
Lanhui Wang
Researcher

Accurate quantification of soil organic matter content using VNIR-SWIR spectra : The role of straw and spectrally active materials
Author
Summary, in English
Soil organic matter (SOM) is crucial for carbon sequestration and sustainable agriculture, yet traditional quantification methods are challenging to apply at large scales. Hyperspectral technology combined with machine-learning offers promising prospects for rapid quantification. This study explores the impact of using VNIR-SWIR spectra on SOM quantification in regions characterized by distinctive soil properties and agricultural activity. Specifically, we propose an innovative approach using 105 soil samples from Yueyang City, China, to refine the range of spectrally active materials and evaluate the effectiveness of iron oxides and straw on SOM quantification. Three feature construction methods (conventional (VNIR-SWIR spectra), optimal (information spectrum subset, ISS), and straw-merged ISS (SISS)) and seven models were employed to evaluate the contributions of iron oxides and straw in SOM quantification. The results indicate that the SISS improved the generalization (RPD and R2) of nonlinear and linear models by approximately 9 % and 4 %, respectively. The relative contributions of straw and iron oxides in modelling are approximately 35 % and 10 %, respectively. Our research successfully developed the SISS by refining the range of spectrally active materials and considering the background formed by the soil properties of the study area. We used it to evaluate the impact of straw on SOM quantification and demonstrated that the spectroscopic characterization of SOM can assess the carbon sequestration benefits of agricultural activities. This approach can be applied to regions with similar soil properties globally, offering a new perspective for SOM quantification.
Department/s
- Dept of Physical Geography and Ecosystem Science
Publishing year
2024-12
Language
English
Publication/Series
Geoderma Regional
Volume
39
Document type
Journal article
Publisher
Elsevier
Topic
- Soil Science
Keywords
- Anthrosols
- Gleysols
- Hyperspectral
- Machine learning
- Soil organic matter
- Spectrally active materials
- Straw
Status
Published
ISBN/ISSN/Other
- ISSN: 2352-0094