Long term predictability of river stages under ENSO influence

The paper explores the bases for alternative approaches to river flow and hydrologic time series forecasting. The precipitation that forces river flow contains significant modes of variability beyond simple noise randomness and these structured statistical properties may be used advantageously in forecasting river flow. Despite the high degree of spatial and temporal intermittency observed in precipitation records, there are often deterministic signals embedded in the time series. These deterministic components of the random variable are related to the physical processes that constitute precipitation formation.