Drought is a recurring feature of the Australian climatic environment.  Drought is commonly defined as a deficit in precipitation compared to expectations for a given climatic environment. However, simply measuring drought by rainfall deficits alone can be problematic and could lead to inaccuracies in agricultural applications.  For example, measuring meteorological drought alone can overlook the concept of ‘rainfall effectiveness’ where factors like rainfall intensity, its seasonal timing, temperatures, evaporation and other in-field factors modify the influence of a rainfall deficit has on crop and pasture growth.

Here we explore different methods of characterizing historical drought in New South Wales (NSW) using output from the NSW Department of Primary Industries’ (DPI) Enhanced Drought Information System (EDIS) Mark II, derived from the ANUClimate 1km2 gridded climate dataset.  EDIS ingests temperature, rainfall, vapour pressure, and solar radiation from ANUClimate, runs it through the DPI AgriMod crop-climate soil water balance model, and outputs meteorologic, hydrologic, and agronomic drought indices.  

This analysis explores a number of aspects of estimating drought including the relationship, configuration and integration of the individual drought indices in NSW since 1970.  It compares traditional percentile estimation of drought with an alternative way of quantifying drought through a non-parametric bimodal distribution. We focus on the manifestation of the 1982-1983 drought, Millennium Drought and current 2017-present drought. The outcome of the analysis points to the existence of operational indeterminacy in drought science, a condition where there are multiple methodologies that can lead to distinct perspectives on drought in both near real time applications and also for analysing long term drought risk.