Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

Venessa Bennett

ASTER MINERAL INDEX PROCESSING USING ERDAS IMAGINE, ER MAPPER, PCI GEOMATICA and EXELIS ENVI w0279429@nscc.ca

INTRODUCTION ER MAPPER ERDAS IMAGINE PCI GEOMATICA EXELIS ENVI


Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) dataset have been routinely used for alteration mineral mapping since the launch of the NASA GENERAL INFORMATION GENERAL INFORMATION GENERAL INFORMATION GENERAL INFORMATION
Terra Satellite in December 1999. ASTER datasets are attractive due to the cost, wide coverage and unique series of sensor bands that are sensitive to alteration assemblages
commonly associated with mineral deposits. A host of image processing techniques have been developed to improve identification of various minerals, including band ratios, ER MAPPER represents part of the Hexagon Geospatial suite of software packages of which includes ERDAS ERDAS IMAGINE is the second in the suite of software packages from Hexagon Geospatial Inc. evaluated PCI Geomatica is a third high-end image processing software suite developed in Canada. Similar to ERDAS A trial version of Exelis ENVI was acquired to evaluate software functionality. ENVI was originally designed
spectral indices, Principal Component Analysis (PCA), spectral matching to reference and custom libraries. Imagine. A single user license costs approximately $CAN5000 (commercial), and was designed initially to focus for carrying out ASTER Band ratio calculations. A commercial single use license is in excess of $CAN10 000, Imagine, a single commercial license is in excess of $CAN 10000. Various add-in modules to the program to process and analyse hyperspectral datasets but has evolve to become a high end image processing and
on needs of image processing for use in mineral exploration. ER Mapper is the least expensive of the four however the software offers greatly expanded functionality compared to ER Mapper. It allows processing increase functionality so that it competes with ERDAS Imagine. PCI Geomatica has inbuilt into the core analysis software suite. A single-user commercial license is also in well in excess of $CAN 10 000. The
The purpose of this study is to: software suites examined and while capable of many of the core image processing tasks (mosaicking, image of most raw image formats and provides various higher end analysis tools (e.g. hyperspectral analysis, software the ability to perform atmospheric corrections using the ATCOR series of algorithms, a capacity software offers several add-ons, including modules for LiDAR and RADAR data processing. In addition to
enhancements, DEM creation, classifications), it lacks some of the high end functionality available in more photogrammetry, radar interferometry and stereo analyst). Image processing specialists from many that is not available in the standard ERDAS Imagine suite (extra add an). The User-interface is a lot simpler private image processing and analysis companies, the target market for ENVI products also includes the
# Conduct band ratio calculation techniques on level 1B orthorectified ASTER data in four different image processing software programs, ER MAPPER, expensive remote sensing software packages. However, for less experienced users from non-remote sensing different disciplines and remote sensing analysts are the target audience for the ERDAS Imagine. The user Fig. 16
than ERDAS Imagine, consisting of a main access toolbar and a workspace (focus) screen ((Fig. 16).
). As a main military. Similar to Erdas Imagine, the ENVI software suite have built in compatibility with ArcGIS products.
ERDAS Imagine 2014, PCI Geomatica and a trial version of Exelis ENVI
backgrounds (i.e. geologists), ER Mapper provides a series of customized wizards to streamline image (Fig. 11
interface (Fig. 11)) consists of a series of menus with ribbons with ribbons to activate the processing tools. competitor with ERDAS Imagine, PCI Geomatica is targeting the same market, image processing specialists Fig. 20
The ENVI interface is similar in appearance to the ArcGIS 10.2 and above interface ((Fig. 20).
). A signifcant
# Compare the workflows/cost and final products derived from image processing in each software suite. processing tasks typically used (e.g. geophysics, ASTER processing, simple RADAR data processing). The Navigation through the ribbon is somewhat cumbersome, however a search tool is provided to quickly and remote sensing analysts. The software is geared towards more user interaction through the EASI drawcard of this software is the searchable toolbox (similar to ArcGIS), that allows rapid searching of
KAOLINITE MINE WORKINGS (Fig. 7
Interface is appealing due to its' simplicity and intuitive menu structure (Fig. 7).
). Tasks are made more efficient access tools. Erdas Imagine also has inbuilt compatibility for ArcGIS. The band ratio calculations were scripting environment and is considerably less button driven than ERDAS Imagine. required tools
LOCATION MAP Site location 1:
by quick access menu buttons that activate common wizard tools. more involved than the automated functions in ER Mapper as custom spatial models had to be built for EXELIS ENVI - Interface
The data were provided by Tarsis Resources Ltd N Kaolinite working each desired ratio (other than the kaolinite model). PCI GEOMATICS 2014 - Interface
Kaolinite
and covers a gold-silver prospect in SE Mexico, workings ER MAPPER INTERFACE with ASTER
Guerrero State and have been projected to NAD83 UTM PROCESSING WIZARDS BAND RATIO PROCESSING
(Fig. 1
zone 14 N (Fig. 1).
). The focus of the band ratio and 1
ASTER data processing in ERDAS Imagine is ERDAS IMAGINE 2014 - Interface
spectral index analysis will be to identify probable
2
BAND RATIO PROCESSING not a wizard-driven focus. The user must be
Coacoyula
areas of Kaolinite, Alunite and Sericite
Sericite.. ER Mapper has a specific ASTER familiar with the relevant tools and be able
The greater spectral resolution of ASTER datasets, Maxela Processing Wizard which displays, to generate custom spatial models to create
