Combining spectral ranges for soil discrimination: A case study in the State of Maranhão - Brazil

Proximal sensing is a tool of relevance to pedology, as it provides suitable and quick information about soil properties. The majority of the studies focus on the stand-alone use of proximal sensors in different wavelengths of the spectrum to get information on soil properties. Although sensors work on a specific spectral range, the combined use of sensors from different spectral ranges makes it easier to explore several kinds of information about the soil. This study aimed at carrying out a discriminant analysis of soil profiles, through the evaluation of data from different spectral ranges along the electromagnetic spectrum. The spectral ranges used included the X-ray (Portable X-Ray fluorescence (pXRF)), visible, near and short-wave infrared (Vis-NIR-SWIR, 350–2500 nm), and mid-infrared (Mid-IR, 2500–25,000 nm) ranges. The study had five main steps: 1) collecting soil samples; 2) acquiring spectral data; 3) identifying the most important spectral ranges for soil discrimination; 4) grouping analysis of soil profiles by color and spectral behavior by cluster analysis; and 5) characterizing and discriminating each group by their spectral behavior. The clustering using the combined spectral ranges (Vis-NIR-SWIR and Mid-IR) joined the soil profiles due to their spectral similarity. The qualitative analysis of the spectral curves allowed us to understand which were the soil properties that influenced the grouping by spectrum. The methodology used in this work was effective for soil discrimination, in terms of soil color, particle size distribution, mineralogy and drainage conditions. The combined use of Vis-NIR-SWIR and Mid-IR showed high efficiency in surveying and detailing information about soil profiles, contributing to its characterization and discrimination. •Similar profiles were grouped by their soil spectral response.•The spectral response was influenced by color, mineralogy, SOC, and texture.•Vis-Nir-SWIR and Mid-IR spectral ranges provided the best results.•Proximal sensors are a powerful tool to soil descrimination.