Discussion: limitations of the study

The geographical coverage of the irrigated crop calendars, their corresponding irrigation water requirement calculations and estimation of irrigation water withdrawal detailed in the previous pages, consists of 167 countries (including two territories) where most irrigation is practiced worldwide. The 32 remaining countries (as well as all other territories), are excluded due to the lack of data available on areas equipped for irrigation and irrigated crops either:

• Countries where irrigation is assumed to be inexistent (because of climatic conditions or limited areas): Faroe Islands, Iceland, Holy See, Liechtenstein, Luxembourg, Monaco, San Marino; or
• Countries where irrigation may be practiced but where not sufficient data is available for the analysis: Andorra, Bahamas, Bosnia and Herzegovina, Comoros, Cook Islands, Croatia, Dominica, Equatorial Guinea, Fiji, Kiribati, Maldives, Marshall Islands, Micronesia (Federated States of), Nauru, Niue, Palau, Papua New Guinea, Samoa, Sao Tome and Principe, Singapore, Solomon Islands, Tokelau, Tonga, Tuvalu, Vanuatu.

For countries included in this study, some limitations exist in the way crop calendars are prepared, irrigation water requirement modelled and water requirement ratios estimated (some already mentioned earlier in this review):

• In the absence of national data for irrigated harvested crops, some crop calendars are based on World agriculture: towards 2030/2050 (FAO, 2012), which are estimations only.
• Exact location of the different types of irrigated crops in the country is unknown (and thus the exact climate conditions under which they are cultivated). However because the areas equipped for irrigation are geo-referenced in the Global map of irrigation areas, their location can be equally distributed among those, reducing thus the uncertainty associated with their climatic conditions, but not where what crop is grown.
• Data on AAIfull are easily confused with data on AEIfull or are often unavailable. In such cases, AAIfull is assumed similar to AEIfull. In countries with a large part of mobile irrigation equipment (towable system, traveller spray, etc.), such as in high income countries, AAIfull becomes technically AEIfull.
• Despite selection of data in such a way that as much as possible similar years for both the crop calendar (and therefore the irrigation water requirement calculations) and the irrigation water withdrawal are used, out of 118 countries for which water withdrawal data are not estimated, only 52 refer to the same year. For the other countries, accuracy is limited by potential discrepancies between years.
• Although water withdrawal for irrigation was preferred when available over agricultural water withdrawal—which also includes water withdrawal for livestock and aquaculture—, in the absence of specific data for irrigation water withdrawal, the agricultural water withdrawal is assumed to refer only to irrigation.
• As previously mentioned in the corrections applied, the irrigation water requirement is distorted by the use of long-term national average precipitation—in particular for countries with moderate climatic conditions or high inter-annual variability for precipitation. Although actual precipitation data could be used when available, the averages were still preferred for later comparison with the total actual renewable water resources which are also long-term national averages. However, this means that for such countries, individual country ratios might misrepresent the reality if the precipitation for the corresponding year is quite different from the long-term average, even though the countries may fall into the 15-85 percent range considered as coherent.
• Distortion arises also from the assumption that optimal plant growth and thus complete satisfaction of the crop water demand always occurs—again, in particular for countries with moderate climatic conditions or high inter-annual variability for precipitation. This also means that for such countries, individual country ratios might misrepresent the reality if the irrigation water differs widely from the irrigation water requirements, even though the countries may fall into the 15-85 percent range considered as coherent.
• Another limitation of the analysis derives from the use of (long-term average) annual renewable water resources which does not account for monthly variations, especially in monsoon climate, and thus the actual availability of water.
• Finally, and in addition to the above limitations, individual country ratios that are not corrected directly reflect the data gathered (and often provided by the respective governments). Even though differences between neighbouring countries with similar context might sometimes seem difficult to explain. However we deliberately choose to keep as much as possible calculations based on ‘actual’ data (i.e. data obtained from national references and statistics), rather than correcting them by sub-regional averages. We seek to improve the coherency of country ratio as soon as more accurate data are available.

Despite a similar exercise undertaken almost 10 years ago, some improvements in accuracy taken on in this review impede to compare their respective results:

• The irrigated crop calendars are established for AEIfull, as opposed to AEItot in the previous exercise, which includes equipped lowlands and spate irrigation.
• The irrigated crop calendars are established for actual figures of AAI, as opposed to the assumption that AAI is 85 percent of AEI in the previous exercise.
• The irrigated crop calendars are established for the exact year of the irrigated crops data, while it was assigned to the year 2000 for all countries in the previous exercise.
• The water requirement ratios are calculated with water withdrawal data referring to the closest year available (to the crop calendar and thus irrigation water requirement), instead of extrapolated water withdrawals relative to the change in area under irrigation as calculated from FAOSTAT (FAO, 2012b) figures of ‘total area equipped for irrigation’ in the previous exercise. This prevents from any distortion introduced by FAOSTAT yearly figures, which in some cases may be different from the AQUASTAT data.