particularly in the shortwave infrared region, make it Site location 2: processes, models and compares the desired band ratios. A general set of band
Kaolinite working
possible to identify surface occurrences of such different sets of ASTER imagery. The indices models are available within the
alteration minerals. Several artisanal and commercial ASTER processing Wizard allows the Fig. 12
Indices tool ((Fig. 12),
), which while providing
Kaolinite ((Fig. 2)) mines exist in the area of interest
Fig. 2 user to identify individual bands that a list of indices and band ratios to choose
(AOI) and provide a coarse assessment of accuracy of the (Fig. 8
form the desired ratio (Fig. 8).
). Figure 7 from is by no means exhaustive. To access
final kaolinite data products generated from band ratios. the spatial model information
information,, the model
must be reviewed to see what processing is
ASTER DATA PROCESSING WIZARD - BAND RATIOS carried out on the data (Fig.(Fig. 13
13).
). The clay index
Figure 1 Figure 2 illustrated in Figure 1313,, represents a ratio of
band 4 to band 6 and thus represents the ratio
used for kaolinite in this study. Similar spatial Figure 20
models were built using the spatial modeler Figure 16
sericite.. Conditional
tool for both alunite and sericite
BAND RATIO PROCESSING
ASTER DATA logic is applied in the model to remove zero
(Figs. 14-15
values (Figs. ).
14-15).
Figure 11
BAND RATIO PROCESSING Band ratio calculations of ASTER data can be carried out using two specific tools in EXELIS ENVI that are used
Band ratio calculations of ASTER data can be carried out in several ways in PCI Geomatica. The simplest is for both simple and complicated raster calculations, Band Ratio ((Fig.
Fig. 21
21)) and Band Math tools ((Fig.
Fig. 22
22).
). Both
INDICES TOOL - ASTER CLAY to use the raster calculator tool ((Fig.
Fig. 17
17),
), which is adequate zero values are removed before performing the were used to generate the three band ratios for kaolinite, alunite and sericite. The Band math tool allows a
The ASTER sensor is an imaging instrument flown on the Terra satellite which was launched in December 1999. ASTER has been designed to acquire land surface temperature
temperature,, calculations. The second tool involves the use of the Ratio Transformation algorithm ((Fig.
Fig. 18
18),
), which provides
emissivity, reflectance, and elevation data and is a cooperative effort between NASA and the JapaneseMinistry of Economy, Trade, and Industry (METI)*1. series of conditional statements to be written, that can result in a series of different tasks to be included in
more user input with respect to input and output parameters. A mask area can be defined and conditions can one statement (for example vegetation, water masking, band ratioing and final threshold definition). A
ASTER data consists of 14 data channels that cover ranges of visible, near infrared (VNIR), shortwave be set with regards to dealing with zero values. A combination of both the raster calculator and the RTR true/false result is converted into binary format, 1 for true and 0 for false.
infrared (SWIR) and thermal infrared (TIR) regions of the electromagnetic spectrum. It provides higher ASTER BAND INFORMATION CLAY SPATIAL MODEL - PRE BUILT (Kaolinite) algorithm was used to create the final three band ratios. Band ratio calculations were easier to perform in
spatial, spectral, and radiometric resolutions than traditional Landsat data (Gabr et al. 2010). Each ASTER PCI geomatica due to not have to build custom spatial models and still allowing the capability to handle zero
Figure 8
values in the denominator. BAND RATIO TOOL
scene covers an area of 60×60 km2. The VNIR bands have a spatial resolution of 15 m, the SWIR bands
have a spatial resolution of 30m and the TIR bands have a spatial resolution of 90 m. The ASTER channels An additional Wizard, The ASTER Mineral and Vegetation processing wizard wizard,, customizes the task a step
are more contiguous in the short wave infrared region than those of Landsat, yielding increased accuracy further and allows the user to specify a particular mineral species they are seeking, rather than selection of
in the spectral identification of rocks and minerals (Gabr et al. 2010). Figure 3 provides a summary of band RATIO TRANSFORMATION TOOL BAND MATH TOOL
(Fig. 9
band ratios (Fig. 9).
). Additionally, the tool automatically removes 0 value raster cells from the calculations to RASTER CALCULATOR
and wavelength information for the ASTER sensor. ensure there are no zero value denominators during band ratio processing. The drawback of this second
wizard is that the band ratios are pre-defined and may not be the selection the user is most interested, as is
A variety of processed products are available for ASTER data and are outlined https://lpdaac.usgs.gov/ Figure 3: Aster/ Landsat
Thermal mapper the case in this study where kaolinite is to be targeting using a band 4 / band 6 ratio, however, the ER mapper
products/aster_products_table. ASTER level 1B products
products,, that were used in this study represent registered Figure 13
wavelength comparison automated wizard offers Band 7/ Band 5. Similarly, Alunite is to be mapped with band 4/ band 6, and ER
radiance at the sensor product and as such contain radiometrically calibrated and geometrically co-registered
Mapper automated wizard offers Alunite-kaolinite – pyrophyllite spectrum defined by
data for the acquired channels*2. The Level-1B data set is produced by applying the radiometric calibration
(band 4 + Band 6)/ Band 5. In order to achieve the same band ratios as outlined in the Workflow section
and geometric correction coefficients to the Level-1A *2. The images are georeferenced to the WGS 84
above, a combination of the two ASTER processing wizards was utilized
utilized.. Masking was also not offered in the Figure 12
datum and Universal Transverse Mercator projection*1.
automated workflow and had to be completed prior to use in ER mapper.
ALUNITE BAND RATIO SPATIAL MODEL
ASTER MINERAL & VEGETATION WIZARD

