A new method for multicolor determination of organic matter in moist soil

• Published in: Catena (Giessen) Vol. 207; p. 105611
• Main Authors: Lin, Lixin, Gao, Zhiqiu, Liu, Xixi, Sun, Yuan
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.12.2021
• Subjects: Moisture-based multicolor reconstruction (MMR); RGB; Soil color; Soil organic matter; Soil color; Soil organic matter; Moisture-based multicolor reconstruction (MMR); RGB
• ISSN: 0341-8162; 1872-6887
• DOI: 10.1016/j.catena.2021.105611

We investigated whether the moisture-based multicolor reconstruction (MMR) method could reduce the influence of soil moisture for the SOM model established from soil color. This method reconstructed the soil color of moist soil samples based on their soil moisture. Fig. 1. Left half: the influence of soil moisture; Right half: the work reported in this study to develop the moisture-based multicolor reconstruction (MMR) model. SOM, soil organic matter; ANFIS, adaptive neuro-fuzzy inference system. [Display omitted] •Multicolor methods for estimating soil organic matter are limited by soil moisture.•A new method called moisture-based multicolor reconstruction was proposed.•This method can reduce the influence of soil moisture by correcting soil color.•The accuracy for soil organic matter estimated from soil color was improved. Soil color is frequently used by researchers to determine soil properties such as soil organic matter (SOM). However, soil moisture can darken soil color, which seriously limits the accuracy of SOM estimates. This study focused on the influence of soil moisture on the SOM estimate by using our new moisture-based multicolor reconstruction (MMR) method. On the basis of RGB color and moisture values of moist soil samples, the optimal MMR (moist soil) model—root mean square error of validation = 4.537 g/kg, residual prediction deviation of validation = 1.681, ratio of performance to interquartile range of validation = 2.939—was obtained to develop the final model, which performed better than the RGB (moist soil) model. Our method can reduce the influence of soil moisture and increase accuracy for multicolor modeling, and explores a new way to accurately determine the SOM content.