Determining Winter Wheat Stand Densities Using Spectral Reflectance Measurements

Remote sensing would be a valuable tool to decision making and crop forecasting groups to gain knowledge of winter wheat acreage, extent of winterkill, and the potential acreage that may be reseeded to spring crop. To simulate winterkill, seven winter wheat (Triticum aestivum L.) stand variables of 0, 10, 20, 40, 60, 80, and 100% of full stand (full stand = seeding rate of 67 kg/ha) were established by seeding to stand. The plots, 20 ✕ 20 m, were located on Williams loam (fine-loamy mixed, Typic Argiborolls) near Sidney, Mont. An Exotech Model 100-A ground truth radiometer, corresponding to LANDSAT multispectral scanner (MSS) bands 4, 5, 6, 7, (0.5 to 0.6, 0.6 to 0.7, 0.7 to 0.8, and 0.8 to 1.1µm), was used to measure reflected radiation at 0930 hours MST on clear or nearly clear days. We chose the comparatively simple, but physically meaningful normalized difference vegetation index, VI7 = (MSS7 − MSS5)/(MSS7 + MSS5), with the variant of substituting MSS6 for MSS7 to illustrate crop cover differences. In an attempt to better distinguish soil background from crops and to improve on the crop-vegetation index relationship, we included the more complex, but also physically meaningful, perpendicular vegetation index, PVI = [(Rgg5 − Rp5)² + (Rpp7 − Rp7)²]¹/² with the variant of substituting MSSG for MSS7. Early in the season, VI7 delineated stands ≥40% from those <40%, allowing a judgment as to whether or not a field would need to be reseeded to spring crops. A winter wheat survival of about 40% is generally considered adequate to provide a reasonably good stand if plants are properly spaced. The 10 and 20% plots became distinguishable from the soil background at a biomass accumulation of about 30 kg/ha (LA1 ≃ 0.06). There was no advantage of using either variant of the PVI over the VI7 when plotted against leaf area index (LAI) or leaf dry matter (LDM). Both yielded a family of straight lines, with indices of determination ranging from 0.796 to 0.942, and slopes decreasing from the 10% plot to similar slopes for the 80 and 100% plots.