Conclusions

The modelling approach described in this review combines data from the AQUASTAT database, such as harvested irrigated areas and crops, cropping patters and intensity, as well as different elements of the climate in order to assess the amount of water diverted and required for irrigation. The development of "irrigated crop calendars" for each country, based on FAO's knowledge of the countries' agriculture is probably one of the most sophisticated ways to ensure reliable assessment of irrigation water requirements.

The robustness of this model is now well recognized and it has been used for a number of significant FAO exercises, such as the World agriculture: towards 2015/2030; an FAO perspective, the World agriculture: towards 2030/2050 (FAO, 2006), the "World agriculture: towards 2050/2080" (FAO, 2011a), and more recently the State of the world's land and water resources for food and agriculture or SOLAW (FAO, 2011b). The accuracy of the approach has also been validated by national statistics that became available in the meantime. This review however has the additional benefits of updated data as well as a handmade meticulous selection and correction of individual data, based on expert judgment.

Worldwide, over 307 million hectares are currently equipped for irrigation, of which 304 million hectares are equipped for full control irrigation and 261 million hectares are equipped for full control and actually irrigated. From these actually irrigated areas and thanks to the higher cropping intensity (CI) permitted by irrigation, over 346 million hectares of irrigated crops are harvested (meaning a global CI of 133 percent), the total irrigation water requirements of which account for 1 500 km3. To meet these requirements, 2 673 km3 are withdrawn (from primary and secondary renewable water resources, fossil groundwater and non-conventional sources of water), resulting in a water requirement ratio of 56 percent. However, both the methodology’s improvement (see last bullet points of the Discussion) and the geographical coverage expansion since the previous modelling exercise (from 90 to 167 countries) prevent from attributing the progress of this ratio (which was 38 percent in 2000, the previous exercise) to refined irrigation or water management.

Despite the fact that this water requirement ratio is used in this review as a tool for estimation of irrigation water withdrawal rather than as a result as such, it could represent the weakness of this analysis at country level, both because of the scarce information available and for conceptual reasons (Perry and Kite, 2003). Conceptually, used at such scale water requirement ratio cannot differentiate between consumptive and non-consumptive flows, productive and unproductive use, and recoverable or unrecoverable flows (Perry et al., 2009). In particular it cannot accommodate recoverable return flows and the unproductive consumptive flows which return to the rivers or the aquifers—either from where it originates or another. In large irrigation schemes or areas, it is therefore certain that the amount of water ‘lost’ in conveyance or in drainage from irrigated fields is reused downstream and that the irrigation scheme ratio or basin ratio can therefore be much higher than field ratio. The problem is particularly relevant in cascade irrigation of paddy rice in areas like South Eastern Asia or with significant use of agricultural drainage water like in Egypt. This study cannot distinguish between basin, scheme and field water requirement ratios by lack of detailed information on the type of irrigation schemes and of geo-referenced water withdrawal data. Another drawback originating from this ratio is the difficulty to account for withdrawal of secondary freshwater, in particular when agricultural drainage water is discharged directly in canals and thus usually not metered. Besides, information regarding the irrigated crops' yield is lacking to account for water productivity. Also water requirement ratio does not consider at all the benefits associated with water (Gleick et al, 2011) such as better water quality, healthier ecosystems or regular agricultural production. Nonetheless, the regional analysis performed in this review tends to erase countries' specificities reflected in the ratio and allows for a meaningful analysis.

Finally, the AQUASTAT team would always give preference to reliable and coherent national statistics over any of the modelled or estimated data obtained in this review. As soon as such data become available, results of this review will be replaced in the database. In addition, estimated and modelled data are proposed to supplement the information already in the database, and will never be used for further estimation or modelling.