Skip to main content
ARS Home » Plains Area » Las Cruces, New Mexico » Range Management Research » Research » Publications at this Location » Publication #341997

Title: Drawing a representative sample from the NCSS soil database: Building blocks for the national wind erosion network

item Levi, Matthew
item WEBB, NICHOLAS - New Mexico State University

Submitted to: ASA-CSSA-SSSA Annual Meeting Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: 8/30/2017
Publication Date: 10/22/2017
Citation: Levi, M.R., Webb, N. 2017. Drawing a representative sample from the NCSS soil database: Building blocks for the national wind erosion network [abstract]. ASA-CSSA-SSSA Annual Meeting. October 22-25, 2017, Tampa, Florida. pg. 31-8.

Interpretive Summary:

Technical Abstract: Developing national wind erosion models for the continental United States requires a comprehensive spatial representation of continuous soil particle size distributions (PSD) for model input. While the current coverage of soil survey is nearly complete, the most detailed particle size classes have categorical of sand, silt, and clay. The National Cooperative Soil Survey (NCSS) database has laboratory data for over 21,000 pedons, but continuous PSDs are not available. Existing soil maps provide the link between NCSS pedon data enable the transfer of detailed laboratory data to spatial representations. Extracting a representative sample from the existing NCSS data for expensive, detailed PSD analysis requires the ability to capture maximum variance with the minimal number of samples. We performed a sensitivity analysis of sample selection to compare the effect of different soil and geographic features on representation of the larger database. A conditioned Latin Hypercube sample design was used to select samples based on their feature space and compared to simple random samples. Comparisons of extracted sample datasets showed little difference for bulk density, water content (field capacity, wilting point, plant available water) and sand fractions. Stronger differences emerged with fewer samples in the dataset. Spatial coverage of all samples effectively represented the geographic coverage of the continental US. This exercise provides insight for how distributions of the PSDs may vary by selection method and also illustrates the ability of different sample sizes for representing the larger database. Our work will inform future sample selection from existing soil databases for modeling at continental scales.