Fifty years of better irrigation models did not make the global withdrawal number converge
WHAT THE EVIDENCE SAYS
A Nature Water analysis compiled 1,624 global irrigation water-withdrawal estimates from 65 studies published between 1974 and 2025 and evaluated them across 360 plausible analytical pathways. Only 17% of pathways showed declining uncertainty over time, falling to 10% when projections to 2100 were excluded; increasing uncertainty was the most common pattern at 40%, followed by unstable trends at 34%.
WHY IT MATTERS / OUR READ
For decision support, more detailed models do not automatically justify narrower confidence. The practical lesson is to preserve uncertainty explicitly—especially when global withdrawal numbers are used to frame water stress, irrigation-efficiency targets or food-system risk—rather than letting a single point estimate become more authoritative than the evidence supports.
CAVEATS / UNCERTAINTY
The study measures convergence and spread among published estimates, not error against a known true global withdrawal volume. Its conclusions depend on the assembled literature and legitimate analytical choices, so the result is best read as an epistemic warning about precision rather than proof that every newer irrigation model is individually less accurate.
SOURCE
Nature Water
View original evidence ↗
An analysis of 1,624 global irrigation-water-withdrawal estimates found that only 17% of plausible analytical pathways showed uncertainty shrinking over time.
Read the Wild Lands assessment →