Satellite images tell tales of changing biodiversity

Analysis of texture differences in satellite images may be an effective way to monitor changes in vegetation, soil and water patterns over time, with potential implications for measuring biodiversity as well, according to new research published by CHANS-Net member Matteo Convertino, of the University of Florida, and colleagues.

The authors designed statistical models to estimate two aspects of biodiversity in satellite images: the number of species in a given region or species richness, and the rate at which species entered or were removed from the ecosystem, a parameter the scientists called species turnover.

The paper, Inferring Species Richness and Turnover by Statistical Multiresolution Texture Analysis of Satellite Imagery, was published in the open access journal PLOS ONE.

The researchers tested their models on data gathered over 28 years in a water conservation area in the Florida Everglades and compared their results to previous reports from the region. They found that their models were nearly 100 percent accurate when predicting species turnover; conventional methods only have 85 percent accuracy.

According to the authors, their automated method using satellite images could help improve the efficiency and decrease the cost of campaigns that monitor biodiversity and guide policy and conservation decisions.

"Texture-based statistical image analysis is a promising method for quantifying seasonal differences and, consequently, the degree to which vegetation, soil, and water patterns vary as a function of natural and anthropic stressors," said Convertino. "The application of the presented model to other fields and scales of analysis of ecosystems is a promising research direction.''