J. H. Everitt, D. E. Escobar, and M. R. Davis

Authors are range scientist, remote sensing specialist, and pilot, USDA-ARS, 2413 E. Highway 83, Weslaco, Tex. 78596.

The authors thank Mario Alaniz for his assistance in obtaining field reflectance measurements and preparation of illustrations, and Jeanne Everitt for her help in word processing.

Manuscript accepted: April 15, 2000

Keywords: remote sensing, color-infrared photography, conventional color photography, noxious brush and weeds

Invasive brush and weeds on rangelands are a primary deterrent to effective management of these areas (Scifres 1980). Many rangelands in the southwestern U.S. support an excessive cover of woody plants and weeds, a problem that must be managed for successful range animal production and for best use of the land for other purposes (Smith and Rechinthin 1964, Scifres 1980). Determining the extent of noxious plant populations on rangelands by ground surveys is difficult because of the generally great expanse and inaccessibility of these areas. The value of remote sensing techniques for rangeland assessment is well established (Carneggie et al. 1983, Tueller 1989, Driscoll et al. 1997). Remote sensing techniques offer the advantage of rapid acquisition of data with generally short turnaround time, 100% observation of the scene, and a procedure that is considerably less costly than ground surveys (Tueller 1982, Everitt et al. 1992). Plant canopy reflectance measurements have been used to distinguish noxious brush and weed species from other rangeland plant species (Gausman et al. 1977a, 1977b, Everitt et al. 1986, 1992). Likewise, aerial photography has been used extensively to distinguish brush and weed species over large and inaccessible rangeland areas (Driscoll and Coleman 1974, Gausman et al. 1977b, Carneggie et al. 1983, Everitt et al. 1995, Driscoll et al. 1997, Anderson et al. 1999).

The objectives of this paper are to: (1) describe the plant canopy light reflectance characteristics of several weed and brush species that occur on rangelands in Texas, southwestern U.S., and Mexico; (2) demonstrate the use of conventional color and color-infrared aerial photography for distinguishing these species; and (3) relate the reflectance measurements to the image tonal responses of the plants on both types of aerial photographs.

General Procedures

Data presented in this paper have been previously published. Aerial photography was obtained under sunny conditions with photographic systems mounted vertically in either a Cessna 206 or an Aero Commander aircraft. Conventional color (0.40 - 0.70 µm) and color-infrared (0.50 - 0.90 µm) films were used in the studies presented herein. Conventional color film is sensitive in the visible blue (0.40 - 0.50 µm), visible green (0.50 - 0.60 µm), and visible red (0.60 - 0.70 µm) spectral regions, whereas color-infrared film is sensitive in the visible green (0.50 - 0.60 µm), visible red (0.60 - 0.75 µm), and near-infrared (0.76 - 0.90 µm) spectral regions. Geographic locations of images and other pertinent information presented here are given with the figure captions. Additional information on the photographic equipment and film can be obtained from the original papers cited.

In the research studies presented here, plant canopy reflectance measurements, ground photographs and reference data (i.e. plant species, density, cover, phytomass, soil surface conditions, etc.) were obtained at the study sites at or near the time photographs were acquired to help interpret them. The interpretation of photographic imagery in terms of how visible (0.45 - 0.75 µm) and near-infrared (0.76 - 0.90 µm) reflected light affected the film emulsion layers is presented and discussed. In conventional color film the yellow, magenta, and cyan emulsion layers are highly sensitive to the visible blue, green, and red spectral light, respectively. In color-infrared film, a yellow filter is used which shifts the yellow, magenta, and cyan emulsion layers sensitivity to the green, red, and near-infrared spectral light, respectively. This paper attempts to explain what generally occurs with the yellow, magenta and cyan film emulsion layers when reflected light from plants impinges the film. This also includes the resulting color produced from the combining of the exposed layers, being the color that eventually would result from the chemical processing of the film. Standard statistical techniques (Analysis of Variance and Duncan's multiple range test) were used to analyze and interpret data (Steel and Torrie 1980).

Silverleaf sunflower (Helianthus argophyllus Torr. and Gray) is an annual, non-palatable, prevalent weed 1-2 m tall found in deep, loose, sandy soils of southern Texas. It is unique in that the entire plant is covered with densely white-tomentose (matted and woolly) pubescence (hairs).

Fig. 1. Field spectroradiometric reflectance measurements over the 0.50 - 0.90 µm waveband for canopies of silverleaf sunflower and 4 associated species on south Texas rangelands.

