Supervised Learning Algorithms Applied in the Zoning of Susceptibility by Hydroclimatological Geohazards

Abstract

Cenit Transporte y Logística de Hidrocarburos (CENIT), operator of about 7000 km of hydrocarbon transport systems, which constitutes it the largest operator in Colombia, has developed a strategic alliance to structure an adaptive geotechnical susceptibility zoning using supervised learning algorithms. Through this exercise, has been implemented operational decision inferences with simple linguistic values. The difficulties proposed by the method considers the hydroclimatology of Colombia, which is conditioned by several phenomena of Climate Variability that affect the atmosphere at different scales such as the Oscillation of the Intertropical Convergence Zone - ITCZ (seasonal scale) and the occurrence of macroclimatic phenomena such as El Niño-La Niña Southern Oscillation (ENSO) (interannual scale). Likewise, it considers the geotechnical complexity derived from the different geological formation environments, the extension and geographical dispersion of the infrastructure, and its interaction with the climatic regimes, to differentiate areas of interest based on the geohazards of hydrometeorological origin, when grouped into five clusters. The results of this exercise stand out the importance of keep a robust record of the events that affect the infrastructure of hydrocarbon transportation systems and using data-guided intelligence techniques to improve the tools that support decision-making in asset management.