BMR Research Newsleller 8
April 1988
Landsat Thematic Mapper: a valuable geological tool, but . In July 1982, ten years after the launch of the first Landsat satellite, NASA launched Land• sat-4 which heralded a new phase of earth resources sensing. Besides the conventional 4band multispectral scanner (MSS), Landsat-4 also incorporated the more advanced Thematic Mapper (TM) scanner. All seven sensing bands of the TM have more spectral, radiometric, and geometric sensitivity than its 4-band predecessors, and thus provide increased sens• ing capability. The aim of this article is to sound a note of caution with respect to the digital processing of TM data, particularly those techniques involving band-to-band man• ipulation of raw data.
TM was the first satellite system with a band incorporated for geological purposes. Band 7 (2.08-2.35 micrometres), is particularly sensitive to absorption of electromagnetic radiation by clay, carbonate, and some mica minerals. Based on studies conducted in the USA, the TM Band 7 wavelengths offered particular promise for locat• ing areas of hydrothermal alteration. Since 1986 Landsat TM data have been recorded in Australia via the ALSICSIROIAMIRA Signal Processing Experiment. Band to band manipulation of satellite scanner data such as 'ratioing' (i.e. dividing one band digital number by another band digital number) has become a common form of raw data enhan-
Table 1. Comparative spectral bands of the Landsat MSS and TM scanners (micrometres) Landsat 1,2,3: MSS
Landsat 4,5: TM
Band No.
Band No.
4 5 6 7
Wavelength 0.~.6
O./Hl.7 O.7--{).8 0.8--1.1
1 2 3 4 5 7 6
Wavelength O.4~.52
O.52--{).60 O. 63--{). 69 O.7/Hl.90 1.55-1.75 2.08--2.35 10.4--12.5
cement and for processing to extract specific thematic information from Landsat MSS data. A more significant application is its potential as a processing method to identify specific geological materials based on their spectral properties. Several 'standard' ratio combinations have evolved as basic detection methods for particular minerals. However, the successful application of these approaches necessitates some form of pre• processing of the raw satellite data. Raw radiance data recorded by the satellite scanners comprise reflected radiation and additive and subtractive components representative of the atmosphere through which they pass. Thus the raw scanner signal, which is ultimately contaminated with various other electronic signals, does not represent a true spectral reflectance value. Com• bined effects of radiance, gain and offset settings, and different relative band sensitivities of the raw data can seriously affect digital manipulations such as ratioing . Problems associated with attempted spectral interpretation of raw radiance data as recorded by Landsat scanners are not new and have been reported by Australian and overseas authors. De• spite this, the raw radiance data are frequently regarded, in digital manipulation, as a form of spectral information . With Landsat MSS data this approach has resulted in some acceptable rock discrimination, but the more sensitive TM data need to be treated with caution . The problem is illustrated in Figure 14 which shows 6-point extrapolated curves of the reflec• tivity of a volcanic rock type as measured on the ground (dashed line) , and equivalent band radiance values from the raw TM data (solid line) as recorded by the satellite over the same rock type. It is apparent that ratio manipulations on such raw data will produce erroneous results, the mag• nitude of the error being band-dependent. Thus before spectral signature processing can be reli• ably performed, the raw data (solid curve) must be corrected to the equivalent values of the dashed curve. Successful TM ratio results have been obtained by BMR (such as accurate discrimination
• •
of mafic and ultramafic rock types) after conver• sion of the raw radiance data to equivalent ground reflectance based on field spectral reflectance measurements at selected ground sites. Other pre• processing techniques can be applied to improve the raw data prior to band-to-band manipulation . The differences between raw radiance values and spectral reflectance values appear more pron• ounced with TM data than has been previously experienced with MSS . One aspect of current BMR remote sensing research is the investigation of possible techniques for direct calibration (con• version of raw radiance data to reflectance) of TM data recorded over the Australian arid zone . It is not the intention of this note to deter users, or potentional users , from using TM data. Enough examples of different geological applications have been investigated by BMR to show that TM data can have significant cost-effective applications in a variety of field programs. From what we have seen it is the best geological remotely sensed spectral data yet recorded by any satellite system , but selectivity is required with the processing techniques employed, especially if band-to-band manipUlations are being contemplated.
For further details contact Mr Colin Simpson, or Mr Taro Macias at BMR (Division of Petrology & Geochemistry) . LANDSAT TM Bands 1 2 3
4
7
30·15/8
0.4
Wave length (micrometras)
2.5
Fig. 14. Reflectivity of volcanics as measured on the ground (dashed line) and equivalent TM raw radiance values (solid line).
Brave new world in cartography This month BMR is scheduled to let a contract for the first stage of a new, state-of-the-art, high-resolution, automated map production system. The conventional environment of ink pens, light tables, and opaquing brushes will be replaced by map publishing software, a laser photoplotter/scanner system and interactive graphics terminals, enabling BMR to retain its place as a world leader in geoscientific car• tography.
BMR's vast collection of map separates at various scales and projections is becoming brittle with age , and keeping it up to date by hand• revision of boundaries, name changes, etc. on each map separate is become prohibitively time• consuming. The new system will combine laser scanning hardware and VAX*-based scanning soft• ware, to provide complete data capture, vector• isation, editing , and plotting capabilities. The development of a digital cartographic database will enable BMR's Cartography Section to meet specific customer requests by accessing any part of it and manipUlating content, symbology, text fonts, scale, projection, or design without actually changing the master design file .
Capturing the data
The high-resolution scanner will convert hard• copy map data to digital format rapidly and • VAX is a trademark of Digital Equipment Corporation.
accurately. It can handle media up to I metre x 1 metre and scan paper or stable-base materials at 12.5, 25, 50, 100, or 200 microns (2000, 1000, 500, 250, or 125 pixels) per inch, to a positional accuracy of less than 2.5 microns per inch. Once scanned, the data are converted to vectors or polygons for use as a digitising backdrop. Text and symbols are recognised, and specialised edit• ing commands smooth the process of conversion. Using interactive graphics terminals and mapping software, cartographers will update existing map data, adding information as required .
New tools
Traditional map printing may require up to 40 overlays, each containing a component of the final map; using positive artwork, the photographer makes masks and open-window negatives to hold out ink where only a single tint should print. In the final stages, separates are sandwiched with dot screens to create a set of four composite positives. The new customised software replaces this entire darkroom process. With perfect registration as• sured, the cartographer can concentrate on the creative aspects of map publishing by simply building a specification table to define the map, e.g. creating digital masks, text haloes, and ban• ded area tints, reversing symbols or text (white text on a coloured area), customising colours, etc. Using the laser photoplotter capability of the
scanner, the cartographer displays the new map at the workstation before plotting any film positives. This eliminates costly trips to the darkroom for colour proofs to verify masking, registration, and content quality . Instead, softcopy proofing re• plicates the map on screen so that changes can be made before final plotting. For final editorial approval, the software includes the option to generate a hardcopy coloured map on an electro• static plotter.
Print-ready output
After final clearance for publication, the scan• ner plots film positives for the standard four• colour printing process (cyan, magenta, yellow, and black) . It can also produce negatives for other multicolor lithographic printing processes. Screens are automatically composited, to plot each positive in a single pass, producing a complete set of film positives ready for lithographic platemak• ing. BMR 's new system will be an end-to-end digital map production system consisting of in• tegrated digitising, attribution , editing, design, symbolisation, and colour separation tools. It is a very exciting system, and new applications are being discovered every day.
For more information contact Mr John Hillier at BMR (Special Projects & Geoscience Services Branch) .
9