It is hypothesised that spatial patterns of dryland vegetation be used to predict catastrophic shifts, and a lot of research tackled this with simple mathematical models of spontaneous pattern formation and remote sensing morphometric analysis. Among simplifications in this type of work, hydrodynamics of overland flow are neglected. In particular, the redistribution of overland flow between bare soil and vegetated areas on storm timescales in not included.

This study asks about the relationship between the characteristics of individual storms and long term vegetation patterns. To do this, ML emulation of the Saint Venant Equations was used to add a realistic representation of overland flow to a pattern-forming model.

We then assessed how resolving overland flow on storm timescales changed predictions made by a simple pattern-forming model.

To bridge the timescale separation between individual storms and longer-term vegetation growth and development, an overland flow model (the SVE model) was coupled to a dryland vegetation model (the Rietkerk/Guttal model).

This approach makes it possible to directly integrate predictions of overland flow and infiltration into a vegetation model.