Author
CHENG, CHEN-WEN - IOWA STATE UNIVERSITY | |
Laird, David | |
MAUSBACH, MAURICE - USDA, NRCS | |
HURBURGH, C - IOWA STATE UNIVERSITY |
Submitted to: Agronomy Abstracts
Publication Type: Abstract Only Publication Acceptance Date: 10/22/1998 Publication Date: N/A Citation: N/A Interpretive Summary: Technical Abstract: Near-infrared reflectance spectra of 725 soil samples collected from 4 Major Land Resource Areas (MLRA) were studied to evaluate the effect of calibration procedures on the accuracy of NIRS-soil analysis. Locally weighted principle component regression (PCR) was used to correlate reflectance spectra with chemical and biochemical soil quality parameters. Preselection of potential calibration sets based on sampling area (MLRA, and State), sampling depth (surface and subsurface), and vegetation type were tested. The results indicate that the preselection procedures did not significantly affect the accuracy of the locally weighted PCR technique, as long as samples from the same MLRA were included in the potential calibration set (R2 of %C varies from 0.73 to 0.85; R2 of %N varies from 0.66 to 0.80). However, the accuracy of the PCR technique was decreased if samples from the same MLRA were excluded (R2 of %C is 0.55; R2 of %N is 0.52). |