Comparison of Satellite Imagery and Ground-Based Active Optical Sensors as Yield Predictors in Sugar Beet, Spring Wheat, Corn, and Sunflower

1 The original use of remote sensing using infrared photography for yield variation was conducted by Colwell (1956). Since the launch of the Landsat 1 imaging satellite in 1972 (Mulla, 2013), satellite imagery has been widely used in agriculture for yield prediction and most lately for site-specific N management. Bhatti et al. (1991) used Landsat imagery and auxiliary data to estimate wheat yield potential. On 11 Feb. 2013, NASA launched Landsat 8, which provides moderate-resolution imagery from 15 to 100 m and operates in the visible, near-infrared, short wave infrared and thermal infrared spectrums (NASA, 2013). Additional satellite imaging systems include: SPOT (SPOT, 2016), MODIS (MODIS, 2016), QuickBird (QuickBird, 2015), RapidEye (RapidEye, 2016), GeoEye (GeoEye, 2016), WorldView (WorldView, 2016), and NOAA-AVHRR (NOAA-AVHRR, 2016). Newer satellite remote sensing systems provide higher spatial resolution, a greater variety of spectral bands, and higher revisit frequency. Satellite imagery in the visible and near infrared (NIR) bands is useful only during daylight when cloud cover does not interfere with the groundcover image. Shou et al. (2007) found that the individual red, green, and blue spectral band reflectance values from QuickBird satellite imagery were highly correlated with winter wheat total N concentration and aboveground biomass. Although in small areas individual spectra can be related to crop biomass, differences in angle of light reflection and other interferences led to the use of vegetation indices, especially normalized differential vegetation index (NDVI), derived from satellite imagery for greatly improved relationships to crop biomass and its use to predict crop grain yield (Rouse et al., 1973) and N status. The NDVI formula is: