A Decision Support System for Matching Irrigation Water Demand and Supply in A Real Time Environment

Recent droughts in Australia and concerns about climate change have highlighted the need to manage agricultural water resources more sustainably especially in the Murray Darling Basin (MDB) which utilizes more than 70% of water for food production. Due to continuous and prolonged drought, many irrigation areas in MDB have been seriously affected, to varying degrees, and this has resulted in a significant reduction of water allocation for growing cereal crops. The reduction in water availability forces farmers and irrigation managers to increase beneficial use of water for crop production and account for every drop of water coming in and going out of the system. However, this would be only possible if the irrigation managers have a reliable and timely information available pertaining the actual water use at every paddock within the irrigation systems. This paper focuses on a decision support system (DSS), the Coleambally Integrated River Information System (Coleambally IRIS), for managing irrigation water demand and supply more sustainably in a real time environment, in the Coleambally Irrigation Area (CIA), which utilizes bulk water for food production. The CIA is located in New South Wales, Australia and covers approximately 79,000 ha of intensive irrigation and comprises of a number of secondary and tertiary canals. All components of water accounting such as land use, rainfall, surface water inflow/outflow, groundwater and evapotranspiration (ET) has been measured in the real time environment for CIA.Accurate maps of various agriculture crops using high spatial resolution satellite images (ALOS AVNIR and Rapid Eye) was developed for each cropping season. Seasonal actual evapotranspiration has been estimated from SAM-ET (spatial algorithm for mapping evapotranspiration) algorithm, developed for an Australian agro-ecosystem, using 18 Landsat 5 TM satellite images andresults being further validated with two Eddy Covariance Systems (ECS) and two Large Aperture Scintillometers (LAS). In addition, this article also describes the active ground truth data campaigns being carried out on a fortnightly basis since 2007 to collect additional data about leaf area index (LAI) from LICOR 2000, soil heat fluxes from HuskeFlux, crop reflectance data from CROPScan and from a thermal radiometer. UAV drone equipped with multispectral scanner and thermal imager was used to get very high spatial resolution Albedo, NDVI and surface temperature maps over the selected farms. This large array of high technology instruments have been used to collect specific measurements within various micro-ecosystems available in our study area. Lastly, the holistic spatial water accounting approach has been applied for CIA on weekly basis in the Coleambally IRIS. Similarly, for demand forecasting, a decision tree has been trained using bayesian network and artificial neural network on a training data set having various non-class attributes including weekly forecasted weather data from Bureau of Meteorology (BOM), and a class attribute which is actual ET. Having trained classifiers and forecast of weather data, our Decision Support System (DSS) can predict future ET and therefore the water demand of a farm. Since, the whole irrigation area has been designed in a hierarchical fashion with subsystem and system, and the DSS can calculate water demand of a node from the water demand of all farms belonging to the node.