The automated reference toolset: A soil-geomorphic ecological potential matching algorithm

Soil Science Society of America Journal
By:  and 

Links

Abstract

Ecological inventory and monitoring data need referential context for interpretation. Identification of appropriate reference areas of similar ecological potential for site comparison is demonstrated using a newly developed automated reference toolset (ART). Foundational to identification of reference areas was a soil map of particle size in the control section (PSCS), a theme in US Soil Taxonomy. A 30-m resolution PSCS map of the Colorado Plateau (366,000 km2) was created by interpolating ∼5000 field soil observations using a random forest model and a suite of raster environmental spatial layers representing topography, climate, general ecological community, and satellite imagery ratios. The PSCS map had overall out of bag accuracy of 61.8% (Kappa of 0.54, p < 0.0001), and an independent validation accuracy of 93.2% at a set of 356 field plots along the southern edge of Canyonlands National Park, Utah. The ART process was also tested at these plots, and matched plots with the same ecological sites (ESs) 67% of the time where sites fell within 2-km buffers of each other. These results show that the PSCS and ART have strong application for ecological monitoring and sampling design, as well as assessing impacts of disturbance and land management action using an ecological potential framework. Results also demonstrate that PSCS could be a key mapping layer for the USDA-NRCS provisional ES development initiative.

Publication type Article
Publication Subtype Journal Article
Title The automated reference toolset: A soil-geomorphic ecological potential matching algorithm
Series title Soil Science Society of America Journal
DOI 10.2136/sssaj2016.05.0151
Volume 80
Issue 5
Year Published 2016
Language English
Publisher Soil Science Society of America
Contributing office(s) Southwest Biological Science Center
Description 12 p.
First page 1317
Last page 1328
Google Analytic Metrics Metrics page
Additional publication details