Modeling cascading drought effects in a national food supply network using minimum cost flow

This paper presents a spatial network optimization approach to model how extreme droughts affect food distribution across Sweden’s national food supply chain. The authors constructed a synthetic, national-scale food supply network with over 830,000 nodes using open geospatial and statistical data from sources including OpenStreetMap, Google Maps, Statistics Sweden, and yield statistics from the Swedish Board of Agriculture. The network was structured into hierarchical layers—production, processing, distribution, retail, and consumers—to capture cascading effects through the system.

Key findings indicate a marked spatial disconnect: agricultural losses were concentrated in southern Sweden, but unmet consumer demand was mainly in northern municipalities. As the network rerouted flows under drought stress, new critical bottlenecks emerged, notably making previously minor hubs highly central, while some established hubs were bypassed. The authors recommend that national food security assessments incorporate both production losses and supply network topology, as optimizing for efficiency can create hidden vulnerabilities and risk secondary cascading failures. The model can help identify emerging bottlenecks for targeted resilience interventions, such as infrastructure reinforcement and buffer-stocking at critical nodes. Future research should validate the model with real-world operational data and integrate temporal and multi-commodity dynamics for more robust risk assessments.