Mapeo Digital de Carbono Orgánico de suelos en el Ecosistema Herbazal de Páramo, Zona Norte del Ecuador

The objective of the study was to perform a digital mapping of organic soil carbon (OSC) in the grassland ecosystem of the moorland of the northern part of Ecuador. Thus, the soils of 213 points were sampled in the field (0-30 cm and 30–60 cm). Each sample was characterized by its OSC content, geological formation, soil taxonomy, altitude, LS factor and 11 spectral indices. These data were used to train an artificial intelligence algorithm and obtain two decision trees (AD) and two non-linear regression models (MRNL) to predict the OSC. In learning, the AD exhibited an accuracy of 93%, while the regression models a R2 of 0.98. The geological formation and soil taxonomy proved to be the most relevant predictors in the prediction of OSC. Thus, the AD and MRNL were used to perform a digital mapping of the SOC throughout the study area. Finding that the minimum and maximum value of SOC in the study area was 30 and 270 g C / kg soil, respectively. The performance of the regression algorithms was validated by comparing the predicted value of the OSC at a depth of 0 - 30 cm at 60 points taken at random with the value reported on the global carbon map in the world prepared by FAO, finding that the Difference between the value on the global map and the predicted value have a zero mean and a standard deviation of 26 Mg C / ha.

• 1. autor: Acosta López, Winer Orley (author)
• Format: bachelorThesis
• Język: spa
• Wydane: 2020
• Hasła przedmiotowe: PARAMO; ÁRBOL DE DESICIÓN
• Dostęp online: http://dspace.unach.edu.ec/handle/51000/6699