Scaling up irrigation for food-system resilience under abrupt sunlight reduction: A global optimisation analysis

Abrupt sunlight-reduction scenarios (ASRSs), such as nuclear winter or massive volcanic eruptions, could suppress temperatures and incoming solar radiation for several years, sharply reducing agricultural productivity and creating a severe threat of global food insecurity. Prior literature indicates that expanding cultivated area can offset a portion of these yield losses, but no estimates exist for the contribution of new and scaled-up irrigation infrastructure to food security under ASRS conditions, whether on existing croplands or on lands suitable for expansion. We quantify the food-security benefit of groundwater-based irrigation investment under a 150-Tg soot scenario using a global 1° gridded framework, with yields and water requirements for six key staple crops drawn from crop simulation models. We assess irrigation investment opportunities on both existing rainfed croplands and suitable expansion areas, allocating budgets sequentially to maximise the marginal caloric return per unit of investment. Early, low-cost gains along the supply curve are dominated by rainfed production on existing cropland and continued legacy irrigation, with new pivot irrigation entering mainly at higher budgets. Expanding cultivation onto herbaceous, shrub, barren, and young-forest land tends to be more expensive per calorie produced, yet becomes increasingly important once lower-cost opportunities are exhausted. At the $8.935 trillion budget, total modelled production is equivalent to feeding about 10.4 billion people per year; of that, the yield gain from new and upgraded irrigation relative to rainfed production on the same parcels is about 2.8 billion people per year, and produces as much grain as today. Assuming labor and input prices remain unchanged, the marginal cost of matching global staple production at current levels on average under ASRS reaches approximately $611 per tonne of dry carbohydrate equivalent, against