Utilizing observations of vegetation patterns to infer ecosystem parameters and test model predictions

Periodic vegetation patterns arise globally in arid and semi-arid environments, and are believed to indicate competing positive and negative feedbacks between resource availability and plant uptake at different length scales. The patterns have become the object of two separate research themes, one focusing on observation of ecosystem properties and vegetation morphology, and another focusing on the development of theoretical models and descriptions of pattern behavior. Given the growing body of work in both directions, there is a compelling need to unify both strands of research by bringing together observations of large-scale pattern morphology with predictions made by various models. Previous attempts have employed spectral analysis on pattern images and inverse modeling on one-dimensional transects of patterns images, yet have not made a concerted effort to rigorously confront predictions with observational data in two dimensions. This study makes the first steps towards unification, utilizing high resolution landscape-scale images of vegetation patterns over multiple years at five different locations, including Niger, Central Mexico, Baja California, Texas, and Australia. Initial analyses of the observed patterns reveal considerable departures from the idealized morphologies predicted by models. Pattern wavelengths, while clustered around a local average, vary through space and are frequently altered by pattern defects such as missing or broken bands. While often locally homogeneous, pattern orientation also varies through space, allowing the correlations between landscape features and changes in local pattern morphology to be explored. Stationarity of the pattern can then be examined by comparing temporal changes in morphology with local climatic fluctuations. Ultimately, by identifying homogeneous regions of coherent pattern, inversion approaches can be applied to infer model parameters and build links between observable pattern and landscape features and the inferred ecosystem and hydrological properties.; Vegetation patterns near Fort Stockton, TX, with regions of coherent pattern morphology highlighted. Areas of yellow soil with green vegetation bands represent regions of pattern coherence. Areas of light gray soil with dark gray vegetation bands represent regions that do not meet criteria for pattern coherence.