A computational framework is described for optimal spatial allocation of land use categories across a geographical region. The relevance of optimisation, its advantages and disadvantages, and details on mathematical background relevant to natural resource management are discussed.Experiments were conducted with a probabilistic optimisation procedure using a prototype software package, with numerical results and visualisation provided for a case study. Results are provided for land use optimisation with an indication of the possible improvement in regional revenue, given data on crop yields and market prices for various commodities. When revenue is used as the performance index for optimisation, the approach reveals a significant improvement is possible when applied to a production region in Victoria, Australia. A conclusion drawn from the experiments is that the problem is quite complex in nature and open ended with many variables. An agenda is suggested for further development of the modelling strategy, by way of incremental improvements, as more environmental information is acquired and further constraint factors are considered.