Light Reflectance and Remote Sensing of Weeds in Agronomic and Horticultural Crops

Abstract

Plant canopy reflectance over the 0.45- to 1.25-μm wavelength (WL) of weed species and crops was recorded with a field spectroradiometer to evaluate the possible use of remote sensing to distinguish weeds from crops. Weed and weed-crop species reflectance differences were generally greater at the 0.85 μm WL in the near-infrared spectral region than at the 0.55 μm WL in the visible region, indicating that color infrared (CIR) aerial photography may be useful to detect weed populations in crops. Canopy reflectance data were more directly related to photographic differences in weed-crop images than were single leaf or inflorescence reflectance data. Aerial photography at altitudes of 610 to 3050 m distinguished climbing milkweed (Sarcostemma cyancboides♯ SAZCY) in orange [Citrus sinensis(L.) Osbeck. ‘Valencia’) trees; ragweed parthenium (Parthenium hysterophorusL. ♯ PTNHY) in carrot (Daucus carotaL., var.sativa‘Long Imperator’); johnsongrass [Sorghum halepense(L.) Pers. ♯ SORHA) in cotton (Gossypium hirsutumL. ‘CP 3774’) and in sorghum (Sorghum bicolorL. Moench. ‘Oro’); London rocket (Sisymbrium irioL. ♯ SSYIR) in cabbage; and Palmer amaranth (Amaranthus palmeriS. Wats. ♯ AMAPA) in cotton. Johnsongrass was also detectable with CIR film in maturing grain sorghum from 18 290 m. Detection of weed species in crops was aided by differential stages of inflorescence and senescence, and by the chlorophyll content, color, area, intercellular space, and surface characteristics of the leaves. Discrete plant community areas were determined by computer-based image analyses from a 1:8000-scale positive transparency with the efficiency of 82, 81, 68, and 100% for Palmer amaranth, johnsongrass, sorghum, and cotton, respectively. The computer analyses should permit discrete aerial surveys of weed-crop communities that are necessary for integrated crop management systems.