Long-Term Drought Forecasting based on Climate Signals using Multi-Channel

Abstract:

Drought is an insidious natural hazard, which may cause severe damage both in natural environment and human society. Timely drought forecasting enables civil protection authorities and public to take actions to reduce the risk of droughts. Thus drought forecasting plays an important role in setting out drought mitigation strategy. Analysis of the dominant oscillations of droughts and large-scale climate indices has shown that climate indices, such as the El Niño Southern Oscillation (ENSO), are significant indicators of drought occurrences in southern United States. It suggests that the climate indices may be used in drought forecasting at a long lead time. In this study, the multi-channel entropy spectral analysis (MCESA) was developed to incorporate the ENSO climate signals to entropy approach for long-term drought forecasting. To focus on the lack of surface water, drought was quantified by standardized streamflow index (SSI) in this study. SSI time series turned out to be stationary and highly autocorrelated, which showed significant 12-month periodicity. As a result, SSI was successfully forecasted using MCESA with ENSO as an indicator for lead times of 4-6 years. The drought forecasting was more reliable for the stations in humid areas than arid areas. Comparison from the retrospective drought forecasts with or without ENSO showed that inclusion of ENSO climate signals reduced the forecasting errors. The forecasts under El Nino (La Nina) condition reduced (increased) drought severity, making the forecasts more accurate.