The proposed methodology does not depend on optical data and, therefore, can create irrigation maps irrespective of cloud cover. As such, the proposed method can complement other existing detection schemes based on optical data. The tool can monitor the progress of measures that support water savings through no water use at all. In other words, the tool can point to previously irrigated parcels that became non-irrigated because of policy support to activities that do not use irrigation, such as conversion to rain-fed cultivations, adoption of cultivars with earlier planting, setting the land aside and others. The tool can also support controls and inspections since the irrigation/non-irrigation maps provide very early detection of irrigation at the plot level for assessing farm compliance against obligations or conditionalities.
The same maps can support an evaluation in three ways. First, they can be used, together with other data sources and other EO tools, in estimating irrigation water needs which is a proxy for the ‘water use in agriculture’ impact indicator. Second, when combined with IACS/LPIS information, they can support the evaluation of agricultural policy measures to reduce water use. Third, the irrigation/non-irrigation map can cross-validate information related to policy effects on crop irrigation and its consequent impacts on water use in agriculture.
The tool is at the proof-of-concept stage of development and was successfully validated at the study region of Riaza in Castilla y León, Spain. A critical step of the tool’s application is the training stage, when the AI algorithms ‘learn’ to recognise an irrigated field for various crops and periods of the year. Learning depends on in situ reference data from field observations, irrigation registries or cadastral maintained by irrigation water associations. Therefore, transferring and validating the tool to other regions and Member States is feasible.
