Classification model of gully erosion process along of railway line through decision tree algorithm and geotechnology

Abstract:

The dynamic analysis of erosion process development in the surroundings of the railways are requirements of environmental legislation for the railway sector. In approaches able to support process analysis, monitoring and control of environmental risks, Decision Tree algorithm (DT) associated with the use of geographic information systems (GIS) have been widely used tools. Such studies generate prediction scenarios procedures, probability measures or classifications of land in relation of degree of susceptibility or vulnerability of areas for their occurrence. However, it is assumed that triggering agents of erosion process may present specific features which contribute to the formation of erosive process of types different, patterns and severity. Classificatory studies based on triggering agents and the characteristics of the processes are still rare in the literature. The main objective of this research is the combination of the use of geotechnology and DT for classification of linear erosion located in the surrounding area of ​​the railway Malha Paulista (SP). The analysis sought to understand how they behave and group the triggering agents of erosion in relation of the type of erosive process developed. The results permit classifies the mapped records sample and identify patterns that distinguish the types of linear erosions.

Reference:
SOUZA, Natália da Costa; PITOMBO, Cira; CUNHA, André Luiz; LAROCCA, Ana Paula Camargo; ALMEIDA FILHO, Gerson Salviano de. Modelo de classificação de processos erosivos lineares ao longo de ferrovias através de algoritmo de árvore de decisão e geotecnologias. BCG Boletim de Ciências Geodésicas, v.23, n.1, 15p, 2017.

Access to the article on the Journal’s website:
www.scielo.br/scielo.php?script=sci_arttext&pid=S1982-21702017000100072&lng=pt&tlng=pt

SUBSCRIBE to our newsletter

Receive our news in your email.

INSCREVA-se em nossa newsletter

Receba nossas novidades em seu e-mail.