@InProceedings{SilvaFonsKört:2017:SpBaAp,
author = "Silva, Alexsandro C{\^a}ndido de Oliveira and Fonseca, Leila
Maria Garcia and K{\"o}rting, Thales Sehn",
affiliation = "{Instituto Nacional de Pesquisas Espaciais (INPE)} and {Instituto
Nacional de Pesquisas Espaciais (INPE)} and {Instituto Nacional de
Pesquisas Espaciais (INPE)}",
title = "A spatiotemporal Bayesian approach to model deforestation",
booktitle = "Anais...",
year = "2017",
organization = "Workshop dos Cursos de Computa{\c{c}}{\~a}o Aplicada do INPE,
17. (WORCAP)",
keywords = "Spatiotemporal, Bayesian Network, Deforestation.",
abstract = "According to the PRODES, a program that monitors deforestation in
the Brazilian Legal Amazon, deforestation rate has increased in
recent years. It jumped almost 30% over last year representing a
sharp increase over the low rates since 2008. Hence, it is
important to monitor the Brazilian Amazon in order to anticipate
preventive actions and establish public policies. Methods of
Bayesian Network have been increasingly used to model
environmental systems. In this work, we present a computer-aided
spatiotemporal Bayesian network method capable to incorporate
experts knowledge and makes predictions about areas more
susceptible to deforestation over time based on spatial and
temporal data observation. The study area is located in the
southwestern of Par{\'a} state around the Novo Progresso
municipally and along the BR-163 highway. The proposed method
takes into account random variables related to the deforestation,
for instance, land use (proximity to pasture and agricultural
areas can influence deforestation), available infrastructure
(roads can give accessibility to the forests) and others spatial
data. This method is intended as a snapshot, modeling what happens
at a certain time in a spatial area. The time line is divided into
a finite number of time instants. An initial instance of the
Bayesian network is designed, which contains the formulation of
the problem at time t=0, that is, the set of random variables and
their dependency relationships. The structure of this initial
Bayesian network is replicated for each instant. In this way, each
network models the annual deforestation of the study area. As
results, we obtained a spatialization of the areas most
susceptible to deforestation over the time. For each studied year
(2010-2015), one Probability Image was computed, in which each
pixel value represents the probability of such area be deforested.
These resulted images can help stakeholders to control or even
reducing deforestation and to support a sustainable development.",
conference-location = "S{\~a}o Jos{\'e} dos Campos, SP",
conference-year = "20-22 nov. 2017",
language = "en",
targetfile = "Silva_spatiotemporal.pdf",
urlaccessdate = "21 maio 2024"
}