Models for the Computational Design of Microfarms.

On microfarms, diversified crops are typically cultivated as densely as possible, and their heterogeneous growth dynamics require complex design and management of crop rotations. Models in computational agroecology may assist farmers in the maintenance of such farms. We use a phenomenological model in which the plants are represented as disks with a growing radius. We show how it can account for the variability observed in the fields and help visualize the spatiotemporal patterns involved in a microfarm cultivated according to the French intensive method. In the last part of the paper, we consider monoculture and a mixture of two vegetables with different maximum radii and growth rates. In the planning strategy, the planting positions and times are randomly chosen, taking inspiration from models of random sequential adsorption. We describe two different event-driven algorithms (1D and 2D) to simulate the dynamics of this system. The steady state of the field consists of disordered configurations of the plants. We study the evolution of the effective planting rate as a function of the nominal planting rate. When the model plants both vegetables with equal probability, simulations show that the proportion of big plants starts decreasing above a given threshold. This model and the algorithms describing the planting strategies may be extended to more species and other planting strategies to suggest original farm designs.