Quantifying of soil organic carbon and total nitrogen through near infrared spectroscopy on forest plantations in Sabah, Malaysia
Understanding the physiochemical properties of soil is critical in ensuring the optimal growth of crops and trees in forest plantation. The utility of near infrared (NIR) spectroscopy has proven as an effective and inexpensive method for soil properties evaluation. Nevertheless, it requires a cal...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/116964/ http://psasir.upm.edu.my/id/eprint/116964/1/116964.pdf |
| Summary: | Understanding the physiochemical properties of soil is critical in ensuring the
optimal growth of crops and trees in forest plantation. The utility of near infrared
(NIR) spectroscopy has proven as an effective and inexpensive method for soil
properties evaluation. Nevertheless, it requires a calibrated technique that must
be developed to account for the heterogeneity of specific regions. Thus, this
study aimed to develop NIR spectral calibrations for quantitation of soil total
organic carbon (TOC) and total nitrogen (TN). Soil samples were collected from
three sites in Sabah, Malaysia, namely Pitas, Kota Marudu, and Tawau districts.
Soil TOC and TN were obtained from 143 soil samples that were collected,
scanned, and analysed using standard chemical method and NIR spectroscopy.
Partial least squares regression (PLSR) analysis using cross validation was
performed and then validated using independent sample set. Calibrations were
generated utilising only the A-horizon samples due to inadequate sampling size
to be used as an independent validation set. The predictions of TOC and TN
concentrations by the PLSR method were statistically sound, with high
coefficients of determination for the validation set (R2
pred TOC =0.77 and R2
pred
TN =0.53) and low root mean square error of prediction (RMSEPTOC =0.44 and
RMSEPTN =0.051). The implementation of these calibrations to offer spatial
evaluation of two contrasting micro-sites within a same area (AFI site) was also
demonstrated. The study also demonstrates the potential utility of NIR
spectroscopy predictions in swiftly and non-invasively characterizing site-soil
attributes, thereby presenting it as an important decision support tool in
managing forest plantations, particularly when combined with NIR spectroscopic
measurement of foliar nutrient levels. |
|---|