Visibility Impairment

Aerosols introduced into the atmosphere can result in visibility impairment that is manifested in two distinct ways: first, as a general alteration in the appearance of landscape features such as color, contiguous contrast between adjacent geologic features, etc., and secondly the aerosol haze may become visible in and of itself. Haze may be visible by the contrast or color difference between itself and its background, or (at great enough optical depths) uniform haze manifests itself as a sem¡transparent curtain that can be seen or perceived as a separate hazy entity disassociated from landscape features. Henry (1987) has referred to the phenomenon as atmospheric transparency, which is psychophysical in nature, and different from atmospheric transmittance.

Perceptibility Parameters for Quantification of Layered Haze (Plume Blight)

Figure 7 illustrates two situations in which a layered haze is visible: (a) when viewed against the sky and (b) when viewed against terrain features. In both cases, the layered haze will be visible as a distinct, horizontal layer if it is sufficiently brighter or darker than the viewing background.

(a) Plume visible against the sky oo

(a) Plume visible against the sky

Figure 7 Two viewing situations in which plumes may be visible.

The simplest way to characterize the relative brightness (or darkness) of plumes is through the use of plume contrast:

PK -hNr

where Nr and hNr are the spectral radiances of the plume and its background, at some distance, r, and at wavelengths in the visible spectrum (0.4 < I < 0.7 |im). A plume is visually perceptible only if it creates a nonzero contrast at different wavelengths in the visible spectrum greater than an observer's perceptibility threshold (generally in the range of ±0.01 to ±0.05).

An object can be perceived because it has a brightness different from that of the background or because it has a different color. Gases and particles in the atmosphere can give rise to coloration by their light-scattering properties (blue sky or white clouds) or by altering the color of objects seen through them (brown coloration due to N02). Several schemes have been used to quantify color. The Commission Internationale de l'Eclairage (CIE) has set colorimeter standards that form the basis of the CIE system of color specification. The most popular CIE index is the so-called AE parameter that not only quantifies differences in color but also differences in brightness. However, the CIE method, while accurate and acceptable for a laboratory situation, may not adequately represent color differences in a natural setting. In any case, a AE of 1 is a just noticeable difference in color and/or brightness in a laboratory setting and AE of 4 can be easily seen by the casual observer.

Layered Haze Thresholds. Psychophysical research (Cornsweet, 1970; Faug-eras, 1979; Hall and Hall, 1977; Henry, 1986; Howell and Hess, 1978; Malm et al., 1987; Ross et al., 1987) has documented the fact that the human eye-brain system is most sensitive to spatial frequencies of approximately three cycles/degree (cpd). Spatial frequency is defined as the reciprocal of the distance between sine-wave crests (or troughs) measured in degrees of angular subtense of a sine-wave grating. Thus, spatial frequency has units of cycles/degree. Any pattern of light intensities, whether it is a sine wave, square wave, step function, or any other pattern, can be resolved by Fourier analysis into a sum of sine-wave curves of different magnitude and frequency. For instance, a rough estimate of the primary spatial frequency of a Gaussian plume can be made as follows. If it were assumed that a Gaussian distribution is nearly identical to a sine-wave pattern, a 2° width of the plume would correspond to the period of the sine wave. The spatial frequency would be the inverse of this, or 0.5 cpd. Figure 8 illustrates several estimates of the sensitivity of the human visual system to sine/square-wave gratings and single Gaussian and square-wave stimuli with various spatial frequencies.

The sensitivity of the human eye-brain system drops off significantly at high spatial frequency (due to visual acuity) and also to a lesser extent at low spatial frequency (i.e., broad, diffuse objects). The human visual system is more sensitive to images with sharp, distinct edges (e.g., square waves) than to images with diffuse, indistinct edges (e.g., sine waves or Gaussian plumes).

Figure 8 Sensitivity curves as reported by Howell and Hess (1978) for sine- and square-wave ratings and for sharp-edged (Malm et al., 1987) and Gaussian plumes (Ross et al., 1990).

Plume Vertical Angular Subtense (degrees)

Figure 8 Sensitivity curves as reported by Howell and Hess (1978) for sine- and square-wave ratings and for sharp-edged (Malm et al., 1987) and Gaussian plumes (Ross et al., 1990).

Ross et al. (1997), based on an extensive literature review, designed a laboratory study to develop the information necessary to predict the probability of detection of plumes with a known size, shape, and contrast. The strategy taken was to develop probability of detection curves for computer-generated plume stimuli that encompasses the various plume geometries that could be encountered in the "real" world and interpolate between these measured thresholds to develop estimates for plumes with other shapes and geometries. In each case, the protocol for observer detection was the same for all experiments, the surround was kept at the same brightness, edge effects were dealt with uniformly, and stimuli representative of Gaussian plume brightness profiles were used.

Sixteen subjects were used for a full-length plume experiment. The stimuli used consisted of plumes with vertical angular sizes of 0.09°, 0.18°, 0.36°, 0.72°, 1.44°, and 2.88° and a horizontal angular extent of 16°. Contrast values of 0.050, 0.040, 0.030, 0.020, 0.017, 0.015, 0.013, 0.011, and 0.005 were used for all sizes. Figure 9 shows the predicted probability of detection curves. As plume contrast increases, the probability of detecting the plume increases. If a plume has a modulation contrast of greater than about 0.01 it will be detected nearly 100% of the time for all size plumes. Furthermore, these curves show that the size of the plume is quite impor-

Figure 9 Predicted probability of detection curves for one subject used in the full-length plume study.

tant! Plumes that subtend an angle of about 3° can be detected more easily than plumes that are larger or smaller. Results for circular and oval-type plumes with Gaussian edges were similar but required higher contrasts to be detected.

To more clearly see how the three shapes compared, the modulation contrast corresponding to 50% probability of detection for each shape is plotted against plume size in Figure 10. Notice that the general trend for all stimuli is the same, with plumes subtending about a 3° width being the easiest to detect. However, observers are most sensitive to full-length plumes and least sensitive to circular stimuli with the oval plumes being intermediate. The full length, oval, and circular plume contrast threshold data have been incorporated into a linear interpolation algorithm that allows plumes of any size to be estimated.

Other studies have been carried out for brighter than background-layered hazes. They identified a 70% detection threshold contrast of 0.02 using photographs of a natural scene with light-colored layered hazes, which varied in size. The evidence for AE thresholds is not as clear-cut. The data of Jaeckel (1973) and Malm et al. (1980) support 70% detection thresholds for AE of three, while the estimates of Latimer et al. (1978) and the more recent data of Malm et al. (1987) and Henry and Matamala (1990) suggest a AE threshold of less than 1. This work is summarized in Table 2.

□ circular x oval o full length

0.001

0.01

Width (degrees)

10.0

Figure 10 Threshold modulation contrast is plotted as a function of plume width in degrees for full-length, oval, and circular plumes. The human observer is most sensitive to all plumes if they have a width that is about 3°. Plumes larger or smaller than about 3° require increased contrast to be seen.

0 0

Post a comment