Multipolarized radar for delineating within-field variability in corn and wheat

In agriculture, there is growing interest in determining field spatial variability for implementing differential management practices, which should generate economic and environmental benefits. To date, the majority of studies involving remote sensing and differential management have focused on optical sensor systems. Less attention has been paid to synthetic aperture radar (SAR), despite the advantages of "all-weather" acquisition enabling information to be collected under cloud cover. This study examined the information content of multipolarization (HH, HV, VV, RR, LL, RL), multitemporal, and multiangle radar for delineating within-field spatial variability. On three dates in 2001, airborne C-band SAR data (35° and 55° incident angles) were acquired over four experimental fields. A series of fuzzy K-means analyses showed that the ability to differentiate zones was dependent upon the crop, the date in the growing season, and the pedodiversity of the field. Consistent with the soil and plant biophysical data, two of the four fields showed no spatial variability in radar backscatter. In the high-pedodiversity cornfield (Zea mays L.), three zones of productivity were discriminated early in the growing season and two zones of productivity in mid-season. Late in the season as a result of saturation of the radar signal, no spatial variability was evident. In corn, the results were similar regardless of the radar polarization. In the wheat (Triticum aestivum L.) field, which was of lower pedodiversity, two zones were identified in early-, mid-, and late-season images. Differences were evident among polarizations, with VV and HV being most sensitive to within-field variation. The delineated zones in both fields were shown to relate to plant and soil parameters, suggesting that radar may be a valuable tool in delineating spatial variation in producer fields and delineating differential management units.