Pedological mapping through integration of digital terrain models spectral sensing and photopedology

Abstract

New tools for soil mapping are needed to increase speed and accuracy of pedological mapping processes.
This study integrated various technologies to map soils of the Piracicaba region in São Paulo State, Brazil. Each
technology was expected to provide different information to design a detailed map. We carried out field survey and soil
sampling for laboratory analysis. Initially, we conducted field visits to obtain soil patterns of a reference site. We applied
the acquired patterns to an validation site, based solely on information obtained from remote sensing and cartographic
databases, namely LANDSAT 7/ETM, digital elevation models (DEM) and aerial photographs. We integrated the
information from each product to generate the map of the validation site, which was validated by field inspection.
Textural classification using satellite imaging ranged from 21-51% of accuracy. Band 5 in the sensor showed the best
performance to discriminate between clayey and sandy soils. Aerial photographs provided the best information because,
besides their own inherent characteristics, they operate on a larger scale and result in a map with up to 50 polygons,
while DEM reached a maximum of 30 polygons. The digital mapping technology generated 45 mapping units. Finally,
the mapping efficiently separated the Latosols from the other classes; however, in some cases there was confusion in the
identification of Cambisols and litholic Neosols.