Premium
Modeling the Spatial Distribution of Soil Texture in the State of Jalisco, Mexico
Author(s) -
Pongpattananurak Nantachai,
Reich Robin M.,
Khosla R.,
Aguirre-Bravo C.
Publication year - 2012
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj2011.0180
Subject(s) - soil texture , spatial variability , kriging , soil map , environmental science , spatial distribution , scale (ratio) , digital soil mapping , soil science , spatial analysis , terrain , multivariate statistics , range (aeronautics) , spatial ecology , soil survey , geostatistics , soil water , cartography , geography , remote sensing , statistics , mathematics , ecology , materials science , biology , composite material
Information on the spatial variability of soil attributes, such as soil texture, is crucial for a wide range of decisions in ecosystem and agricultural research and management. Surveys designed to collect such information across large geographic regions may not capture the fine to moderate scales of variation in soil properties. As a result, soil properties may exhibit spatial dependencies at scales smaller than the scale at which sampling was performed. This lack of spatial structure in the sample data makes it difficult, if not impossible, to use optimal predictors such as ordinary kriging for modeling spatial variability in a data set. We developed a new approach for modeling soil texture fractions across large geographic regions at a fine spatial resolution. The method of three‐stage least squares is used to model the large‐scale variability in soil texture, while the small‐scale variability is modeled using multivariate regression trees. This approach was used to model the spatial distribution of soil textural classes in the Mexican state of Jalisco. Independent variables used in the modeling process included terrain data, climatic data, and satellite imagery. Our results indicate that the sand model accounted for 62% of the variability observed in the sample plots, while the clay model accounted for 56% of the observed variability. Maps of soil attributes obtained from this study can serve as a useful surrogate (i) explaining the spatial variability in soil attributes across large geographic regions and (ii) supporting applications of precision forestry and agriculture for site‐specific management across both small and large geographic regions.