Evaluating multiple indices of agricultural water use efficiency and productivity to improve comparisons between sites and trends

Approximately 70% of global available freshwater supplies are used in the agricultural sector. Increased demands for water to meet growing population food requirements, and expected changes in the reliability of freshwater supplies due to climate change, threaten the sustainability of water supplies worldwide - not only on farms, but in connected cities and industries. Researchers concerned with agricultural water use sustainability use a variety of theoretical and empirical measures of efficiency and productivity to gain insight into the sustainability of agricultural water use. However, definitions of measures, or indices, vary between different natural and political boundaries, across regions, states and nations and between their respective research, industry, and environmental groups. Index development responds to local data availability and local agendas, and there is debate about the validity of various indices. However, real differences in empirical index measures are not well-understood across the multiple disciplines that study agricultural water use, including engineering and hydrology, agronomy, climate and soil sciences, and economics. Nevertheless reliable, accessible, and generalizable indices are required for planners and policymakers to promote sustainable water use systems. This study synthesizes a set of water use efficiency and productivity indices based on academic, industry and government literature in California and Australia, two locations with similarly water-stressed and valuable agricultural industries under pressure to achieve optimal water use efficiency and productivity. Empirical data at the irrigation district level from the California San Joaquin Valley and Murray Darling Basin states of Victoria and New South Wales in Australia are used to compute indices that estimate efficiency, yield productivity, and economic productivity of agricultural water use. Multiple index estimates of same time-series data demonstrate historical spread in efficiency and productivity measures in different agricultural regions. Individual indices consistently over- or under- estimate trends in efficiency and productivity by their construction, and may provide inaccurate results in years with extreme climatic events, such as droughts. By treating multiple indices as an "ensemble" of measures, analogous to the treatment of multiple climate model predictions, this study quantifies likely "true" states of efficiency and productivity in the selected agricultural regions, and error in individual indices. While different individual indices are preferable at different scales, and relative to the quality of available input data, ensemble indices can be more reliably used in comparative study across different agricultural regions, and for prediction.