We present a decomposition strategy for mixed-integer linear programming (MILP) models that are formulated on the basis of geographic information system (GIS) data. Our algorithm relies on decomposing the MILP into two levels, a master problem and a slave problem between which we iterate until a termination criterion is satisfied. The former is constructed using a K-clustering statistical aggregation method that reduces the computational burden of the model. The solution of this level is used to guide the search in the slave model. A case study that addresses the optimal design of sewage sludge amendment in Catalonia (NE of Spain) is introduced to illustrate the capabilities of the proposed approach.