A nonparametric algorithm to model movement between polygon subdomains in a spatially explicit ecosystem model

Many modern spatially explicit ecosystem models use modelling subdomains of different shape and size (‘polygons’) to resolve space, and movement of biomass between them forms an important part of the modelling effort. In marine applications, a flow field grid describing the water movement usually forms the basis for movement of passive or nearly passive biomass. Grid-based advection algorithms are not designed to model movement on the larger scale of polygons, resulting in disproportionately large computational costs and difficult communication between model layers. In this paper, a simple and effective algorithm to model movement at the polygon level is proposed, preserving the general properties of biomass distribution in comparison to a grid-scale model. A nonparametric description of inter-polygon movement in the domain is generated which is used to approximately replicate the observed movement pattern. To estimate the movement description, the moves of passive numerical drifters between polygons are observed. The resulting algorithm outperforms the conventional polygon-based transport equation approach both in artificial and realistic scenarios.