateau in China, the Yellow River source area (YRSA) is the river's major water source area and therefore it is called “water tower” of the Yellow River. However, the instrumental data are only 50-years, based on which the hydrological response at 100-1000 years cannot be understand properly. Although many results of streamflow reconstruction are published, the results of high-resolution suspended sediment load (SSL) reconstruction have not been reported yet. Here we show that the Wanxiang Cave δ18O record (Zhang et al., 2008) can be used to reconstruct the SSL of the YRSA along with the mean annual streamflow, annual maximum discharge and precipitation. Based on the data of the overlapping period (1961-2001) after 5-year moving, calibration equations were established, using regression method. The R of annual stream flow, SSL, maximum discharge and precipitation with 5-year moving δ18O is − 0.768, − 0.763, − 0.692 and − 0.803, respectively, at p < 0.000001. After validation using Leave-One-Cross Validation method, we reconstructed the 1800-year variations in these variables. Based on statistical analysis, we proved the existence of the following cause-effect chain: the weakened (or strengthened) Asian Summer Monsoon recorded by cave δ18O → increased (or decreased) water vapor to the YRSA → increased (or decreased) precipitation → increased (or decreased) streamflow → increased (or decreased) SSL. These links provide the scientific basis for the paleo-hydrological reconstruction. Over the past 1800 years, the Qinghai-Tibet Plateau's water tower effect on the Yellow River showed an overall decreasing trend, with the 5-year moving streamflow decreased at a rate of 4.81 million m3/year. Eight stages of alternative decrease and increase were identified, of which two long continuing decline stages are likely one of the factor that caused the collapse of the Tang and Ming dynasties. The uncertainties that should be considered to improve the model in the future are also discussed.