Skip to main content
ARS Home » Research » Publications at this Location » Publication #139301
Title: USE OF SPECTRAL REFLECTANCE MEASUREMENTS FROM HYPERSPECTRAL RADIOMETRY IN PREDICTION OF ANIMAL GAINS ON PASTURE

Author
item PERKINS, ASHLEY - REDLANDS C.C.
item Brown, Michael
item Starks, Patrick
item APPEDU, LISA - SWOSU
item DILWORTH, CARLA - REDLANDS C.C.
item DAWKINS, DAVID - REDLANDS C.C.
item BILES, CHARLES - EAST CENTRAL U.

Submitted to: Research Day Abstracts: Regional Universities Research Day
Publication Type: Abstract Only
Publication Acceptance Date: 8/1/2002
Publication Date: 10/1/2002
Citation: PERKINS, A., BROWN, M.A., STARKS, P.J., APPEDU, L., DILWORTH, C., DAWKINS, D., BILES, C. USE OF SPECTRAL REFLECTANCE MEASUREMENTS FROM HYPERSPECTRAL RADIOMETRY IN PREDICTION OF ANIMAL GAINS ON PASTURE. RESEARCH DAY ABSTRACTS: REGIONAL UNIVERSITIES RESEARCH DAY. 2002. Abstract p. 82.

Interpretive Summary: Abstract Only.

Technical Abstract: Supplementation of grazing animals with grain or hay without regard to nutrients received from the forage can be inefficient when the supplement is not needed. However, there are times within the grazing season that supplementation can be economically efficient, if those times could be determined in advance. In a research project at the Grazinglands Research Laboratory, spring-born lambs (n=47) were used to evaluate the potential for predicting lamb growth directly from hyperspectral radiometer data taken in bermudagrass pastures grazed by lambs. Lambs were randomized to each of four, 1.6-hectare predominantly bermudagrass pastures so that each pasture was stocked with approximately 12 lambs. At the time initial lamb weights were taken (June 3, 2002), a mobile hyperspectral radiometer was used to measure the spectral reflectance of each pasture in 252 wavebands in the visible through the near-infrared portion of the electromagnetic spectrum (368.4 to 1113.7 nm). Forage spectral reflectance for each field was measured eight times along a line transecting the field and each of the eight samples was subsampled three times. Lambs were weighed again on June 11, June 25, July 9, July 22, and July 31. At the time of each weight, spectral reflectance from the pastures were measured as described above. Animal growth data was estimated in five time periods: June 3 to June 11; June 11 to June 25; June 25 to July 9; July 9 to July 22; and July 22 to July 31. Reflectance data were converted to absorbance by taking the logarithm to the base 10 of the reciprocal of reflectance. Relationships of animal gain to 252 spectral absorbance estimates were investigated by pairing averages for gain and spectral absorbance for each pasture at the beginning of each week or period. Stepwise regression was then performed using the twenty observations (4 pastures x 5 weeks) to identify putative best multiple linear regression models that would account for the largest proportion of total variance in animal gain. A linear combination of ten independent spectral absorbance variables ranging from absorbance at 732.3 nm to absorbance at 1025.6 nm accounted for over 94% of the total variance in animal gains, which should be sufficient for reasonably accurate prediction. While further research is needed to verify these results, it appears that precision supplementation of livestock based on performance predicted from field-level hyperspectral radiometry may be feasible.