The synoptic character of Landsat and other large-area coverage satellite remote sensors is proving especially favorable in geologic mapping and other geologic applications. Land cover/use, agriculture, urban monitoring, and similar non-geologic activities related to human endeavors for the most part relate to areal coverage at smaller scales. Geologic phenomena typically are spread over wider scenes, so that the ability to see the "regional picture" is a powerful attribute of space imagery. This perception is supported by the discussion of landforms analysis that makes up most of Section 17.
This view is somewhat tempered by the "interference" of vegetative cover in trying to single out geologic features. Aerial photography and space imagery work especially well on geologic subjects when the scenes they portray are minimally covered with vegetation. Early spring pictures are usually optimal (before leafing). Also, best results obtain when rocks are exposed (outcrop), are covered with little soil or are bare, and are not buried by alluvial deposits. The best images seen in this Section are in the western United States and arid regions elsewhere in the world.
Geologists have used aerial photographs for decades to serve as databases from which they can do the following:
1. Pick out rock units (stratigraphy)
2. Study the expression and modes of the origin of landforms (geomorphology)
3. Determine the structural arrangements of disturbed strata (folds and faults)
4. Evaluate dynamic changes from natural events (e.g., floods; volcanic eruptions)
5. Seek surface clues (such as alteration and other signs of mineralization) to subsurface deposits of ore minerals, oil and gas, and groundwater.
6. Function as a visual base on which a geologic map is drawn either directly or on a transparent overlay.
With the advent of space imagery, geoscientists now can extend that use in three important ways:
1) The advantage of large area or synoptic coverage allows them to examine in single scenes (or in mosaics) the geological portrayal of Earth on a regional basis
2) The ability to analyze multispectral bands (especially hyperspectral data sets) quantitatively in terms of numbers (DNs) permits them to apply special computer processing routines to discern and enhance certain compositional properties of Earth materials
3) The capability of merging different types of remote sensing products (e.g., reflectance images with radar or with thermal imagery) or combining these with topographic elevation data and with other kinds of information bases (e.g., thematic maps; geophysical measurements and chemical sampling surveys) enables new solutions to determining interrelations among various natural properties of earth phenomena.
While these new space-driven approaches have not yet revolutionized the ways in which geoscientists conduct their field studies, they have proven to be indispensable techniques for improving the geologic mapping process and carrying out practical exploration for mineral and energy resources on a grand scale.
The very first practical use of ERTS-1 (Landsat-1) imagery in any discipline was the drawing by Dr. Paul D. Lowman, Jr, of a geologic structures map superimposed on the first color composite image, based in part on already known field information and in part on his interpretation of this scene. He is a geologist at Goddard Space Flight Center, and an expert on space photography (he prepared Section 12 on Astronaut Imagery in this Tutorial). The image was of the central California coast around Monterey Bay, acquired 3 days after launch.
This map confirmed predictions from his studies of astronauts' photos that Landsat would be an efficient tool for recognizing faults and other known structural trends in small-scale imagery. In spite of lower resolutions, these images excel in portraying regional geologic settings and are easily enhanced by digital processing.
We now consider several examples of geologic applications using these new approaches. We concentrate initially on how Landsat Thematic Mapper (TM) data for a local region in Utah are manipulated to identify different rock types, map them over a large area using supervised classification, and correlate their spatial patterns with independent information on their structural arrangement. Next, our focus changes to examination of geologic structures, particularly lineaments, as displayed in regional settings in the U.S., Canada, and Africa. Then, in Section 5, we will look at how space-acquired data fit into current methods of exploring for mineral and hydrocarbon deposits by considering a case study of a mineralized zone in Utah and at a large-area Landsat scene in Oklahoma. In Section 17, we will return to a geologic theme by examining landforms at regional scales, (so-called Mega-geomorphology), as a prime example in considering how remote sensing is used in basic science studies.
Most geologic maps are also stratigraphic maps, that is, they record the location and identities of sequences of rock types according to their relative ages. The fundamental rock unit is the formation (abbreviated as Fm or fm), defined simply as a distinct mappable set of related rocks (usually sedimentary) that has a specific geographic distribution. A formation typically is characterized by one or two dominant types of rock materials.
The term "formation" is most commonly associated with strata, namely layers of sediments that have hardened into sedimentary rocks. Under most conditions, sediments are laid down in horizontal or nearly so layers on sea floors, lake bottoms, and transiently in river beds. Here is a typical set of sedimentary layers exposed in a road cut (note that the layers have been cut and slightly offset by a break which is termed a "fault"):
If we see sedimentary rocks inclined at more than a few degrees from the horizontal, we should suspect that these are involved in displacements from their original horizontal state by forces (tectonic) that cause the rocks to bend and curve (folds) or break (faults). Here is a roadcut along a Maryland highway that is passing through the fold belt of the Appalachians.
As we shall see later in this Section, inclined layers can produce curved structures called anticlines (uparched) and synclines (downarched), Here is an example of the latter:
Any given formation is developed and emplaced over some finite span of geologic time. We can approximate its age by the fossils (evidence of past life) that were incorporated within the soft layers (which become strata or beds) during the time in which these life forms existed. Age dating by determining the amounts of radioactive elements and their decay-daughter products can usually produce even more accurate age estimates. Another, less precise, approach to fixing the age (span) of a rock unit is to note its position in the sequence of other rock units, some of whose ages are independently known. We can correlate the units with equivalent ones mapped elsewhere that have had their ages worked out. This method tends to bracket the time in which the sedimentary formation was deposited but erosional influences may lead to uncertainties. The association of sedimentary layers with specific time intervals constitutes the field of stratigraphy. Igneous and metamorphic rocks also have time significance and are treated as rock units on geologic maps (which show all stratigraphic and crystalline units in a legend).
