U.S. flag

An official website of the United States government


Main content area

Determining soil properties in Amazonian Dark Earths by reflectance spectroscopy

Araújo, Suzana Romeiro, Söderström, Mats, Eriksson, Jan, Isendahl, Christian, Stenborg, Per, Demattê, JoséA.M.
Geoderma 2015 v.237-238 pp. 308-317
plateaus, anthropogenic soil types, soil organic carbon, cation exchange capacity, soil map, calcium, exchangeable calcium, models, least squares, reflectance spectroscopy, soil analysis, pH, nutrients, prediction, phosphorus, Amazonia, Brazil
In the Brazilian Amazon patches of anthropogenic soils known as Amazonian Dark Earths (ADEs) occur. These soils are rich in carbon (C) and plant nutrients compared to the naturally occurring strongly weathered soils. In this paper we explore the potential of visible to near infrared (vis–NIR) and mid infrared (MIR) spectroscopy as an alternative to traditional soil analysis of ADE properties for predicting and assessing spatial distributions. We also test whether partial least square regression (PLSR) models generated from soil data at one ADE site can serve as a basis for predictive assessments of soil characteristics at another. The study was carried out at two locations on the Belterra Plateau, Pará state, Brazil, each including soils that displayed typical ADE characteristics. Laboratory analyses confirmed the occurrence of general properties typical of ADE: elevated pH, phosphorus (P), exchangeable calcium (Ca), cation exchange capacity (CEC), and soil organic carbon (SOC) in the study areas. For C and CEC, MIR models were more efficient than those based on vis–NIR (R2=0.90 and 0.82 vs. 0.72 and 0.63). The soil maps produced from the PLSR models adequately described the spatial pattern of SOC, CEC, and Ca. However, useful models of soil P could not be produced. We conclude that spectroscopy can be useful for assessing the spatial distribution of some of the most important ADE properties. MIR spectroscopy models in particular have the potential to be an alternative to traditional soil analysis.