Simulation and Management of On-Demand Irrigation Systems: A combined agrological and remote sensing approach

Rational use of water resources in agriculture requires improvements in the efficiency of irrigation. Many irrigation systems, particularly in Mediterranean regions, have been enhanced by replacing open channel conveyance systems with pressurised pipelines. This allows to provide water on-demand. Increased demand of water for civil and industrial uses and a progressive reduction of available water resources compel a more efficient use of irrigation water. To achieve this goal irrigation managers need to understand and to monitor the processes which determine the operation of an irrigation system.In this thesis a procedure integrating the agrohydrological aspects of irrigation with hydraulic and management aspects has been developed. The procedure named SIMODIS (SImulation and Management of On-Demand Irrigation Systems) is based on the integration of different tools such as agrohydrological and hydraulic simulation models, remote sensing and GIS techniques.An irrigation system is described as a set of elementary (e.g. individual fields) connected by the pressurised conveyance system. The spatial distribution of soil water deficit in each elementary unit is computed daily by combining the soil water model SWAP with occasional satellite-based estimates of crop water requirements. A methodology has been developed to obtain spatially distributed input data for the soil water model SWAP i.e. the soil hydraulic properties and the upper and lower boundary conditions.Multispectral satellite images are used to map the crop coefficients needed for the definition of the SWAP upper boundary condition in each elementary unit of the irrigation district. Two different approaches have been proposed. The first is based on classification techniques, where clustering algorithms are applied to derive the spectral classes corresponding to different crop coefficient values. In the second approach, the crop coefficient is analytically related to the canopy variables determining the potential evapotranspiration i.e. leaf area index, surface albedo and crop height. At-surface directional spectral reflectance are used to estimate these canopy variables from which the value of crop coefficient is calculated.The spatial distribution of farmers' water demand is derived on a daily basis from the soil water deficit according to predefined irrigation scheduling criteria. Before applying this farmers' water demand distribution for the given day, the SIMODIS procedure assess whether water demand is consistent with the available amount of water resources and with the structural and operational constraints imposed by the conveyance and distribution system. For this purpose a steady-state simulation model of pipeline hydraulics is used in SIMODIS. The final distribution of farmers' water demand is then resulting from a three-tiered adaptation of irrigation schedule considering: i) the limitation of flow rate at delivery outlets, ii) the limitation of available water resources, iii) the required minimum hydraulic head at the delivery outlets.The procedure SIMODIS has been applied in the Gromola irrigation district of approximately 3000 ha in southern Italy. Measurements of irrigation volumes were used to identify the parameters driving irrigation scheduling. Irrigation efficiency indicators were calculated from the spatial distribution of actual transpiration rates and of the corresponding irrigation volumes applied. To illustrate the use of SIMODIS in support of irrigation decision making, alternative scenarios of water management were simulated and compared.The development of SIMODIS demonstrated that agrohydrological simulation models and remote sensing can be effectively combined to describe the operation of an irrigation system. These techniques have reached a sufficient degree of reliability to be transferred to practical applications. The estimation of crop coefficients by means of remote sensing techniques is of general usefulness in the definition of the upper boundary condition of distributed hydrological simulation models and it can be applied to evaluate with satisfactory accuracy the crop water requirements at regional scale. In the future new types of satellite sensors will probably allow for a more precise determination of the canopy variables, thus providing novel opportunities in the integration between agrohydrological simulation models and remote sensing techniques