Correlation between digital Landsat levels With Uranium content in soil using Support Vector Machines Regression Algorithms

Volume : IV, Issue : IV, April - 2015

Luis Hern N Ochoa, Luis Hern N Alvarez

Geophysical and Geochemical Field surveying is a hardly and costly activity where remote sensing tools can be a very 

useful tool to increase coverage and resolution in that kind of studies. Here we present the results obtained using a 

Support Vector Machine Regression model that relates contents of Uranium measured in laboratory to field soil samples 

withlocation coordinates and digital Landsat levels at the sample site. Soil samples were taken during a geochemical field survey made by 

the "InstitutoColombiano de Geología y Minería – INGEOMINAS“in the Vichada and Guainía Departments – Colombia in 2006. The maximum 

correlation factor obtained was 0.67 which is quite low but indicates the presence of an important relation between Uranium content and digital 

Landsat levels. This research can be considered as an important contribution to the classical geostatistical models because let’s to increase model 

resolution in areas where sampling density is low getting valuable information from inaccessible areas using remote sensing higher spatial 

variabilit