V. SOME RESEARCH RESULTS AND APPLICATIONS


The Need for Luminescence Spectra

The first example illustrates the need for both reflectance and luminescence spectra. Figure 17 (top) shows the reflectance characteristics of two fabrics, A and C, and of a typical green leaf, curve B. Note the deep water absorption bands at approximately 1.4 and 1.9 µm, whieh are typical of green turgid vegetation. Fabric C is a reasonable match in the visible and out to about 1100 nm. But, it distorts the 1400 nm water absorption band, misses the 1900 nm band, and would be easy to detect. Fabric A, however, mimics the vegetation throughout the spectral range. Based on the reflectance spectra, it would not be detectable against a vegetal background, and would, in fact, be classified as vegetation. At this point, one needs other spectral information, such as luminescence, or infrared thermal emittance. In Fig. 17B, the luminescence intensities of the fabrics, vegetation, and soils were plotted for the indicated Fraunhofer lines. Fabrics A and C not only have strong signals as compared to soils and vegetation, but have different distributions. Thus, they would be detectable in a milieu of soil and vegetation, and also distinguishable from each other. When bruised, the vegetation showed a strong luminescence that persisted for hours, indicating a possibility for detecting passage of traffic.

Figure 18A shows the reflectances of four differently dyed areas of a fabric. Although they, or their composite signal, would be distinguishable against a background of vegetation, the contrast would not be strong against a mixture of soil and different vegetation types and conditions. On a luminescence basis, however, the fabric has a signal intensity that not only greatly exeeeds that of soils and vegetal backgrounds, but occurs at different wavelengths--making detection a certainty if the areal extent is sufficient. These relations are shown in graphs B and C. For airborne detection, the signal threshold is about 1,500 units, and the fabric's luminescence peak is 81,000 units--the "hottest" we had measured to date. In general, the luminescence seldom goes above 12,000 for healthy vegetation. As vegetation senesces, the luminescence increases, but rarely reaches 20,000. The vegetal sample in Fig. 18C, has a peak of 11,000. The iso-intensity contour plot, graph C, shows that the luminescence of the two samples, vegetal and fabric, occurs in different wavelength bands. The original fabric, designed by the U.S. Army Natick Laboratories, matched the luminescence background. What happened? What happened was, the fabric was laundered, and the soap brightener, which luminesces, coupled to the material.

Vegetal Luminescence

Figure 19 shows some of the differences that take place in vegetal luminescence as a function of drying and of aging. In each example, the upper graph is a three-dimensional display of emission versus excitation, and the lower graph is an iso-contour plot. Figure 19A is typical of healthy, turgid, green leaves, i.e., five distinct luminescence peaks--a group of four with one peak above them. Drying (Plot B) results in a loss of three peaks, reduced intensity in two, and the development of a peak in another area of the plot. Senescence (Plot C) results in the loss of the original five peaks, and the development of two new peaks. Table 3 summarizes these results.

Luminescence and Pollen

An interesting observation was made by W.R. Hemphill, USGS (personal communication), that when looking out over the terrain with the FLD instrument, there was a strong luminescent signal just above the horizon. It was during the pollen season, and he wondered if the two were connected. Subsequent measurements at TEC confirmed the possibility, as shown in Fig. 20. Implicit in this illustration are a potential application and a potential problem. First, a possible technique for detecting and monitoring airborne pollen loads, i.e., measuring atmospheric quality; and second, a resulting problem--i.e., such an airborne load can reduce contrast, or otherwise interfere with the recording of terrain surface signals.

Luminescence Characteristics

All materials have spectral reflectance characteristics; but, not all materials have luminescence characteristics. Although we have examined but a portion of what is available, some general statements can be made. Soils measured to date do not show useful luminescence. About 75 percent of the vegetal samples and 30 percent of the fabrics have detectable and diagnostic luminescence peaks. For healthy turgid vegetation, the luminescent peaks fall in the wavelength range between 640 and 800 nm. As vegetation dries out, these peaks decrease in intensity and peaks develop in the wavelength region between 400 and 600 nm. Intensity distributions are related to material type and condition, and the peak intensities can be sorted into fairly distinct groups based on emission wavelengths. As shown in Fig. 21, healthy herbaceous vegetation falls into one assemblage, and everything else, e.g., paints, fabrics, pollen, dry vegetation, senesced vegetation, etc., falls in another. So far, with reference to peak emissions, there are three diagnostically useful excitance bands. These are centered at 400, 460, and 660 nm. The important emission bands are centered at approximately 690 and 730 nm.

Reflected Photons Versus Luminesced Photons

Keep in mind that sensors cannot distinguish between reflected photons and luminesced photons. If sun induced luminesced photons are present, they will be recorded along with the reflected photons, causing a slightly higher DN value or slightly darker gray tone. It is not a large contribution; but, it is a contribution. A spectrometer trace made from a surface illuminated with full-spectrum light, such as the sun, or a halogen lamp, can differ from a record made of a surface illuminated with narrow-band energy at each measurement step. The former situation is characteristic of airborne systems such as AVIRIS, and of field spectrometers, whereas the latter is characteristic of many laboratory instruments, or airborne systems using lasers for illumination, or excitation. For example, luminesced photons from camouflage fabric (refer to Fig. 18) can add to the reflected photon stream to give a slightly higher intensity over the 400-600 nm range. This is shown in Fig. 22, which compares traces recorded from a surface in full illumination and in monochromatic illumination. There can be as much as 8-10 percent absolute difference and 35 percent relative difference. The slight increase in the near infrared region might also be caused by luminescence. It has long been known that vegetation luminesces in the near infrared as the result of the strong absorption at about 645 nm and at about 430 nm (refer to Fig. 19). These added luminesced photons could cause a slight increase in slope steepness of the rapidly increasing reflectance edge just beyond 700 nm, the so-called red edge.


J.N. Rinker, USA TEC; Knowledge Sciences, Inc.