Perishable agricultural supply chains face significant uncertainties in production yield, market demand, and availability of resources. This paper presented a multi-period, multi-echelon, and multi-product mathematical model for a perishable agricultural supply chain network comprising farms, cold storage facilities, and wholesale markets. The model determined optimal allocation of cultivation area, harvest timing, duration of storage, and product flow distribution to minimize total supply chain costs, including production, transportation, storage, and import costs. A key innovation is the integration of interval programming to handle uncertainty in production costs, transportation rates, and water availability. Three metaheuristic algorithms including Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Simulated Annealing (SA) were employed to solve the nondeterministic polynomial (NP)-hard problem. A comprehensive case study of three major perishable crops, i.e., potato, onion, and tomato, across 31 Iranian provinces and 24 candidate cold storage facilities attested to the effectiveness of the model. Results demonstrated that PSO outperformed GA and SA in solution quality by achieving a total cost of 64.14 trillion Rials, with 93.7% of costs attributed to production, 5.2% to transportation, and 1.1% to storage. The model reduced final product prices to one-third of market averages. Sensitivity analyses revealed that production cost was the most sensitive parameter, with a 100% increase triggering economically viable imports. Water availability reduction below 10% of baseline rendered domestic production infeasible. The model achieved 19.9% average water savings (1.062 billion cubic meters annually) through optimized cultivation patterns. The proposed framework provided agricultural policymakers with a quantitative tool for balancing domestic production, investment in storage capacity, and import decisions under uncertainty.