Mapping Tropical Dry Forest at High Spatial Resolution in an Andean Region with High Environmental Variability

27 Pages Posted: 19 Feb 2024

See all articles by Jose Camilo Fagua

Jose Camilo Fagua

Universidad Nacional de Colombia

Patrick Jantz

School of Informatics, Computing, and Cyber Systems, Northern Arizona University

Abstract

Tropical Dry Forests (TDFs) are highly threatened. It is estimated only between 6% and 10% of their original cover remains in the tropical countries of South and Central America. Despite this high degree of threat, the mapping of TDF is uncertain and complex due to environmental and taxonomic inconsistencies in its definition. We present a reliable methodology to construct forest maps that discriminate between TDFs and Tropical Moist Forests (TMFs) in an Andean region with high environmental variability, where both types of forests converge. Using one of the most reliable decision-tree learning algorithms (Random Forests-RF), field surveys, environmental data (climate and soil variables), Synthetic Aperture Radar-SAR imagery (Copernicus Sentinel-1), and multispectral imagery (Copernicus Sentinel-2), we built an accurate map at 10m spatial resolution that discriminates between TDF and TMF in the department of Caldas (Colombia). We estimated ~54 km2 of TDF (2.2% of the forest cover), which represents 9.2% of its potential habitat. TDF gradually changes to TMF from two main low altitude river valleys to higher altitudes in the Andes Mountains. Since the phenology of TDFs justifies their distinction as a biome (Dexter et al., 2018) but their environmental limits vary widely (e.g. by climate, soil, and altitude), we hypothesized that TDF-spatial limits can be defined spatially using leaf phenology rather than environmental variables. With modification, our methods could be applied to assess the mapping of TDF in other tropical regions.

Keywords: Andes Mountains, Deciduous leaf phenology, Evergreen phenology, Random Forest, Tropical Moist Forest, ecological gradients.

Suggested Citation

Fagua, Jose Camilo and Jantz, Patrick, Mapping Tropical Dry Forest at High Spatial Resolution in an Andean Region with High Environmental Variability. Available at SSRN: https://ssrn.com/abstract=4731849 or http://dx.doi.org/10.2139/ssrn.4731849

Jose Camilo Fagua (Contact Author)

Universidad Nacional de Colombia ( email )

Carrera 30 # 45 - 03
Bogota, 111321
Colombia

Patrick Jantz

School of Informatics, Computing, and Cyber Systems, Northern Arizona University ( email )

PO Box 15066
Flagstaff, AZ 86011
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
27
Abstract Views
70
PlumX Metrics