An integrated approach for multi-year irrigation benchmarking using satellites, surveys and on-farm measured data.

Benchmarking is an effective management tool to improve irrigation performance through comparison with other irrigation management units; however, its application is often limited by the data available to support the analysis. This study developed a benchmarking system that reduces reliance on local data by using satellite-based estimates and quantifies benchmarking uncertainty due to alternative data sources. We benchmarked the relative irrigation supply (RIS satellite) (i.e. the ratio of irrigation supply to crop net irrigation demand) for more than 300 farms growing corn/maize, cotton and rice crops in the Coleambally Irrigation Area in Australia from 2011 to 2019. Three key inputs to RIS satellite , namely irrigated cropping area (ICA), start and end dates of a season (SOS/EOS) and crop coefficient (K c), were derived from Landsat time series data. To understand the uncertainty in satellite-based benchmarking, RIS satellite was compared with RIS non-satellite estimates derived from non-satellite datasets. The variation in RIS with respect to input data sources and the reliability of data sources were also quantified. RIS satellite was highly variable across farms in any particular year, and over-irrigation (RIS satellite >1) was common. In addition, each farm tended to have a consistent RIS satellite ranking across years, indicating the role of farmers' irrigation management practice in RIS. ICA from different datasets contributed most to the random variation in RIS, while satellite-based ICA presented more reliable estimates with a median absolute percentage error of 5 %, which is half the error for survey-based ICA. We conclude that our satellite-based benchmarking system is effective in overcoming data limitations and is likely to be more reliable than systems using survey data. This system can support improved irrigation performance monitoring in this district and is transferable to other irrigation areas with irrigation supply information available. • We developed an irrigation benchmarking system that uses satellite and observed water supply data. • Over-irrigation is found in most farms across crops and years (RIS>1). • Farms tend to have similar RIS when growing the same type of crops in two different years. • Crop area is the input leading to the largest variation in RIS. • Satellite-derived cropping areas show good consistency with ground truth (R2=0.96). [ABSTRACT FROM AUTHOR]