Fig. 2. Color-infrared photographic positive print of a rangeland area infested with silverleaf sunflower. The arrow on the print points to the characteristic pinkish image of silverleaf sunflower. The photograph was taken near Nixon, Texas, on June 29, 1976, at an original scale of 1:11,000.

Gausman et al. (1977b) surmised that the pubescence would facilitate the identification of silverleaf sunflower on color-infrared aerial photography. Laboratory spectrophotometric leaf reflectance measurements showed that silverleaf sunflower had significantly higher (P = 0.01) visible (0.50 - 0.75 µm) and near-infrared (0.76 - 0.90 µm) reflectance than common sunflower (Helianthus annuus L.). Field spectroradiometric canopy light reflectance measurements supported those made in the laboratory. Everitt et al. (1984) further studied the canopy light reflectance characteristics of silverleaf sunflower and 4 associated rangeland plant species (Fig. 1). Silverleaf sunflower had significantly higher visible and near-infrared reflectance than the 4 associated species. Gausman et al. (1977b) and Everitt et al. (1984) attributed the high visible reflectance of silverleaf sunflower to its dense, white pubescent foliage, whereas its higher near-infrared reflectance was primarily due to its greater vegetative density than associated species (Myers et al. 1983).

Broom snakeweed [Gutierrezia sarothrae (Pursh.) Britt. and Rusby] is an abundant, noxious shrublet 15 to 90 cm tall found on southwestern rangelands in the United States and Mexico (Lane 1985). It often reduces forage production (Ueckert 1979), is poisonous, and causes great losses of livestock (McGinty and Welch 1987).

Everitt et al. (1987) described the light reflectance characteristics of broom snakeweed and associated species and showed that color-infrared aerial photography could be used to distinguish infestations of this weed. Figure 3 shows canopy reflectance data obtained in August 1984 for broom snakeweed and 8 associated plant species and mixtures over the 0.50 - 0.90 µm spectral region. The visible reflectance of broom snakeweed was similar to that of several other associated species, but its near-infrared reflectance was significally lower (P = 0.05). Broom snakeweed had a similar reflectance pattern on several other dates between June and September when it was in its mature vegetative stage. The low near-infrared reflectance of broom snakeweed was attributed to its erectophile (erect-leaf) canopy structure. The similarity in visible reflectance among broom snakeweed and several associated species was caused by their comparable green foliage colors and in-canopy shadowing (Richardson et al. 1975, Gausman 1985).

Fig. 3. Field spectroradiometric reflectance measurements over the 0.50 - 0.90 µm waveband for broom snakeweed and 8 other associated species or mixtures on south Texas rangelands in August 1984.

Fig. 4. Color-infrared photographic positive print of a rangeland area infested with broom snakeweed near Tatum, New Mexico, in August 1984. The arrow on the print points to the typical dark brown image of broom snakeweed. The print is a 1.5X enlargement of part of an original 70-mm photograph (original scale 1:10,000).

The erectophile canopy structure of broom snakeweed produced a deep, dark brown tonal color rendition on color-infrared aerial photographs (arrow on Figure 4) that can be easily distinguished from the various shades of magenta, red, and light brown of other plant species and the white image of soil. Ground truth surveys on 12 scattered sites selected from color-infrared photographic transparencies (1:5,000 to 1:10,000 scales) of rangeland areas in Texas and New Mexico resulted in 100% correct visual identification of broom snakeweed at all locations. The spectral reflectance of broom snakeweed (Fig. 3) supported its color-infrared photographic color tonal response in Figure 4. The low reflectivity of broom snakeweed in both the near-infrared and visible spectral regions, as a result of light entrapment due to its erectophile canopy structure and in-canopy shadowing, did not markedly affect the color-infrared film emulsion layers. This caused the cyan layer, as well as the yellow and magenta layers, to become highly saturated in the film, thus producing the dark image tonal response of broom snakeweed in the color-infrared photographs.

Huisache

Huisache [Acacia smallii (L.) Willd.] is a woody legume found on rangelands in southern Texas and northeastern Mexico. It is estimated to occupy over 1 million hectares of grasslands in southern Texas (Scifres 1980). Huisache is adapted to a variety of soil types but reaches its greatest density on medium to heavy textured soils in the Coastal Prairies and eastern portions of the South Texas Plains (Scifres et al. 1982). Although huisache has some value for wildlife (Drawe 1968), it rapidly becomes a serious brush problem on rangeland following soil disturbance by mechanical brush control methods such as rootplowing (Mutz et al. 1978).

