Submitted to: Geoderma
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: July 17, 2011
Publication Date: August 19, 2011
Repository URL: http://
Citation: Myers, D.B., Kitchen, N.R., Sudduth, K.A., Miles, R.J., Sadler, E.J., Grunwald, S. 2011. Peak functions for modeling high resolution soil profile data. Geoderma. 166(1):74-83. DOI:10.1016/j.geoderma.2011.07.014. Interpretive Summary: Soil physical and chemical properties can be costly to characterize since they can vary significantly with depth and by location in the landscape. This is a critical problem since more soil property data is needed to improve understanding of how variable soils affect plant growth, land-use sustainability, landscape hydrology, and ecosystem services. Compounding this problem is the traditional method of describing soils where similar areas are grouped, named, and described with qualitative terms. Higher resolution methods are needed to quantify soil properties more continuously in three dimensions without significantly increasing the cost of acquiring such data. The objective of this work was to evaluate the use of mathematical “peak” functions to model sensor-measured soil profile properties, and to use the parameters of these functions to quantitatively and continuously describe landscape variation in soil properties. We found two functions that successfully fit clay, silt, and pH data for an example soil profile. One of the functions was further used to model clay depth distribution in soil profiles for different landscape positions. All models fit the data well, and function parameters described differences within the landscape. For example, one parameter indicated the depth in the profile to peak clay content. Another gave the abruptness of the clay peak. Combined these two are important for describing the surface and subsurface flow of water across the landscape. This procedure, along with innovative in-field soil sensing methods currently being developed, offers those who manage soil systems a whole new way of mapping soils. Detailed maps of surface and subsurface properties could be used by farmers and consultants for site-specific inputs and for targeting conservation practices. Site-specific management based on more detailed soil property data will improve cropping efficiency and establish more sustainable food, fuel, and fiber production systems. Higher resolution soil property maps can be used by those who perform numerical modeling or simulations of crops, carbon sequestration scenarios, climate change, hydrology, and water quality to refine predictions in these domains.
Technical Abstract: Parametric and non-parametric depth functions have been used to estimate continuous soil profile properties. However, some soil properties, such as those seen in weathered loess, have complex peaked and anisotropic depth distributions. These distributions are poorly handled by common parametric functions. And while nonparametric functions can handle this data they lack meaningful parameters to describe physical phenomena in the depth distribution of a property such as a peak, an inflection point, or a gradient. The objective of this work is to introduce the use of asymmetric peak functions to model complex and anisotropic soil property depth profiles. These functions have the advantage of parameters which can quantify or describe pedogenic processes. We demonstrate the application of the Pearson Type IV (PIV) and the logistic power peak (LPP) functions to high resolution soil property depth profiles measured by diffuse reflectance spectroscopy in a claypan soil landscape of Northeastern Missouri, USA. Both peak functions successfully fit clay, silt, and pH data for an example soil profile from a summit landscape position (R^2=0.90 for pH and 0.98 for silt and clay). The LPP function was further demonstrated to fit clay depth distribution for a shoulder, backslope, footslope, and a depositional landscape position (R^2=0.98, 0.96, 0.96, 0.91). Relationships in the fitted parameters were useful to describe morphological features in the soil profiles and show promise to continuously describe pedogenic processes in three dimensions. Peak functions are a useful companion to high resolution soil profile data collected by sensors and their combined use may allow more intensive mapping and better explanation of soil landscape variability.