Publication Date:
2022-01-14
Description:
Abstract
Description:
Nitrogen (N) is one of the most essential elements in agriculture and ecology due to its direct role in determining crop yield and grain quality, as well as its association with canopy photosynthetic capacity and carbon-nitrogen cycling in the earth ecosystem. Remote sensing provides a useful way to capture canopy nitrogen and biomass with high spatial and temporal resolution. However, seasonal dynamics of plant morphophysiological variation hinder the simultaneous estimation of canopy N concentration (%N) and biomass using a traditional method such as vegetation indices because of the distinct dynamics of canopy biochemical and physical traits. In contrast, multivariate analysis method offers the capability of calibrating a model with multiple dependent variables of interest. Therefore, the main objective of this study was to, simultaneously, estimate canopy %N and biomass of rice using the partial least squares regression (PLSR) model. A field experiment was conducted for paddy rice fertilized with five N rates across five growth stages in 2008, located in the Sanjiang Plain, China. Results showed that the PLS regression model simultaneously explained 84% and 91% of the variation in %N and biomass, respectively, across the five growth stages. Our results also suggest that biomass is the dominant factor that affects the link between canopy dynamical traits and canopy reflectance measures. This study demonstrates that, by incorporating with PLSR for retrieving biophysical and biochemical properties from the full-spectrum analysis, to what extent canopy %N and biomass can be simultaneously estimated from canopy reflectance measurement.
Keywords:
Nitrogen
;
Biomass
;
Hyperspectral
;
Remote Sensing
;
Agriculture
;
550 Earth sciences
Language:
English
Type:
Text
,
Workshop paper
Format:
5 Pages
Format:
1130 Kilobytes
Format:
application/pdf
Permalink
