Long-term memory of the hydrological cycle and river runoffs in China in a high-resolution climate model

The hydrological cycle in China is analysed on the basis of a 250-years present-day climate simulation with a high-resolution (T63, approximate to 2 degrees x 2 degrees) coupled atmosphere-ocean circulation model (ECHAM5/MPI-OM). The analysis of the annual data in the model simulation reveals long-term memory (LTM) on decadal time scales in some components of the hydrological cycle. LTM is characterised by a scaling exponent beta>0 in the power spectrum S(f) similar to f(-beta) for low frequencies f and is determined by detrended fluctuation analysis (DFA). The simulated annual precipitation and atmospheric near-surface temperature fields show, as in the observations, a white low-frequency spectrum and, hence, no long-term memory in East Asia. However, simulated river flows of the Yangtze and the Huang He reveal LTM with scaling exponents beta = 0.3-0.4 (similar to the observations and that of the river Nile) extending beyond the decadal time scale. The model soil temperature indicates restricted memory up to time scales of approximately 30 years. In addition, the model's soil wetness, evaporation, and local runoff show memory on this time scale in a zonal belt at the latitude of Mongolia. Copyright (C) 2006 Royal Meteorological Society.