WORKFLOW Figure 18

A cloud-free Level 1B ortho-rectified ASTER scene was provided in three separate files based on wavelength grouping (VNIR,
(VNIR, SWIR and TIR
TIR).
). For this study only the VNIR and
SWIR datasets were used. The imagery was acquired on January 24, 2006 and was pre-processed by PhotoSat Inc. The assumption has been made that several key pre-
processing steps, including an atmospheric correction to convert radiance to spectral reflectance has been performed on the data set used in this study. A schematic work flow Figure 22
of image processing procedures performed for this study is provided in Figure 4. The scene was subset to a smaller Area of Interest (AOI) encompassing the Erika Property of
Tarsis Resources Ltd. Note that Crosstalk is an effect in ASTER imagery caused by signal leakage from band 4 into adjacent bands 5 and 9*3. A cross talk correction (not done is
this study) can be applied to the data using a custom tool called ERSDAC Crosstalk 3. Figure 17
Figure 13

ASTER DATA PROCESSING - GENERAL WORKFLOW SERICITE BAND RATIO SPATIAL MODEL OUTPUT RESULTS Figure 21
THIS STUDY Figure 9
Once the models were run, the data were output as floating point data to an .img file. The files were
Obtain IMAGE ROTATION ATMOSPHERIC imported into ArcGIS 10.2.2 and conditional statements were applied to extract cells exceeding specific
ASTER data
CROSSTALK DATA
and
ORTHORECTIFICATION RADIANCE CORRECTION/
MASKING BAND
After selecting the appropriate bands in ASTER Wizard 1 and Wizard 2, the results are output initially to a thresholds for each mineral. The final map results are identicalto those produced from ERDAS imaging and OUTPUT RESULTS
CALIBRATION DARK PIXEL
Correlation IMPORT CORRECTION (Vegetation,
Cloud, Water)
RATIO
GENERATION
THRESHOLD
DEVELOPMENT FINAL
temporary blue-red colour ramp image. Each dataset is then permanently saved for final processing thusare not presented. The coincidence of high probability kaolinite occurrence with known kaolinite mines
MAP PRODUCT and threshold derivation of the data for interpretative purposes. Upon calculation of band ratios for sericite, kaolinite and alunite, the data were imported to ArcGIS 10.2.2
Free for
US, Aust
Conversion of red - predicted Kaolinite north of Coacoyula are presented in Figure 19
19.. and the same thresholding process that was used for the ERDAS and PCI products were applied to the ENVI
satellite orientation 1. ATCOR correction Generation of
Stand alone open Single band import to ground Rescaling of to convert radiance Masking different Creation of thresholds
$80.0 per scene source software and creation of spatial reference DN to Top of to spectral reflectance image components that of specific band to identify high Final map products. The final output maps are identical to those presented in Figure 15 and are therefore not displayed
international
Most typical -
to correct for
Signal leakage from
band 4 into band 5
three separate
Multispectral layer
frame Atmosphere radiance.
or
will likely generate
false analysis results
ratios relevant to
problem tobe
analysed
zones of high probability
of alteration mineral
product for use in a
variety of GIS platforms OUTPUT RESULTS N
again.
Level 1B and band 9
stacks Projection to desired 2. Dark Pixel correction due to mixing of spectral e.g. alteration occurrence KAOLINITE MINES vs. PREDICTED KAOLINITE OCCURRENCES
VNIR, SWIR, TIR co-ordinate system (min value in each dataset
is subtracted from the
signatures mineral ID Three maps presented in Figure 10 illustrate the
Format - HDF NAD 83 UTM z14 data) distribution of possible occurrences of Kaolinite, Alunite and
Sericite for the subset area in Guerrero State, Mexico. The
PRE PROCESSING ALREADY COMPLETED maps have a threshold applied to the data to delineate only Figure 14
N
Figure 4 those areas representing the highest probability that the
mineral of interest may occur at any given location. ER Mapper OUTPUT RESULTS red - predicted Kaolinite
DISCUSSION
DATA PREPARATION ASTER band ratio results are rescaled from a floating point Once the models were run, the data were N
value to a 0 – 255 range. The statistical distribution of each output as floating point data to an .img file. Despite the uncertainty of the thoroughness of the pre-processing procedures completed by PhotoSat Inc.
Necessary pre-processing of ASTER datasets prior to any band math calculations, includes (i) Resampling to (15 m or NDVI dataset was reviewed and the highest 1 % of the dataset
