Matching regional water supply and demand can be improved by allocating local renewable water resources through decentralized water supply networks (WSNs). The feasibility of decentralized WSNs depends on the costs for the required pipeline infrastructure. The lowest costs for pipeline infrastructure depend on the local landscape characteristics. We present a model that designs decentralized WSNs to supply water with regional supply sources. The objective of the model is to include the effects of landscape characteristics on infrastructure costs and to minimize overall WSN costs. We tested the model on a case study in the fresh-water scarce region of Zeeuws-Vlaanderen in the southwestern part of The Netherlands with known (hydro)geological, geographical and climate data. The model was tested to supply a large industrial water user with groundwater resources operated within sustainable yields. The generated WSNs cover a demand between 0.5 and 5.5 million m3 year-1. Between 1 and 12 supply locations are needed to cover the demand. The pipeline infrastructure needed ranges from 25.1 to 114.5 km. The model determines the optimal pipeline route, the amount of water flowing over each pipeline segment, and reveals if a small increase in demand causes a relatively large increase in costs. The results can be used to determine if water transport is preferred over other water supply options, such as wastewater re-use or desalination of saline water resources.