Identifying vegetation patterns for a qualitative assessment of land degradation using a cellular automata model and satellite imagery

TL;DR

This paper combines a cellular automata model with satellite imagery to qualitatively assess land degradation by analyzing vegetation patterns, fragmentation, and recovery phenomena related to environmental stress.

Contribution

It introduces a novel approach integrating stochastic cellular automata with satellite-derived vegetation indices to evaluate land degradation and vegetation health.

Findings

Vegetation fragmentation correlates with land degradation states.

Seasonal effects influence vegetation recovery and fragmentation.

Vegetation clusterization indicates environmental stress levels.

Abstract

We aim to identify the spatial distribution of vegetation and its growth dynamics with the purpose of obtaining a qualitative assessment of vegetation characteristics tied to its condition, productivity and health, and to land degradation. To do so, we compare a statistical model of vegetation growth and land surface imagery derived vegetation indices. Specifically, we analyze a stochastic cellular automata model and data obtained from satellite images, namely using the Normalized Difference Vegetation Index (NDVI) and the Leaf Area Index (LAI). In the experimental data, we look for areas where vegetation is broken into small patches and qualitatively compare it to the percolating, fragmented, and degraded states that appear in the cellular automata model. We model the periodic effect of seasons, finding numerical evidence of a periodic fragmentation and recovery phenomenology if the model parameters are sufficiently close to the model's percolation transition. We qualitatively recognize these effects in real-world vegetation images and consider them a signal of increased environmental stress and vulnerability. Finally, we show an estimation of the environmental stress in land images by considering both the vegetation density and its clusterization.