Global irrigation contribution to wheat and maize yield

The rising global population and increasing demand for food necessitate substantial increases in crop production. Expanding cropland into forested areas carries significant environmental consequences, making sustainable intensification crucial. Climate change further complicates the challenge by impacting crop yields at regional and global scales. Improving irrigation is a key strategy to enhance yields in water-limited regions and improve resilience to climate variability. Despite the acknowledged importance of irrigation for cereal yields, its contribution at global scales remains uncertain. Previous studies, based on hydrological models and climate analogues, have yielded varying estimates of yield gains from irrigation. The benefit of irrigation varies considerably with climatic conditions, necessitating a global-scale analysis of this relationship. This study addresses this knowledge gap by integrating climate analogue (CA) and global gridded crop model (GGCM) approaches within a Bayesian framework to analyze the relative differences between irrigated and rainfed yields for wheat and maize.

Two main approaches have been used to estimate the impact of irrigation on crop yields: climate analogues (CA) and process-based crop models. The CA approach analyzes yield data from censuses and surveys, classifying climatic zones and irrigation extent to estimate attainable yields with and without irrigation. While this approach implicitly accounts for various interacting factors, its spatial extrapolation relies on simplified climatic indices and neglects soil characteristics. Global gridded crop models, on the other hand, offer mechanistic representations of crop growth, considering soil characteristics and benchmarked against field data. However, they may have limitations in representing the diversity of crop varieties, management practices, and soil properties. This study leverages the complementary strengths of both approaches.

This study integrates climate analogue (CA) and global gridded crop model (GGCM) data using Bayesian Model Averaging (BMA). The CA data is updated from Mueller et al. (2012), while GGCM data comes from the Global Gridded Crop Model Intercomparison project, incorporating results from 10 global gridded crop models. BMA is employed to combine the CA and GGCM results without making strong assumptions about driving factors, leading to improved estimates compared to either approach alone. The performance of the reanalyzed data is evaluated against an independent dataset of irrigated and rainfed yields from the US Department of Agriculture (USDA). Spatial analyses investigate the relationship between the relative difference in yields (ΔY) and climatic variables (precipitation and potential evapotranspiration). Finally, the study compares irrigation demands required to achieve the estimated ΔY with available river discharge to assess the potential for sustainable yield gap closure.

The Bayesian reanalysis shows that at the global scale, irrigation increased wheat yields by 34 ± 9% and maize yields by 22 ± 13% compared to rainfed yields. However, there is significant spatial variation in this impact. For wheat, the largest yield increases from irrigation were observed in semi-arid and subtropical regions, whereas in major wheat-producing regions with sufficient precipitation, the benefit of irrigation was limited. For maize, high yield increases were observed in semi-arid and summer-dry regions. Partial correlation analyses revealed that spatial variation in ΔY is primarily driven by precipitation rather than potential evapotranspiration, indicating that water availability is more critical than water demand in determining the impact of irrigation. A comparison of irrigation water demands with river discharge showed that a substantial portion (30–47%) of contemporary rainfed wheat and maize cropland lacks sufficient water resources to achieve yield gap closure using current river discharge alone, highlighting the need for additional water management strategies.

The findings highlight the significant but spatially heterogeneous contribution of irrigation to wheat and maize yields globally. While irrigation significantly boosts yields in water-limited regions, the availability of sufficient water resources is a major constraint. The dominance of precipitation over evapotranspiration in determining the spatial variation of ΔY underscores the importance of water supply rather than water demand in maximizing the benefits of irrigation. The substantial area where irrigation demands exceed available river discharge raises concerns about the sustainability of yield gap closure through irrigation alone. This necessitates the consideration of alternative strategies such as improving irrigation efficiency and exploring cross-basin water transfer projects.

This study provides a comprehensive global assessment of irrigation's contribution to wheat and maize yields, integrating diverse data sources and advanced statistical methods. The results reveal significant spatial heterogeneity in irrigation's impact, with water resource limitations posing a considerable challenge to sustainable yield enhancement. Future research should focus on improving irrigation practices, exploring innovative water management strategies, and further investigating the economic and social dimensions of irrigation expansion. More precise estimation of yield differences between irrigated and rainfed agriculture can better inform policy and decision-making in the nexus of hydrology, agriculture, and economics.

The analysis has several limitations. The water budget calculations are based on annual averages, neglecting spatial and seasonal variations in river discharge. Hillslope constraints on water availability for hillside croplands are not considered. The study only considers wheat and maize, neglecting other irrigation-demanding crops. The estimation of rainfed cropland area may underestimate the area needing additional irrigation. The analysis assumes that all river discharge can be used for irrigation without accounting for environmental flow requirements. The study does not consider groundwater as a source of irrigation water.