30 m) to match the spatial resolution between the SWIR and VNIR, (ii) Radiance Calibration
Calibration,, (iii) Atmosphere Correction
Correction,, The files were imported into ArcGIS 10.2.2 and and the lack of any masking procedures, the ASTER band ratios exhibit some important results. The
were retained for final map output. conditional statements were applied to extract coincidence of high probability kaolinite and alunite occurrences derived from the ASTER dataset with known
(iv) Image area subsetting and (v) Vegetation Masking
Masking.. All steps were conducted with the exception of vegetation blue - outline of kaolinite mine
masking. Upon review of the NDVI index map for the ASTER image (Fig. (Fig. 55),
), only limited areas in the southeastern corner cells exceeding specific thresholds for each active kaolinite mines within the AOI indicates the band ratio technique is successful in detected highly
N
of the study area had any significant vegetation cover (i.e. NDVI >0.2
>0.2).
). Vegetation cover during January in this part of A comparison of the highest probability occurrences of Kaolinite as mapped by band ratios from the ASTER Fig. 15
mineral(Fig.
mineral( 15).
). Similar to ER Mapper, the Kaolinite anomalous surface zones of kaolinite and alunite. If only the highest threshold values are considered, several
Mexico is typically minimal and any addition to surface reflectance values due to dense vegetation is considered imagery with known locations of Kaolinite mine workings in the area north of the town of Coacoyula is Kaolinite band ratio identifies the known kaolinite Mine anomalous areas can be identified from the dataset. Figure 23 illustrates 3-4 coincident kaolinite-alunite
minimal. The pre-processed ASTER data for the AOI, then became the input data for the band ratio procedures carried partly dependent on timing of ASTER data acquisition. A review of Google Earth indicates that at the time mines that were operating during the time of anomalies that are worthy of further scrutiny as they are associated with highly favourable geological features
out in this study in the four image processing software packages. the ASTER data was acquired (Jan 24, 2006) the two large Kaolinite mines to the NW of the town were not ASTER data acquisition
acquisition.. In the case of sericite, consistent with sediment-hosted gold mineralization. Ground truthing of the preliminary results would be
yet in operation are were characterized by significant vegetation cover. The main kaolinite workings in the band ratios mostly correspond with water the only way to validate the datasets.
2006 were immediately north of the town of Coacoyula and the kaolinite band ratios defined by ER Mapper bodies and vegetation, rather than known
IMAGE ANALYSIS – BAND RATIOS and INDICES identify this zone with reasonable accuracy. ground occurrences. Yellow - outline of kaolinite mine
The sericite band ratio was not useful for ground targeting purposes as the final output results coincided
Band ratio calculation is a technique in which one or more bands are divided by other bands in order to enhance with water and vegetation, which were not masked prior to data processing.
spectral differences and assist in delineation of specified surface units. The band with the higher reflectance is assigned ER Mapper ALUNITE ER Mapper KAOLINITE ER Mapper SERICITE ERDAS SERICITE ERDAS ALUNITE ERDAS KAOLINITE
to the numerator and the band with diagnositic absorption trough is assigned to the denominator
denominator.. In the case of kaolinite ASTER ANOMALIES ASTER + GEOLOGICAL MAP
(Fig. 6
and alunite alteration minerals (Fig. 6;; Rouskov et al. 2005), bands in which strong absorption occurs (5 and 6;
denominator) are divided by bands with high reflectance (4; numerator). The resultant grey-scale image represents a N N N N N N
relative band intensity image and areas within the image that appear bright are regions that represent the highest Figure 19 N N N
possibility for that specific mineral species to occur
occur.. The benefits of band ratio techniques include removal of gold-in-soil gold-in-soil
topographic shading, suppression of albedo effects, and enhancement of relative colour differences (Rencz, 1999). anomaly anomaly
To ensure optimal results from band ratio methods, the dataset must be corrected from system contributions or
environmental signals that will add to the DN values measured (Rencz, 1999).

