Aerosol-producing global catastrophes such as nuclear war, super-volcano eruption, or asteroid strike, although rare, pose a serious threat to human survival. Light-absorbing aerosols would sharply reduce temperature and solar radiation reaching the earth's surface, decreasing crop productivity including for locally adapted traditional crop varieties, i.e. landraces. Here, we test post-catastrophic climate impacts on barley, maize, rice, and sorghum, four crops with extensive landrace cultivation, under a range of nuclear war scenarios that differ in the amount of soot injected into the climate model. We used a crop growth model to estimate gradients of environmental stressors that drive local adaptation. We then fit genotype environment associations using high density genomic markers with gradient forest offset (GF offset) methods and predicted maladaptation through time. As a validation, we found that our GF models successfully predicted local adaptation of maize landraces in multiple common gardens across Mexico. We found strong concordance between GF offset and disruptions in climate, and landraces were predicted to be the most maladapted across space and time where soot-induced climate change was the greatest. We further used our GF models to identify landrace varieties best matched to specific post-catastrophic conditions, indicating potential substitutions for agricultural resilience. We found the best landrace genotype was often far away or in another nation, though countries with more climatic diversity had better within-country substitutions. Our results highlight that a soot-producing catastrophe would result in the global maladaptation of landraces and suggest that current landrace adaptive diversity is insufficient for agricultural resilience in the case of the scenarios with the greatest change to climate.