@Article{XimenesAmaMonAlmVal:2021:MaTeEc,
author = "Ximenes, Arimat{\'e}a de Carvalho and Amaral, Silvana and
Monteiro, Antonio Miguel Vieira and Almeida, Rodolfo Maduro and
Valeriano, Dalton de Morisson",
affiliation = "{Instituto Nacional de Pesquisas Espaciais (INPE)} and {Instituto
Nacional de Pesquisas Espaciais (INPE)} and {Instituto Nacional de
Pesquisas Espaciais (INPE)} and {Universidade Federal do Oeste do
Par{\'a} (UFOPA)} and {Instituto Nacional de Pesquisas Espaciais
(INPE)}",
title = "Mapping the terrestrial ecoregions of the Purus-Madeira interfluve
in the Amazon Forest using machine learning techniques",
journal = "Forest Ecology and Management",
year = "2021",
volume = "488",
pages = "e118960",
month = "May",
keywords = "spatial ecology, ecoregions, biodiversity, self-organizing map,
landscape ecology, geoprocessing.",
abstract = "An ecoregion is a region with similar environmental conditions.
However, many ecoregions represent regional habitat heterogeneity,
and areas with little fieldwork information can benefit from
ecoregion mapping by providing information about their
biodiversity distribution. This work presents the procedure
adopted to map the terrestrial ecoregions of the Purus-Madeira
interfluve, in the Brazilian Amazon using Machine Learning
techniques. A methodological approach with Self-Organizing Map and
K-means algorithms is proposed for the ecoregion mapping and the
resulting model is discussed. The final ecoregion map was built up
from a set of variables including altitude, slope, drainage
density, percentage of tree cover and a vegetation map.
Discriminant analysis identified the extent to which the variables
are similar or different between the ecoregions, with a Kappa
index of 0.86. This indicates that the methodological approach is
reliable and thus can reproduce valid results over different
areas. We produced a map with 14 ecoregions to account for the
environmental diversity in the Purus-Madeira interfluve. This map
can be used for the planning of biodiversity conservation.",
doi = "10.1016/j.foreco.2021.118960",
url = "http://dx.doi.org/10.1016/j.foreco.2021.118960",
issn = "0378-1127",
language = "en",
targetfile = "ximenes_mapping.pdf",
urlaccessdate = "09 maio 2024"
}