Assessing opportunities for water savings in large-scale pressurized irrigation systems using actual evapotranspiration retrieved by surface energy balance and remotely sensed data

Remote sensing allows the observation of large land stretches and the acquisition of worthwhile information that can be used efficaciously in agro-hydrologic systems. Satellite imagery associated to computational models provide a reliable resource in estimating evapotranspiration (ET) fluxes based on surface energy balance. On irrigated crops, quantifying the spatial distribution of actual ET enables a broad range of applications such as irrigation management, monitoring water distribution, assessing crop water status and irrigation system performance. The general objective of the research was to propose a methodology to estimate ET by using Landsat Thematic Mapper (TM) images and surface energy balance (SEB) models, thus allowing the monitoring of current irrigation practices and crop water status. The proposed methodology was applied in the irrigation district of ‘‘SAT Llano Verde’’, Albacete, Castilla-La Mancha (Spain) characterized by annual horticultural crops, for irrigation seasons (from May to September) 2006, 2007 and 2008, as well as in the irrigation district 1/A, managed by “Consorzio di Bonifica Agrigento 3”, Castelvetrano, Sicily (Italy), characterized by sparse perennial vegetation (mainly olives and grape vines) during irrigation seasons 2009 and 2010. Two satellite-based image-processing SEB models were used for estimating actual ET, i.e., the single source Surface Energy Balance Model for Land (SEBAL) and the Two-source Surface Energy Balance model (TSEB). The first model was applied on both case studies, while the second was applied only to the Italian case study, because of its suitability to sparse vegetation. The models were applied to quantify instantaneously, daily, monthly and seasonal actual ET over the available TM cloud-free images. Successful applications of SEBAL model provided direct estimations of the spatial distribution of the main energy fluxes, at the instant of the satellite overpass. Daily Evapotranspiration values (ETa,ID) were obtained from instantaneous values by assuming an invariable evaporative fraction for the entire day of acquisition. Monthly and seasonal ET values were then estimated from ETa,ID, by assuming that the latter varies in proportion to the reference evapotranspiration (ET0) based on the data acquired by meteorological stations located in both the study-areas. In this way it was possible to account for day to day variations in meteorological forcing. In the case of the SAT Llano Verde irrigation district, the model application allowed the comparison between the monthly water volumes distributed by each hydrant, as measured, with the corresponding actual evapotranspiration volumes, taking into account a value of irrigation efficiency equal to 85%, as indicated by the district irrigation management. This comparison allowed assessing the irrigation performance at both hydrant level and the whole irrigation district. It was shown that the majority of farmers use to apply amounts of water higher than those retrieved by SEBAL, with the exception of May in all investigated years, in which a significant amount of precipitation occurred. The comparison at seasonal time-scale evidenced that a considerable amount of water could have been saved, corresponding to 26.2, 28.0 and 16.4 % of total water consumption, evaluated respectively for years 2006, 2007 and 2008. On the other hand, the application of TSEB did not provide convincing results due to several factors, among which I believe the high sensitivity of this method to uncertainties in satellite radiometric temperature retrievals, as well as to the additional input parameters required to run the model. In the irrigation district 1/A (Italy) the actual ET fluxes retrieved by SEBAL were also compared to the corresponding estimated by the eddy covariance tower located in an olive orchard within the district. The comparison evidenced the general reliability of daily ET retrieved by the model and consequently the validity of the self-preservation hypothesis applied to upscale instantaneous ET values. However, the temporal upscaling between acquisition dates strongly depended on the number of cloud-free available images and, mainly in the Italian case-study, failed to take into account preceding and succeeding rainfall and/or irrigation events. Finally, the comparison between SEBAL outputs and the maximum daily crop evapotranspiration (ETc) estimated with the FAO 56 Penman-Monteith approach for the image acquisition days, evidenced remarkable stress levels on both the investigated crops, with the exception of May during which significant rainfall events occurred in both the considered years.