Functional soil mapping for site-specific soil moisture and crop yield management Q. Zhu a, b, ⁎, H.S. Lin b , J.A. Doolittle c a State Key Laboratory of Lake Science and Environment, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, 73 East Beijing Road, Nanjing 210008, China b Department of Crop and Soil Sciences, 116 ASI Building, The Pennsylvania State University, University Park, PA 16802, United States c USDA–NRCS, National Soil Survey Center, 11 Campus Blvd., Suite 200, Newtown Square, PA 19073, United States abstract article info Article history: Received 28 March 2012 Received in revised form 1 February 2013 Accepted 7 February 2013 Available online 20 March 2013 Keywords: Geophysics Hydropedology Electromagnetic induction Precision agriculture To advance site-specific management, detailed functional (application-orientated) soil maps are desirable. This study presents an example of a functional soil map for the management of crop yields and soil moisture in an agricultural landscape located in central Pennsylvania, USA. A high-intensity soil map was prepared using electromagnetic induction (EMI), terrain attribute, and soil core data. Two soil properties, A horizon texture and depth to clay layer were found to be significantly correlated (p b 0.05) with soil moisture and crop yields (corn and soybean). A functional soil map was generated by overlaying these two properties on the high-intensity soil map. Spatial and temporal variations in soil moisture and crop yield within different functional map units were statistically compared with those within different map units on the second-order and the high-intensity soil maps. While soil moisture and crop yield did not show obvious difference among map units of the second-order soil map, clear differences were observed among the functional soil map units developed in this study. This study demonstrates that the accuracy and utility of second-order soil maps for site-specific management can be improved using EMI and functional soil mapping approaches. © 2013 Elsevier B.V. All rights reserved. 1. Introduction Soil maps provide the foundation for our understanding of soil– landform relationships and soil variability across different landscapes. Traditional soil maps are created using conceptual models of soil variation across landscapes. These models are generalized from direct field observations, remotely-sensed data, and tacit knowledge of soil, landform, geology, vegetation, and land use (Dijkerman, 1974; Hudson, 1992). In preparing these maps, direct field observations are made at a very limited number of points in the landscape that are selected to confirm interpretations and conceptual models (Dijkerman, 1974; Hudson, 1992). Second-order soil maps (with a cartographical scale of 1:12,000 to 1:31,680, and a minimum delineation size of 0.6–4.1 ha) have been developed widely in the United States using this soil mapping approach (Soil Survey Division Staff, 1993). However, traditional methods result in qualitative delineations that produce broad schemes to separate and classify the soil continuum, and do not adequately convey the quantitative nature of soil variation (Cook et al., 1996). While highly valuable for general land-use planning and many other applications, second-order soil maps have encountered considerable challenges when used for more precise applications in soil-landscape studies, precision agriculture, hydrology, and ecosystem studies at the catchment- or field-level scales (e.g., Franzen et al., 2002; Lin et al., 2005a,b; Robert, 1993). In recent years, quantitative pedologic models have provided more intensive and quantitative approaches for mapping and modeling the spatial distribution of soils and soil properties across landscapes. Examples include the use of environmental correlation modeling (e.g., McKenzie and Ryan, 1999; Park and Vlek, 2002), landscape- guided soil mapping (e.g., Heuvelink and Webster, 2001), and fuzzy- logic approaches (Zhu et al., 2001). In these approaches, terrain attributes and land use/land cover are used to partition the landscape and provide more site-specific information on soil variation. However, soil properties and core observations were generally not directly used in these quantitative soil mapping models and approaches because soil sampling and analysis were time consuming and costly. Noninvasive geophysical tools provide copious, direct or indirect data on soil physiochemical properties as input parameters in soil mapping. Electromagnetic induction (EMI) is a geophysical tool that is used to measure the soil apparent electrical conductivity (ECa) (Johnson et al., 2001; Sudduth et al., 2001). This tool offers a fast and convenient method to collect large quantities of soil-related data. Since ECa is affected by a number of soil properties including clay content and mineralogy, soil moisture, and salinity, it has been widely used as a surrogate to quickly and comprehensively map the spatial distribution of soils and soil properties (e.g., Auerswald et al., 2001; Corwin and Lesch, 2005; Rhoades et al., 1976; Robinson et al., 2009; Saey et al., 2009; Zhu and Lin, 2009). The use of EMI to differentiate soil type and identify soil boundaries has been documented in several studies. Anderson-Cook et al. (2002) used maps of ECa and crop yield to accurately locate soil Geoderma 200–201 (2013) 45–54 ⁎ Corresponding author at: State Key Laboratory of Lake Science and Environment, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, 73 East Beijing Road, Nanjing 210008, China. Tel.: + 86 25 86882139; fax: + 86 25 57714759. E-mail address: qzhu@niglas.ac.cn (Q. Zhu). 0016-7061/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.geoderma.2013.02.001 Contents lists available at SciVerse ScienceDirect Geoderma journal homepage: www.elsevier.com/locate/geoderma