Abstract:
Global water and energy budget studies as well as long term soil moisture observation records have shown that the time scales of soil moisture anomaly dissipation are about 2-3 months. This implies that streamflow should, in principle, be partly predictable at seasonal time scales whenever precipitation is predictable (based on El-Nino or the MJO, for example). Here we investigate the hypothesis of seasonal streamflow predictability for Sri Lanka. Gridded 0.25 degree monthly precipitation observations for Sri Lanka were merged with 0.5 degree 6-hourly bias-corrected global reanalysis data for the period 1979-1993 and used to force the NASA Catchment Land Surface Model (CLSM) over Sri Lanka. CLSM simulations streamflow compared well with station observations. One-month lagged autocorrelation of simulated soil moisture was high (spatial average of 0.75), with anomaly dissipation time scales of 3-5 months. The lagged cross-correlation between simulated total runoff efficiency and soil moisture indicates predictability of streamflow up to 3 months in advance in the drier seasons (April to June and July to September).