85th AMS Annual Meeting

Wednesday, 12 January 2005: 8:45 AM
Long-Lead Drought Forecasting – Lessons Learned in the Murray-Darling Basin, Australia
A. P. Barros, Duke University, Durham, NC; and G. Bowden
An operational model to forecast meteorological drought at seasonal to interannual time scales for the Murray-Darling Basin (MDB), Australia is presented. The Murray-Darling river system has a catchment area of approximately 1 million km2, or nearly 14% of Australia’s land area and is the fourth longest river system in the world. Droughts are an accepted feature of the climate in this part of the world and the ability to reliably predict such events would provide considerable improvement in the management of the basin’s water resources. For this case study, drought severity is described by the standardized precipitation index (SPI), a measure of meteorological drought. The objective is to develop a model to predict monthly spatial patterns of the SPI within the MDB at long lead time-scales, taking advantage of the strong linkages with ENSO in Australian climate variability. The forecasting methodology involves the spatial analysis of historical records of precipitation, near-global sea surface temperature anomaly (SSTA) patterns over the Indian and Pacific Oceans, and early indicators of El Niño/Southern Oscillation (ENSO) onset including: temporal and spatial gradients of outgoing longwave radiation (OLR) in the Pacific Ocean, and the far western Pacific windstress anomaly. These derived indices are used to build a data-driven model combining a self-organizing map (SOM) and multivariate linear regression techniques. We present results for a 12-month lead-time operational demonstration, this showing that it is possible to produce reliable operational forecasts at adequate lead times useful for effective water resource management. This study also provided an opportunity to examine the stability of the heuristic cause-effect relationships of highly nonlinear and nonstationary systems, and the challenge this poses to climate prediction.

Supplementary URL: