USE OF SPECTRAL RADIANCE FOR CORRECTING NITROGEN DEFICIENCIES AND ESTIMATING SOIL TEST VARIABILITY IN AN ESTABLISHED BERMUDAGRASS PASTURE

The use of variable rate technology has become increasingly popular fo r applying plant nutrient elements. The most widely used method for de termining variable fertilizer rates is presently based on soil testing and yield mapping. Three field studies (Burneyville 1995, Burneyville 1996, and Ardmore 1996) were initiated in established Midland bermuda grass [Cynodon dacrylon (L) Pers.] pastures to determine the relations hip between spectral radiance at specific wavelengths with forage nitr ogen (N) removal and biomass, and to determine field variability of so il test parameters. Variable N (applied to 1.5 x 2.4 m subplots within 2.4 x 45.7 m main plots), fixed N and check treatments were evaluated at each location. Spectral radiance readings were taken in the red (6 71+/-6 nm), green (570+/-6 nm), and near infrared (NIR) (780+/-6 nm) w avelengths. The normalized difference vegetation index (NDVI) was calc ulated as NIR-red/NIR+red. Variable N rates were applied based on NDVI . The highest fixed variable N rate was set at 224, 336, and 672 kg N ha(-1) for Burneyville, 1995, 1996, and Ardmore, 1996, respectively. A t Burneyville, soil samples were collected in all variable rate plots (1.5 x 2.4 m) and analyzed for various soil test characteristics. NDVI , red, green, and NIR spectral radiance readings were correlated with bermudagrass forage N removal and yield, Correlation of forage yield a nd N removal with red, NIR, and NDVI were best with maximum forage pro duction, however, when forage production levels were low correlation d ecreased dramatically for the red wavelength compared with NIR and NDV I. Forage yield and forage N removal in variable rate treatments incre ased when compared to the check while being equal to the half-fixed an d fixed rates where higher N rates were applied. Also, variability abo ut the mean in variable rate plots was significantly lower than half-f ixed and fixed rates which supports adjusting N rates based on indirec t NDVI measurements. Variable N rate plots reduced fertilizer inputs b y 60% and produced the same yield as fixed rate plots, while fixed and half-fixed rates did not increase N content in the forage over that o f the variable rate treatment. Soil sample data collected from small c onsecutive plots (<4 m(2)) was extremely variable indicating that inte nse sampling would be needed if variable fertilizer application were t o be based on soil test results.