Remote-sensing displays, whether they are aerial photos, space-acquired images, or classification maps, show the surface distribution of the multiple formations usually present and, under appropriate conditions, the type(s) of rocks in the formations. The formations show patterns that depend on their proximity to the surface, their extent over the surveyed area, their relative thicknesses, their structural attitude (horizontal or inclined layers), and their degree of erosion. Experienced geologists can recognize some rock types just by their appearance in the photo/image. They identify others types from their spectral signatures. Over the spectral range covered by the Landsat TM bands, the types and ages of rocks show distinct variations at specific wavelengths. This is evident in the following spectral plots showing laboratory-determined curves obtained by a reflectance spectrometer for a group of diverse sedimentary rocks (collected from their field formations [named from a geographic locality near where the fromationas unit was first described) from Wyoming:
2-1 From these spectra, predict the general color of these four rock units: Niobrara Fm; Chugwater Fm; Frontier Fm; Thermopolis Fm. ANSWER
2-2: What spectrally distinguishes the Mowry Fm from the Thermopolis Fm (both dark in the field); the Jelm Fm from the White River Conglomerate? ANSWER
Several of these spectral signatures are composite, in that more than one mineral substance is present. Spectra of individual minerals can be quite different, as shown in the next illustration:
The Near-IR wavelength interval from 2.0 to 2.4 Ám is especially sensitive to absorption of radiation; note that all four minerals share a negative peak (trough) around 2.32 Ám. But a word of caution: there are several thousand mineral species known from Earth. The spectra of many of these can be similar, making field and even laboratory distinction of just what mineral species is being looked at difficult to assess (X-ray diffraction patterns are usually better identifiers). When remote sensing is involved in onsite surveys, knowledge beforehand of the particular minerals likely in the sensed scene is usually a benefit. Thus, when an area is being surveyed for specific mineral content (as in exploration for ore minerals; Section 5), this information aids in identifying the mineral species being mapped. This is also true for rock types; whenever possible, field spectra such as shown above for Wyoming formations are gathered as Ground Truth (Section 13) so that their diagnostic characteristics can be used to calibrate the image analysis (as an example, referring to the rock spectra above, the averaged reflectances of the 6 non-thermal TM bands can be calculated for, say, the Muddy Sandstone; this knowledge could then be fed into a computer classification as a discriminator function which looks for all pixels with similar multiband values).
A common way of mapping formation distribution is to rely on training sites at locations within the photo/image. Geologists identify the rocks by consulting area maps or by visiting specific sites in the field. They then extrapolate the rocks' appearance photographically or by their spectral properties across the photo or image to locate the units in the areas beyond the training sites (in effect, the supervised classification approach).
In doing geologic mapping from imagery, we know that formations are not necessarily exposed everywhere. Instead they may be covered with soil or vegetation. In drawing a map, a geologist learns to extrapolate surface exposures to underneath covered areas, making logical deductions as to which hidden units are likely to occur below the surface. In working with imagery alone, these deductions may prove difficult and are a source of potential error. Also, rock ages are not directly determined from spectral data - only material types are determined, so that identifying a particular formation requires some independent information (knowledge of a region's rock types and their sequence).
In exceptional instances, such as those to be shown on the next three pages, when geologic strata are turned on their side (from folding; discussed on page 2-5) so that the successive geologic units are visible as a sequence, the changes within and between each discrete unit can be measured in terms of some spectral property, as for example, variations in the reflectance of a given band, or a ratio of bands. When plotted as shown below the result are tracings that resemble (analogously) those made from well logging of such properties as electric resistivity, permeability, magnetic intensity and other geophysical parameters. Here are two figures, the top showing the succession of sedimentary strata exposed along the Casper Arch in central Wyoming; the bottom being reflectance "logs" derived from spectral traverses along one of the lines in the upper image:
In the lower diagram, the bottom unit is the Permian Phosphoria Formation, extending upward from the Triassic Chugwater Formation to the Frontier sandstone (Cretaceous) at the top. On the right the left tracing is of TM Band 3 (red), with 0% reflectance on the right base extending to 70% on the left, and the right tracing, for Band 1, goes from 0% on the left to 50% on the right.
Before looking at some specific examples of the use of space imagery for geologic structure analysis, this is a good point to introduce one particular advantage of having space observing systems that can repetitively cover the same large regions over the four seasons. Two Landsat images appear below: one taken during the southern Winter in South Africa; the other during the height of Spring. The area includes Johannesburg, some of the gold mines in the Witwatersrand district, and the Pilanesburg pluton (near the top). In the wintertime, some of the underlying rock units fail to show distinctly because the entire scene has its vegetation (mostly grasslands) dormant. But with plant reawakening in Spring, different units have different vegetation types and these variably modify the colors displayed, thus revealing the more complex structures in the region.
This geobotanical phenomenon (the differential distribution of various plant types as a function of soils developed on different rock types) is sometimes used as a mapping or prospecting device (example; the element Selenium is associated with Uranium; certain plants thrive on Selenium enrichment and are thus indicators of subsurface Uranium enrichment).