Huisache has a prominent appearance during its flowering period in late February or March, producing a profusion of small orange-yellow flowers that encompass the entire plant giving it a striking appearance. It was surmised that this troublesome shrub might be distinguishable on aerial photographs when it flowers. This would be useful for monitoring its spread and delineating areas needing control.

Plant canopy reflectance measurements over the 0.45 - 0.90 µm spectral range for flowering huisache and 5 associated species and mixtures of species in March are presented in Figure 5 (Everitt and Villarreal 1987). At the 0.45 µm visible blue wavelength, huisache had similar reflectance to Mexican paloverde (Parkinsonia aculeata L.) and mixed herbaceous species. However, at the 0.55 µm visible green and 0.65 µm visible red wavelengths, huisache had significantly higher (P=0.05) reflectance than the other associated species and mixtures of species. The orange-yellow flowers of huisache produced higher visible green and red reflectance than the green foliage of the other species (Gausman 1985). The near-infrared (0.85 µm) reflectance of huisache did not differ from that of Mexican paloverde or honey mesquite.

Fig. 5. Field spectroradiometric reflectance measurements over the 0.45 - 0.90 µm waveband for flowering huisache and 5 other associated species or mixtures of species on south Texas rangelands in March 1985.

Fig. 6. Conventional color photographic positive print of a rangeland area infested with huisache near Sinton, Texas, in March 1986. The arrow on the print points to the characteristic golden image of huisache in flower. The print is a 4X enlargement of part of an original 70-mm photograph (original scale 1:6,000).

The reflectivity of flowering huisache affected the conventional color film (Fig. 6) emulsion layers differently in comparison to the other associated species. The low blue spectral reflectance (Fig. 5) of huisache had a minimal affect on the yellow emulsion layer causing it to saturate the film. Conversely, the high green and red reflectance of huisache highly affected the magenta and cyan emulsions, causing low concentrations of these layers in the film. Hence, the combinations of the saturated yellow emulsion with the moderately low concentrations of magenta and cyan emulsions rendered the bright orange-yellow coloration of huisache in the conventional color photograph. The low blue and red visible light reflectance of the associated species only slightly affected the yellow and cyan film emulsion layers, respectively, leaving relatively high concentrations of these emulsions in the film. In contrast, their moderately high visible green reflectance strongly affected the magenta emulsion, thus causing low concentrations of this emulsion in the film. As a result, the yellow and cyan film emulsions combined to produce the associated plant species variable green color renditions.

Each year poisonous plants affect 3 to 5% of the cattle, sheep, goats, and horses that graze the western rangelands of the United States (James et al. 1980). The most widespread poisoning of livestock in the western United States is by locoweed (James et al. 1981, Ralphs 1987). The genus Astragalus is represented by a number of species that cause locoweed disease (Kingsbury 1964, Ralphs 1987), which is caused by the alkaloid swainsomine (Molyneux and James 1982). Big Bend locoweed [Astragalus mollissimus var. earlei (Rydb.) Tidestr.], also referred to as Woolly locoweed or Earle loco, and Wooton locoweed (Astragalus wootonii Sheld.) or garbancillo are 2 species of Astragalus that often cause serious livestock poisoning in the Trans-Pecos area of west Texas (Sperry et al. 1964, Freeman et al. 1982). Big Bend locoweed is a perennial of a few years duration, whereas Wooton locoweed is an annual or biennial (Correll and Johnston 1970). Big Bend locoweed primarily affects cattle, horses, and sheep, while Wooton locoweed is most toxic to horses (Kingsbury 1964, Sperry et al. 1964, James et al. 1980).

Everitt et al. (1994) conducted a study to determine the feasibility of using remote sensing techniques to detect Big Bend locoweed and Wooton locoweed infestations on rangelands. Table 1 shows mean visible red (0.63 - 0.69 µm) and near-infrared (0.76 - 0.90 µm) reflectance for Big Bend locoweed, Wooton locoweed, 5 associated species and mixtures of species, and bare soil for 2 dates. In April 1991, Big Bend locoweed had a similar red reflectance value to that of smooth sotol (Dasylirion leiophyllum Engelm.), while the red reflectance value of Wooton locoweed did not differ from those of Javelina bush [Condalia ericoides (Gray) M. C. Johnst] and purple prickly pear (Opuntia violacea Engelm.). At the near-infrared wavelength both Big Bend and Wooton locoweed had higher reflectance than the associated plant species and soil. Reflectance measurements made in February 1992 followed a similar pattern to that shown in April 1991. The visible red reflectance of Big Bend locoweed was similar to that of purple prickly pear and smooth sotol, while the visible reflectance value of Wooton locoweed could not be separated from that of Javelina bush. The near-infrared reflectance values of Big Bend locoweed and Wooton locoweed were higher than those of the associated species and soil. The inability to separate Big Bend locoweed and Wooton locoweed from the associated species at the visible red wavelength was attributed to their similar foliage colors (Gausman 1985). The greater near-infrared reflectance of Big Bend locoweed and Wooton locoweed than the associated species was attributed to their denser canopies (Myers et al. 1983).

