Using high spatial resolution spectral data to study spatial and temporal variability in corn and soybean management systems

In the last two decades, with the advent of satellite technology and fast computers, the application of remote sensing in agriculture has gained importance. In the past, most studies in the application of remote sensing to agriculture have used spatial resolutions ranging from 1.1-Km to 10-m. The main purpose of this study was to evaluate and characterize high spatial resolution (1-m) spectral data to study spatial and temporal variability of corn and soybean management systems.; In this study, three types of reference data sets were acquired to help calibrate/interpret multispectral (12-band) data sets that were acquired using a mechanical scanner aboard an aircraft. Multispectral images were acquired 5 times in 1994, targeted to pick differences in land cover and crop growth development. The reference data included: (1) biweekly field samples consisting of soil and plant information; (2) high accuracy ("{dollar}<{dollar}5-cm") Digital Elevation Model (DEM) depicting the micro-topography of a 50-ha crop field; and (3) on-the-go yield data for field variability studies.; Both remotely sensed images and on-the-go yield map were co-registered with the DEM for multi-layer spatial analysis in a Geographic Information Systems (GIS) environment. Temporal spectral responses of landcovers were developed. The relationships between spectral and reference data were determined. The spatial structures of spectral data were investigated using correlograms.; The relationships between soil-plant-water system variables and spectral data depended on the land cover type and degree of cover at the time of image acquisition. Corn and soybean crops were spectrally separated at all dates. Open spaces and shadows in a corn canopy contributed for the low reflectance observed. Temporally pooled data sets showed improved correlation than individual dates (r = 0.92 vs. r = 0.77 for soybeans; r = 0.84 vs. r = 0.70 for corn). Correlation between spectral data and the combined water from the soil and plants decreased beyond a thin layer (0.5-cm) of soil. Mean yield class values correlated well (r = 0.99) with the corresponding mean spectral values. Yield classes were also related to micro-topography differences (r = 0.92). Spectral bands with the highest spatial auto-correlation at shorter spatial lags contained most of the spectral information.