Estimation of spatial and temporal variability of crop water productivity with incomplete data

Abstract

Crop water productivity (WP) in irrigated agriculture is a key for food and environmental security, in particular when water becomes scarce, as has been predicted for downstream regions in the Aral Sea basin. The assessment of WP for each field crop is hampered when farmers cultivate several crops at the same time and on the same fields and cannot record water allocation for each crop. Since results from the commonly used Ordinary Least Squares (OLS) approach turned out to be unreliable and in some cases even had negative values, we combined a behavioral approach with mixed estimation methods to estimate water allocation for each crop over larger areas and over various years with limited data availability. Unobserved crop specific input data was derived from aggregated data using the mixed estimation method for the case study region Khorezm, located in northwestern Uzbekistan. On the basis of actual water usage, spatial (for different administrative districts in Khorezm) and temporal (for the years 2004-2007) distributions of WP for cotton, wheat, rice, vegetables (including melons), and fruits were estimated and visualized through contour diagrams. All crops, except forage, used much more irrigation water than the recommended amount, with cotton and rice being the highest water consumers. For example, cotton was almost 64% over-irrigated compared to the recommended amounts. Even though rice was cropped on a relatively small share of the total land in Khorezm (less than 10% of the total arable land), about 30% of the total crop irrigation water was applied on rice. WP depends on district or farms' location declining proportionally to the distance from the water source, due to high conveyance losses and low yields monitored at the tail ends of the irrigation system. Extremely high water losses on both conveyance and field level revealed much scope for water saving by implementing different water-wise options such as lining canals, introducing best water use practices and producing less water consuming crops. This would, in turn, allow increasing crop yields and WP particularly in downstream districts. Furthermore, the comparison of WP among different crops and different districts allowed determining the potential suitability of certain districts for certain crops, which suggests that a more regionally differentiated cropping portfolio, according to water availability, soil quality and similar parameters, would improve overall system WP, and hence sustainability of the agricultural production.