Table 1. Mean canopy reflectance of Big Bend locoweed, Wooton locoweed, 4 associated species and mixtures of species and soil for the visible red and near-infrared bands on two dates. Reflectance measurements were made in rangeland areas near Alpine and Marfa, Texas.

Fig. 7. Color-infrared photographic positive print (1:3,000 scale) of a rangeland area populated with Wooton locoweed near Marfa, Texas, in March 1991. The arrow on the print points to the bright red image of Wooton locoweed

Fig. 8. Conventional color photographic positive print (1:7,000 scale) of an infestation of Chinese tamarisk along the Rio Grande River near Lajitas, Texas, in December 1988. The arrow on the print points to the yellow-orange color of Chinese tamarisk.

The vivid red color tonal image response of Wooton locoweed in the color-infrared photograph was primarily attributed to its high near-infrared and relatively low visible red reflectance values. Although visible green reflectance measurements were not obtained for Wooton locoweed, the strong absorption of red visible light usually decreases a plant's green visible light (see silverleaf sunflower image discussion). As shown in Table 1, the visible red light reflectance of Wooton locoweed was low, thus, its visible green light reflectance would have probably also been low. The high near-infrared reflectance of Wooton locoweed strongly affected the cyan emulsion layer (causing low concentrations in the film), whereas its low visible light reflectance slightly affected and caused the saturation of the yellow and magenta emulsions. The mixture of the 3 emulsions produced the red signature of Wooton locoweed, but this tonal response is attributed more to the combining of the yellow and magenta emulsion layers.

Chinese tamarisk (Tamarix chinensis Lour.), also known as saltcedar, is a shrub introduced to the U. S. from Asia for use as an ornamental and for erosion prevention of streambanks (Baum 1967). Chinese tamarisk is an invader of riparian sites in the southwestern U. S. and northern Mexico, where it forms dense, low thickets that displace native vegetation, impede water flow, increase sedimentation, use excessive water, and increase soil salinity (Horton and Campbell 1974, Deloach 1990).

During late fall and early winter, the foliage of Chinese tamarisk turns a yellow-orange to orange-brown color prior to leaf drop and the plant is very conspicuous in contrast to other vegetation. It was surmised that Chinese tamarisk populations might be distinguishable on aerial photos during this phenological stage.

Plant canopy reflectance measurements made on Chinese tamarisk (dormant), athel tamarisk [Tamarix aphylla (L.) Karst.], and willow baccharis (Baccharis neglecta Britt.) at 4 wavelengths in January - February in south Texas are given in Table 2 (Everitt and Deloach 1990). At the 0.45 µm visible blue wavelength, Chinese tamarisk had similar reflectance to that of athel tamarisk. For the visible green (0.55 µm) and red (0.65 µm) wavelengths, however, Chinese tamarisk had significantly higher (P=0.05) reflectance than athel tamarisk and willow baccharis. The near-infrared (0.85 µm) reflectance of Chinese tamarisk did not differ from that of the 2 associated species. The higher visible green and red reflectance of Chinese tamarisk was attributed to its yellow-orange foliage which absorbed less green and red light than the evergreen foliage of the other two species (Gausman 1985). The similar blue reflectance of Chinese tamarisk to that of athel tamarisk was apparently due to the high absorption of the blue light by the caroteniod pigments in the foliage of both species (Gausman 1985). The inability to separate the near-infrared reflectance values of the 3 species was primarily due to their comparable vegetative densities (Myers et al. 1983). Reflectance measurements made on 3 other dates during the growing season showed that both the visible and near-infrared reflectance values of Chinese tamarisk could not be separated from those of 1 or more associated species on all dates (Everitt and Deloach 1990).

Table 2. Field spectroradiometric-measured canopy reflectance of Chinese tamarisk and 2 associated species in January-February 1988 at 4 wavelengths. Reflectance measurements were made near Rio Grande City, Texas.