Crop Vegetation Nitrogen Status - Sensitive Signatures for Proximal Remote Sensing.

Rationale. Nondestructive methods for detecting vegetation nitrogen concentration generally depend on change in reflectance or transmittance of radiation in the Visible-Near Infrared  (VNIR) range of 400-1100 nm. These methods tend to depend on change in chlorophyll concentration or to the red-edge, both of which respond to the statuses of various nutrients besides nitrogen and to various environmental stresses. In this study, we sought to develop a method for detecting vegetation nitrogen status using reflectance in the Short-Wave Infrared (SWIR) range of 1100-2500 nm. The SWIR contains bands affected by protein level, but is not strongly affected by chlorophyll. Most nitrogen in rice leaves is found in protein. Methods. Ten rice genotypes of diverse origin, including temperate and tropical japonicas, indicas, and NERICA in replicated field research plots at Beaumont, Texas, USA. Wide range of nitrogen fertility: 0, 112, or 224 kg N/ha.  SWIR reflectance collected from individual leaves using Ocean Optics NIRSpec with controlled light source. Leaf reflectance spectra and leaf N concentration related using two opposite approaches: (1) All Possible Ratios and (2) Contiguous waveband clustering followed by Partial Least Squares Regression (PLSR). Results & Discussion. The cluster of ratios of wavebands of best fit included a waveband associated with protein. The highest-loading cluster of the PLSR is influenced by protein. Most protein bands were relatively high loading in the PLSR. The predictions based on the PLSR using the top six factors explaining variation in N concentration provided both good precision and accuracy. A good linear relationship between SWIR reflectance and leaf N concentration was obtained over a wide range of leaf N concentration suggesting the ability to detect leaf N concentrations at levels in which remediation practices can be applied before economic damage occurs.