Developing predictive soil C models for soils using quantitative color measurements

摘要

Rapid low cost methods to quantify soil C concentrations are needed to support local through global resource inventories. Color is a key indicator of many soil properties, with a strong linkage between darkness and soil organic matter (SOM), making it an important indicator for soil taxonomy, soil quality, and fertility. This study investigated relationships between quantitative measurements of soil color and C in similar to 1900 forest soil samples, representing a wide range of soil development, parent material (PM), and C concentrations. Utilizing a hand held chromameter and the CIELAB color space, soil darkness (L) was employed as a continuous predictor of C in simple models (C similar to darkness) with a slope similar to -0.1 across the entire population. Grouping samples by taxonomy and PM influenced model coefficients with strong correlation in weakly developed soil groups (Inceptisols r = -0.9) and more felsic PM. Soil redness (A) has a strong influence on model performance, altering slope and increasing data scatter. Including redness in multivariate relationships greatly increased fit, aligning model slopes at -0.1 for all PM. Ordinary least squares models reached predictive accuracy of <0.5% (RMSE) for specific soils, certain PM classes, and in samples with <4% C. These results demonstrate the utility of quantified soil color to drive predictive relationships and support data development to refine ecosystem C budgets and quantify soil C credits.