Identifying optimal spectral bands to assess soil properties with VNIR radiometry in semi-arid soils

Abstract

Mediterranean soils are critically endangered by natural and human induced degradation processes. Rapid and cost effective techniques for in situ diagnosis of soil properties are highly valuable for soil assessment and management studies. The combined use of remote sensing techniques and Geographic Information Systems (GIS) has introduced a new era for soil resources assessment and monitoring. This work deals with soils in SE Spain and explores the use of spectral derivative analysis to predict several soil properties using field based visible-near infrared (VNIR) spectroradiometry. Robust regression models were obtained for electric conductivity (E.C.), carbonates, soil organic matter (SOM) and sand content with the first and/or second derivative approximation. Future research stages will include evaluating the technique presented in this paper for other semi-arid soils and scaling-up field based regression model predictions for soil properties to the resolution of satellite remote sensors.