Mapping Vegetation in Fine Detail Across the Americas

Two ambitious mapping projects show how modern vegetation classification can be combined with land cover mapping to describe ecosystems in far more detail than conventional land cover products, according to a study published in Vegetation Classification and Survey.

Natural resource management and biodiversity conservation both depend on inventories of vegetation that span multiple management or political jurisdictions. Although remote sensing data and analytical tools have enabled maps of increasing spatial resolution and reliability, the authors note that there are limited examples where national or continental-scale maps represent vegetation at high thematic detail.

The study, led by Patrick J. Comer and colleagues, presents two case studies: a national mapping effort covering the United States and a continental effort spanning North and South America. Together, these studies cover conditions from subpolar to tropical latitudes of the Americas.

Both case studies used a supervised modeling approach based on the International Vegetation Classification (IVC) to produce maps with greater thematic detail. Georeferenced locations of vegetation types are used by machine learning algorithms to train a predictive model and generate a distribution map.

The United States case study draws on LANDFIRE, the Landscape Fire and Resource Management Planning Tools Project. According to the authors, it illustrates how a history of vegetation-based classification and the availability of key inputs can come together to generate standard map products covering more than 9.8 million square kilometers, which they describe as unsurpassed anywhere in the world in terms of spatial and thematic resolution.

That achievement comes with a clear constraint. Mapping at the thematic resolution of the IVC Group and finer requires very large and spatially balanced inputs of georeferenced samples. Even with extensive prior data collection efforts, the authors report that these samples remain a key limitation.

The second case study is a NatureServe effort for the Americas, an area encompassing 22 percent of the global land surface. The authors report that this project demonstrates methods and outputs suitable for worldwide application at continental scales.

The authors conclude that continued collection of the input data used in these case studies could enable mapping at these spatial and thematic resolutions around the globe. For conservation planners and land managers, the work suggests that bridging the gap between traditional land cover mapping and modern vegetation classification is technically feasible, provided that the underlying field-referenced sample data continue to grow.

Subject of Research: Integrating vegetation classification with land cover mapping at national and continental scales in the Americas.

Article Title: Integration of vegetation classification with land cover mapping: lessons from regional mapping efforts in the Americas

Article References: Comer, P. J., Hak, J. C., Dockter, D., & Smith, J. (2022). Integration of vegetation classification with land cover mapping: lessons from regional mapping efforts in the Americas. Vegetation Classification and Survey, 3, 29-43. https://doi.org/10.3897/vcs.67537

Image Credits: AI Generated

DOI: 10.3897/vcs.67537

Keywords: vegetation classification, land cover mapping, International Vegetation Classification, LANDFIRE, remote sensing, biodiversity conservation, machine learning, NatureServe, Americas, map accuracy, Integration, vegetation