Multivariate split moving windows and magnetic susceptibility for locating soil boundaries of Sa∼o Paulo, Brazil

Multivariate split moving window (MSMW) is a tool to automate soil mapping and to assess uncertainty in soil boundaries. In this paper, we propose a new approach to locate soil boundaries. We investigated the potential of (i) MSMW associated with Mahalanobis D2 and (ii) split moving window (SMW) associated with magnetic susceptibility (MS) as tools for validation of soil boundaries. A transect was lined in Guatapar ' a city (Sao Paulo state, Brazil), and 172 soil samples were collected from 86 locations, at a depth of 0-25 and 25-50 cm, for physical and chemical analyses. Additionally, categorical properties - geology, land use, and altitude - were assessed at the same sampling points. All the data were organized into four groups of properties (G1, G2, G3, and G4) and analyzed by principal component analysis, MSMW analysis - to delineate map units using D2 -, and SMW analysis using MS. MSMW and SMW were compared regarding their potential to locate soil boundaries along the transect. The MS peaks in SMW presented a correlation with peaks of t-statistics and D2 in MSMW: (r = 0.56; p < 0.01 - r = 0.69; p < 0.01) and (r = 0.55; p < 0.01 - r = 0.64; p < 0.01) for both depth intervals (0-25 cm and 25-50 cm), respectively. MSMW was more sensitive than SMW in the detection of soil boundaries in areas with lower clay content. Compared to conventional soil surveys, MSMW considerably improved the prediction of boundaries in mapping units. We concluded that the proposed method is a promising strategy for soil surveyors and can be used to assist pedological cartography.

Automated summary

The multivariate split moving window (MSMW) is proposed as a tool to automate soil mapping and assess uncertainty in soil boundaries. The method was compared with split moving window (SMW) associated with magnetic susceptibility (MS) for validation of soil boundaries. MSMW was found to be more sensitive than SMW in detecting soil boundaries in areas with lower clay content, providing an improved prediction of boundaries in mapping units.