Login | Request Account (DAF staff only)

Rapid assessment of soil carbon and nutrients following application of organic amendments

Share this record

Add to FacebookAdd to LinkedinAdd to XAdd to WechatAdd to Microsoft_teamsAdd to WhatsappAdd to Any

Export this record

View Altmetrics

Farrar, M. B., Wallace, H. M., Tahmasbian, I., Yule, C. M., Dunn, P. K. and Hosseini Bai, S. (2023) Rapid assessment of soil carbon and nutrients following application of organic amendments. CATENA, 223 . p. 106928. ISSN 0341-8162

Full text not currently attached. Access may be available via the Publisher's website or OpenAccess link.

Article Link: https://doi.org/10.1016/j.catena.2023.106928

Publisher URL: https://www.sciencedirect.com/science/article/pii/S034181622300019X

Abstract

Regular and routine soil analysis is vital to properly understand agricultural soil evolution and sustainably manage nutrient applications over time. Hyperspectral imaging (HSI) is a fast-emerging technology and has shown potential to rapidly detect soil nutrient levels. However, it remains unknown to what extent processing soil samples and applying organic amendments, such as biochar, to soil will affect HSI model prediction accuracy and reliability. This study aimed to: 1) compare prediction accuracy of models developed using images of sieved and ground soils; and 2) assess prediciton reliability for samples where biochar had been applied. Specifically, we developed partial least squares regression (PLSR) models to predict soil total carbon (TC), nitrogen (TN), water extractable phosphorus (P), potassium (K) and sodium (Na) using pooled data collected over six years. Additionally, prediction reliability using best-fit models for samples with nil biochar were compared to samples where biochar was applied. Best-fit PLSR models were developed using images of ground soils and improved prediction of TN (R2p from 0.78 to 0.84; ratio of performance deviation (RPD) from 1.74 to 2.37) and Na (R2p from 0.78 to 0.83; RPD from 1.97 to 2.23) compared with using images of sieved soils. In general, prediction accuracy of TC, TN, P and Na was influenced by processing by grinding, whereas this was not the case for K. Images of ground soils predicted reference values more accurately due to reduced particle size and light scattering, resulting in increased mean reflectance and sample homogeneity. Biochar addition to individual samples did not impede prediction reliability in the external test dataset. Reliable TN and Na prediction occurred in 79.4% and 84.3% of samples containing biochar, and 73.7% and 78.4% with no biochar, respectively. Therefore, hyperspectral imaging represents a promising tool to rapidly predict nutrients in agricultural systems where organic amendments and biochar have been applied.

Item Type:Article
Business groups:Animal Science
Keywords:Biochar Biochar fertiliser Hyperspectral imaging Organic amendments Partial least squares regression (PLSR) Soil nutrients Soil sample processing
Subjects:Agriculture > Agriculture (General) > Agricultural chemistry. Agricultural chemicals
Agriculture > Agriculture (General) > Soils. Soil science
Agriculture > Agriculture (General) > Soils. Soil science > Soil chemistry
Live Archive:16 Jan 2023 02:09
Last Modified:16 Jan 2023 02:09

Repository Staff Only: item control page