Desarrollo de una metodología basada en reflectancia difusa para determinar carbono orgánico en suelos de la cuenca del chambo

The objective of this research is the development of a methodology based on diffuse reflectance to determine the organic carbon in the soil (SOC) and the type of soil (Mollisol and Andosol). To achieve our objective, 106 soil samples were collected in the field and 4240 spectra of diffuse reflectance were recorded in the range of 350 nm to 2500 nm. The spectra were segmented in windows or spectral predictors of 10, 20, 30, 40 and 50 nm. Five classes of carbon and two of soil were defined. With this information, 200 decision trees (AD) were generated. The set of 30 nm predictors led to an AD with 6 nodes and 5 relevant predictors that allowed to predict the COS with an overall accuracy of 95%. Likewise, the set of 50 nm predictors led to an AD with 5 nodes and 4 relevant predictors that allowed predicting soil type with an overall accuracy of 97%. In the prediction of COS it was found that the most relevant predictors are the accumulated reflectance values in the spectral ranges that start at 380, 650, 770 and 920 nm. While in the prediction of the type of soil, the ranges starting at 400 and 1850 nm are important. Diffuse reflectance measurements have allowed us to identify spectral regions of interest in the prediction and quantification of COS and differentiate between Andosoles and Molisoles, facilitating the development of a methodology for the COS prognosis.

• Autor principal: Unach (author)
• Altres autors: Sinaluisa Pilataxi, Bryan Alexander (author)
• Format: bachelorThesis
• Idioma: spa
• Publicat: 2019
• Matèries: CARBONO ORGÁNICO; ANDOSOL; MOLLISOL; ÁRBOL DE DESICIÓN
• Accés en línia: http://dspace.unach.edu.ec/handle/51000/5652