Machine learning application to assess deforestation and wildfire levels in protected areas with tourism management
Felipe Roberto da Silva
Samuel Façanha Câmara
Francisco Roberto Pinto
Francisco José da Costa
Leonardo Martins de Freitas
José Gilmar Cavalcante de Oliveira Júnior
Thiago Matheus De Paula
Marcelo de Oliveira Soares
Abstract
This study aims to identify the influence of management plans, management boards and tourism management on the relationship between performance indicators in the management of protected areas and degradation processes. To understand these relationships, 283 protected areas (PAs) in Brazil were analyzed. The first stage of the research used classification models based on machine learning algorithms, which revealed that predictive variables were a promising way to assess PAs vulnerability to deforestation and wildfire, giving decision-makers an 87.5% and 72.8% chance, respectively, of correctly identifying the PAs more susceptible to these threats. The predictive variables more relevant to deforestation were biome, area, and tourism management, while for wildfire, governance and PA type were the most relevant. Predictive variables were also a promising way to assess PAs management, giving decision-makers a 79.7% and 78.1% chance, respectively, to correctly identify the PAs with higher levels of effectiveness and governance. In addition, in the second stage, to empirically reinforce the models, multivariate analyses were performed, through which it was possible to confirm that deforestation levels are significantly higher in areas of sustainable use than in fully protected areas and determine how the positive interaction with tourism management contributes to the reduction in deforestation records and improves effectiveness. Therefore, it is understood that tourism management can strongly influence the sustainability of natural resources, and it is of utmost importance to generate tourism management policies with potential value generation for natural spaces.
Silva F.R.d., Câmara S.F., Pinto F.R., Costa F.J.d., Freitas L.M.d., Oliveira Júnior J.G.C.d., De Paula T.M. and Soares M.d.O. (2023) Machine learning application to assess deforestation and wildfire levels in protected areas with tourism management. Journal for Nature Conservation 74: 126435. 10.1016/j.jnc.2023.126435
@article{Silva2023,
Title = {Machine learning application to assess deforestation and wildfire levels in protected areas with tourism management},
Author = {Silva, Felipe Roberto da and Câmara, Samuel Façanha and Pinto, Francisco Roberto and Costa, Francisco José da and Freitas, Leonardo Martins de and Oliveira Júnior, José Gilmar Cavalcante de and De Paula, Thiago Matheus and Soares, Marcelo de Oliveira},
Editor = {},
Journal = {Journal for Nature Conservation},
Year = {2023},
Pages = {126435},
Volume = {74},
Doi = {10.1016/j.jnc.2023.126435},
Abstract = {This study aims to identify the influence of management plans, management boards and tourism management on the relationship between performance indicators in the management of protected areas and degradation processes. To understand these relationships, 283 protected areas (PAs) in Brazil were analyzed. The first stage of the research used classification models based on machine learning algorithms, which revealed that predictive variables were a promising way to assess PAs vulnerability to deforestation and wildfire, giving decision-makers an 87.5% and 72.8% chance, respectively, of correctly identifying the PAs more susceptible to these threats. The predictive variables more relevant to deforestation were biome, area, and tourism management, while for wildfire, governance and PA type were the most relevant. Predictive variables were also a promising way to assess PAs management, giving decision-makers a 79.7% and 78.1% chance, respectively, to correctly identify the PAs with higher levels of effectiveness and governance. In addition, in the second stage, to empirically reinforce the models, multivariate analyses were performed, through which it was possible to confirm that deforestation levels are significantly higher in areas of sustainable use than in fully protected areas and determine how the positive interaction with tourism management contributes to the reduction in deforestation records and improves effectiveness. Therefore, it is understood that tourism management can strongly influence the sustainability of natural resources, and it is of utmost importance to generate tourism management policies with potential value generation for natural spaces.},
}
TY - JOUR
AU - Silva, Felipe Roberto da
AU - Câmara, Samuel Façanha
AU - Pinto, Francisco Roberto
AU - Costa, Francisco José da
AU - Freitas, Leonardo Martins de
AU - Oliveira Júnior, José Gilmar Cavalcante de
AU - De Paula, Thiago Matheus
AU - Soares, Marcelo de Oliveira
TI - Machine learning application to assess deforestation and wildfire levels in protected areas with tourism management
T2 - Journal for Nature Conservation
PY - 2023
SP - 126435
VL - 74
DO - 10.1016/j.jnc.2023.126435
AB - This study aims to identify the influence of management plans, management boards and tourism management on the relationship between performance indicators in the management of protected areas and degradation processes. To understand these relationships, 283 protected areas (PAs) in Brazil were analyzed. The first stage of the research used classification models based on machine learning algorithms, which revealed that predictive variables were a promising way to assess PAs vulnerability to deforestation and wildfire, giving decision-makers an 87.5% and 72.8% chance, respectively, of correctly identifying the PAs more susceptible to these threats. The predictive variables more relevant to deforestation were biome, area, and tourism management, while for wildfire, governance and PA type were the most relevant. Predictive variables were also a promising way to assess PAs management, giving decision-makers a 79.7% and 78.1% chance, respectively, to correctly identify the PAs with higher levels of effectiveness and governance. In addition, in the second stage, to empirically reinforce the models, multivariate analyses were performed, through which it was possible to confirm that deforestation levels are significantly higher in areas of sustainable use than in fully protected areas and determine how the positive interaction with tourism management contributes to the reduction in deforestation records and improves effectiveness. Therefore, it is understood that tourism management can strongly influence the sustainability of natural resources, and it is of utmost importance to generate tourism management policies with potential value generation for natural spaces.
ER -
