Hyperspectral Mapping Of The Geology Of Afghanistan

This post submitted to the Accretionary Wedge # 48 hosted by Earth-like Planet. The theme is "Geoscience and Technology" and this post is on the use of multi and hyperspectral remote sensing for geological mapping.

Coinciding with the 40th anniversary of the Landsat series of remote sensing satellites, two maps of the surface distribution of several distinctive minerals covering a large portion of Afghanistan has been released by the USGS.

 Source: USGS Pub A

These maps have been prepared by processing the reflectance properties of surface materials captured by sensors aboard a plane. Conventional satellite mapping like that prepared from Landsat data does the same thing but it generally captures less information. For example, most conventional commercial satellites will capture reflected energy in the visible and the near infra red portion of the spectrum in 4 - 7 bands.

This type of remote sensing of the reflected and emitted energy from surface material is termed multispectral sensing. Recently, new satellites have started capturing hyperspectral data. Here, the energy from the visible to infrared spectrum is collected at very narrow intervals or channels. For example, NASA's Hyperion sensor aboard the EO-1 satellite is capable of collecting spectral information in 220 spectral bands from between the 0.4 to 2.5 µm (micrometer) bandwidth with a 30-meter ground resolution.



Satellite imagery is created by combining the information collected in different bands and then assigning the bands a particular color. This is called a False Color Composite.  These are the images  you are used to seeing at NASA and other websites or put up as a wall decoration in countless geology departments. These images can be used to visually identify the surface materials.  Experienced geologists can identify broad rock categories visually based on the tone (how dark or light) the outcropping and weathering pattern, bedding or regional foliation (mineral orientation) and other properties.

In this post, when I say identification of materials I am referring to their identification strictly from their spectral properties and not when the data is mapped out visually as in a satellite image.

The image below shows conceptually how multiple bands are collected of the same location and  processed to produce a reflectance curve for a particular location.

Source: MicroImages Inc

The identification is based on materials having a characteristic spectral curve formed by say more absorption in the green band and less absorption in the red or infrared band or vice versa. Using image processing software you can plot, as shown in the image above, the surface reflectance of objects located for example in the northeast corner of the image and study their reflection curve. From that you can, based on comparison to a library of spectral signatures, make an identification of the material in the northeast corner.

For the map of Afghanistan, the USGS used a sensor known as HyMap imaging spectrometer loaded aboard an aircraft. It collected spectral data covering 128 bands of 15-20 nm  (nanometer) bandwidth in the 0.4 to 2.5 µm (micrometer) range i.e the visible and the near infrared spectrum with a 5 meter ground resolution. Minerals have distinctive absorption signatures,  meaning that when sunlight strikes the surface of the earth the O-H or C-O3 or Si-O2 or Fe-OH bonds in the mineral absorb energy at a distinct wavelength, each covering a very narrow portion of the spectrum . Because conventional multispectral sensing collects energy averaged over a  broad interval it cannot discriminate between individual minerals. Hyperspectral sensing is fine grained enough (15-20 nm bandwidth) to be able to resolve the distinctive absoption signatures of several minerals.

The image below shows the types of surface objects identifiable from the broad band multispectral sensors aboard conventional satellites like Landsat and SPOT.

Source: MicroImages Inc

And the second image below shows the typical minerals that can be identified using hyperspectral mapping.

Source: MicroImages Inc

Using this advantage of a hyperspectral scanner the USGS has compiled two maps - a) of carbonates, phyllosilicates (clay minerals), sulphates and other altered minerals and b) of iron bearing minerals.

How can such maps help in searching for precious metals? After all the maps don't show where the gold or copper or other metals are located. Rather you have a map of mostly minerals formed by the weathering of rocks.Take the example of clay minerals. A distribution of various clay minerals in a terrain can point to two things, one is the nature of the source rock and the other is the intensity of chemical weathering.

Clay minerals like montmorillonite which is a Mg rich clay form by the weathering of mafic and ultramafic rocks. Clays like illite which is a K (potassium) rich clay form by the weathering of granitic rocks. Mafic and granitic rocks may host under certain situations concentrations of specific precious metals. Another clay called kaolinite forms under very intense chemical weathering wherein most of the more soluble elements get leached away from the soil. Kaolinite rich terrains may contain what are known as residual elements which resist being leached away. These are for example the rare earth series prized for their many applications in high-tech electronics and magnets. Significant portions of the massive rare earth deposits in China are in these deeply weathered kaolinite rich clay deposits.

Other alteration products like sulphate or dolomite along with a knowledge of their associated host rock can point to the ancient  presence of  particular types of hydrothermal fluid systems which along with altering the rock may have deposited metals.

So, having such a map of the weathering of Afghanistan considerably eases the job of identifying the more promising areas for metal deposits. Such remotely sensed information will not eliminate the need for actual field work along with rock sampling and physical estimation of metal abundance, but it does save time by rejecting certain areas and pointing to the more likely ones.