REFERENCES
Kaolinite Figure 6:
Representative *1 http://www.yale.edu/ceo/Documentation/ASTER.pdf
ASTER spectral
reflectance *2 https://lpdaac.usgs.gov/products/aster_products_table/ast_l1b
signatures for
Kaolinite and *3 http://www.ga.gov.au/webtemp/image_cache/GA7833.pdf - ASTER Mineral Index Processing
Alunite (from -
http://www.isprs.org Manual
/proceedings/XXXV/
Alunite congress/comm7/ Rencz, A.N (editor) 1999. Manual of Remote Sensing, Vol. 3: Remote Sensing for the
papers/15.pdf) Earth Sciences. Wiley, 202 pp.

Gabr, S., Ghulam, A., and Kusky., T. 2010. Detecting Areas of high-potential gold mineralization
Figure 23
using ASTER data. Ore Geology Reviews 38: 56 -69.
Three band ratios were applied to the ASTER scene to test for the presence of Kaolinite, Alunite and Sericite.
The band ratios used for the different mineral species include: Rouskov, K., Popov, K., Stoykov., S. and Yamaguchi., Y. 2005. Some Applications of the Remote The four image processing software packages all produce very similar results for the three simple band ratios
examined in this study. ER Mapper is the simplest and most automated of the four, and ENVI provides the
Alunite (Advanced Argillic alteration): Band 4/ Band 5 Sensing in Geology by using ASTER Images. Retrieved http://www.space.bas.bg/sens/Ses2005/r5.pdf most straightforward method to conduct more complicated raster calculations. ERDAS Imagine provides the
Kaolinite (Intermediate Argillic alteration): Band 4/ Band 6 option of spatial model generation as opposed to scripted models, although these were more time-consuming
Figure 5: NDVI map for AOI; Values >0.2 vegetation
Figure 10 Figure 15 to create than the pre-built band ratio tools in PCI Geomatica and Exelis ENVI. ER Mapper produced a 0-255
Sericite (Phyllic alteration): (Band 5 + Band 7)/band 6 This project is a student project and was completed for education purposes. scaled output whereas the other three provided an option for floating point outputs.
DISCLAIMER The poster should not be reproduced or distributed in any format.

You might also like