This past quarter at Stanford, I took on a research project (CEE 299) that pulled together two things I care a lot about: water sustainability and agriculture. The question driving it: how much room does the world actually have to grow more food without draining its water resources dry?
Agriculture already uses more freshwater than any other human activity, and as the planet warms and populations grow, the pressure to produce more food isn't going away. But "just irrigate more" isn't a free lunch. Water is unevenly distributed, and pushing irrigation past what a region can sustainably support just trades a food problem for a water crisis. I wanted to know, country by country, where there's real headroom to intensify agriculture sustainably, and where there isn't.
This work fed into a bigger effort called the Natural Capital Index (NCI), which is trying to score how close a country's land management is to the most efficient possible tradeoff between agricultural output and things like carbon storage, nutrient cycling, and biodiversity. Picture a pareto frontier: a curve showing the best possible outcomes you can get for agriculture production versus, say, carbon storage. A country sitting right on that curve is using its land about as efficiently as it can. One sitting far off it has room to improve.
My specific piece was figuring out how to filter that frontier by water availability. Not every point on the curve is realistic. If it takes more irrigation than a country can sustainably supply, it's not really an option.
I pulled freshwater consumption data by country from Rosa et al. (2018), which calculates what it would take to close current agricultural yield gaps under sustainable and unsustainable irrigation scenarios. I paired that with irrigation area data from Portmann, Siebert & Döll (2010) to calculate an irrigation application rate, essentially how many millimeters of water per year a country is applying per unit of irrigated land, for three scenarios: current sustainable, current unsustainable, and future sustainable.
One key modeling choice: I applied the water-availability filter only to the intensification scenario, not expansion. That's because when agriculture expands into forested land, it's usually rainfed, and evapotranspiration actually drops compared to forest cover. Intensifying existing cropland, on the other hand, is where added water demand really shows up.
The United States stood out. Its current irrigation application rate is 601 mm/year, but the future sustainable rate comes out to 1,601 mm/year, a 2.6x increase. That means there's substantial room to intensify irrigation in the US without tipping into unsustainable territory, assuming the infrastructure and practices catch up.
Other countries told different stories. China's future sustainable rate (145 mm/yr) barely budges from its current sustainable rate (137 mm/yr), suggesting it's already operating close to its water ceiling. Russia showed the most dramatic jump, from 185 mm/yr to nearly 2,995 mm/yr, hinting at a huge amount of underused sustainable capacity.
This approach doesn't account for the downstream effects of converting forest to cropland, things like lost carbon storage and biodiversity impacts, that could offset some of the water gains. Future work on the NCI project would need to fold those tradeoffs back into the optimization rather than treating water as the only constraint. But even as a first pass, this gives a much clearer, country-level picture of where sustainable agricultural intensification is actually possible.
From Closing the Yield Gap While Ensuring Water Sustainability (Rosa et al 2018), we have taken the data listed in Table S4. This data shows the freshwater consumption by country for sustainable and unsustainable crop production. With this data, we were able to calculate the amount of sustainable and unsustainable water available per country per year.
Table S4. Freshwater consumption by country under current and maximized crop production.