Quantifying irrigation water use with remote sensing: Soil water deficit modelling with uncertain soil parameters.

Water sharing plans have been implemented to allocate water fairly between irrigated agriculture and other stakeholders. Water accounting is an important feature of successful water management and sharing. Remote regions are often neglected with metering infrastructure and therefore remote sensing is an option for the quantification of irrigation water use. The guides published in FAO Irrigation and Drainage paper No. 56 (FAO56) can provide some information in data poor regions although it has its own limitations without direct observations. This study uses the constellation of Landsat satellites (5–8) to monitor crop conditions via the vegetation index to assess crop growth through a crop coefficient (K c) based on an actual field condition. This remotely sensed input is then used in soil water deficit modelling based on the FAO56 approach over two fields, an almond plantation and a vineyard, located in South Australia. Soil parameters, such as readily available water (RAW), are taken from in-situ observations and digital soil maps for comparison. The results closely matched metered irrigation time series with only small changes in results when interchanging in-situ soil properties or digital soil maps. Following this an uncertainty analysis using a Monte Carlo approach was performed using the range of parameter values for RAW. A small period of this study (9 months of 2015/16) overlapped with the Sentinel-2a operational period, which was investigated for its improved spatial resolution and differences in spectral band width of key vegetation observation bands. When comparing these results to previous studies, which did not consider soil water deficits, the improvements are substantial (improvements ranged from 3% to 15% monthly and 56% to 68% annually). These improvements require extra data, which has the limitation of comprehensive field data being difficult to obtain and digital soil maps being potentially unreliable. The choice to include soil water deficit modelling or not is dependent on the required accuracy to effectively use the quantified irrigation for the intended use. • Remote sensing provides a potential tool for quantifying irrigation water use. • Soil water deficit modelling is effective at improving simulations. • Digital soil maps provide the opportunity for spatial scaling. • Monte Carlo simulations indicate the sensitivity of soil map parameters. • As newer satellites become available, results will likely improve. [ABSTRACT FROM AUTHOR]