Document Type

Thesis - University Access Only

Award Date

1998

Degree Name

Master of Science (MS)

Department / School

Agronomy

Abstract

Near infrared reflectance spectroscopy (NIRS) is routinely used for chemical analysis of forages and other agricultural crops. Principles of NIRS are to statistically relate reflectance spectra with sample composition. Portable devices are now marketed that permit precise detection of reflected solar radiation. This instrumentation may allow for remote and nondestructive analysis of plants in the field. The objective of this research was to develop NIRS equations for analysis of pasture quality and forage crops based on remotely gathered solar reflectance spectra. Three types of established perennial grass mixtures were examined; cool-season species, warm-season species, and a mixture of both cool- and warm-season species. Cool-season species included orchardgrass (Dactylis glomerata L.), intermediate wheatgrass (Thinopyrum intermedium Host, Barkworth and D. R. Dewey), and creeping foxtail (Alopecurus. arundinaceus Poir.). Warm-season species were big bluestem (Andropogon gerardii Vitman) and switchgrass (Panicum virgatum L.). In April 1996, four nitrogen (N) fertility levels (0, 381, 749, and 1, 135 gN m-2 ) were applied to subplots in all three replicates ofeach grass mixture. On 24 June, 31 July, and 4 September1996, five solar reflectance spectra were collected in eachsubplot with a field radiometer (350-2500 nm). Herbagesamples were collected from 0.2 m2 quadrats for each spectraand dried at 60 ° C. Out of a total of 540 field-basedspectra, 226 were used for calibration development.Calibration equations were developed for crude protein (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF) and carbon isotope discrimination (􀁘). Results from field based calibration statistics were very promising. For forage quality traits, SECV values ranged from 13 g kg- 1 for

CP to 32 g kg- 1 for NDF and 3.7 °/oo for 􀁘- These levels of error represent 16, 5, 8% of the grand mean for CP, NDF, and ADF, respectively, and 35% for 􀁘- The highest R2 and 1-VR values occurred for CP; 0.91 and 0.80, respectively. The NIRS calibration equations were validated with known values from 274 samples. Standard errors of prediction were 23, 33, and 29 g kg- 1, for CP, NDF, and ADF, respectively, and 5.5 °/ 00 for 􀁘- These results suggest a field radiometer may be used for nondestructive analysis of grass canopies using NIRS development techniques to determine chemical composition.

Library of Congress Subject Headings

Forage plants -- Analysis Pasture plants -- Analysis Near infrared reflectance spectroscopy

Format

application/pdf

Number of Pages

104

Publisher

South Dakota State University

Download

Share

COinS