PROCEEDINGS ARTICLE | October 2, 2008
Proc. SPIE. 7104, Remote Sensing for Agriculture, Ecosystems, and Hydrology X
KEYWORDS: Mid-IR, Statistical analysis, Spectroscopy, Nitrogen, Reflectivity, Imaging spectroscopy, Vegetation, Infrared radiation, Chemical analysis, Electromagnetism
Imaging spectroscopy can provide real-time high throughput information on growing crops. The spectroscopic data can
be obtained from space-borne, air-borne and handheld sensors. Such data have been used for assessing the nutritional
status of some field crops (maize, rice, barely, potato etc.). In this study a handheld FieldSpec 3 spectroradiometer in
the 350 - 2500 nm range of the electromagnetic spectrum was evaluated for its use to estimate sugarcane leaf nitrogen
concentrations. Sugarcane leaf samples from one variety viz., N19 of two age groups (4-5 and 6-7 months) were
subjected to spectral and chemical measurements. Leaf reflectance data were collected under controlled conditions and
leaf nitrogen concentration was obtained using an automated combustion technique (Leco TruSpec N). The potential
of spectroscopic data for estimating sugarcane leaf nitrogen status was evaluated using univariate correlation and
regression analyses methods with the first-order reflectance across the spectral range from 400 to 2500 nm. The variables
that presented high correlation with nitrogen concentration were used to develop simple indices combining reflectances
of 2-wavelengths. Simple linear regression was then used to select a model that yielded the highest <i>R</i><sup>2</sup>. These were the
R<sub>744</sub> / R<sub>2142</sub> index for the 4-5 months old cane crop and the (R<sub>2200</sub> - R<sub>2025</sub>) / (R<sub>2200</sub> + R<sub>2025</sub>) index for the 6-7 months old
cane crop, with <i>R</i><sup>2</sup> of 0.74 and 0.87, respectively.
