+ All Categories
Home > Documents > Paper Final Ronen Alid

Paper Final Ronen Alid

Date post: 17-Nov-2015
Category:
Upload: fanakiri-asa
View: 16 times
Download: 1 times
Share this document with a friend
Description:
Paper Final oleh Ronen Ali.
Popular Tags:
27
This article was downloaded by:[Gersman, Ronen] [University of Southern California] On: 16 June 2008 Access Details: [subscription number 788774280] Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK International Journal of Remote Sensing Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713722504 Mapping of hydrothermally altered rocks by the EO-1 Hyperion sensor, Northern Danakil Depression, Eritrea Ronen Gersman ab ; Eyal Ben-Dor b ; Michael Beyth c ; Dov Avigad a ; Michael Abraha d ; Alem Kibreab d a Institute of Earth Sciences, The Hebrew University, Jerusalem, Israel b Department of Geography, Tel Aviv University, Israel c Geological Survey of Israel, Jerusalem d Eritrean Department of Mines, Asmara, Eritrea Online Publication Date: 01 January 2008 To cite this Article: Gersman, Ronen, Ben-Dor, Eyal, Beyth, Michael, Avigad, Dov, Abraha, Michael and Kibreab, Alem (2008) 'Mapping of hydrothermally altered rocks by the EO-1 Hyperion sensor, Northern Danakil Depression, Eritrea', International Journal of Remote Sensing, 29:13, 3911 — 3936 To link to this article: DOI: 10.1080/01431160701874587 URL: http://dx.doi.org/10.1080/01431160701874587 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.
Transcript
  • This article was downloaded by:[Gersman, Ronen][University of Southern California]

    On: 16 June 2008Access Details: [subscription number 788774280]Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

    International Journal of RemoteSensingPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713722504

    Mapping of hydrothermally altered rocks by the EO-1Hyperion sensor, Northern Danakil Depression, EritreaRonen Gersman ab; Eyal Ben-Dor b; Michael Beyth c; Dov Avigad a; MichaelAbraha d; Alem Kibreab da Institute of Earth Sciences, The Hebrew University, Jerusalem, Israelb Department of Geography, Tel Aviv University, Israelc Geological Survey of Israel, Jerusalemd Eritrean Department of Mines, Asmara, Eritrea

    Online Publication Date: 01 January 2008

    To cite this Article: Gersman, Ronen, Ben-Dor, Eyal, Beyth, Michael, Avigad,Dov, Abraha, Michael and Kibreab, Alem (2008) 'Mapping of hydrothermally altered rocks by the EO-1 Hyperionsensor, Northern Danakil Depression, Eritrea', International Journal of Remote Sensing, 29:13, 3911 3936

    To link to this article: DOI: 10.1080/01431160701874587URL: http://dx.doi.org/10.1080/01431160701874587

    PLEASE SCROLL DOWN FOR ARTICLE

    Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

    This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expresslyforbidden.

    The publisher does not give any warranty express or implied or make any representation that the contents will becomplete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should beindependently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with orarising out of the use of this material.

    http://www.informaworld.com/smpp/title~content=t713722504http://dx.doi.org/10.1080/01431160701874587http://www.informaworld.com/terms-and-conditions-of-access.pdf

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    Mapping of hydrothermally altered rocks by the EO-1 Hyperion sensor,Northern Danakil Depression, Eritrea

    RONEN GERSMAN*{{, EYAL BEN-DOR{, MICHAEL BEYTH, DOVAVIGAD{, MICHAEL ABRAHA" and ALEM KIBREAB"

    {Institute of Earth Sciences, The Hebrew University, Givat Ram, Jerusalem, 91904,Israel

    {Department of Geography, Tel Aviv University, Ramat Aviv, 69978, IsraelGeological Survey of Israel, 30 Malkhei Israel St., Jerusalem, 95501

    "Eritrean Department of Mines, Liberty Avenue, 213, PO Box 272, Asmara, Eritrea

    (Received 15 March 2007; in final form 1 June 2007 )

    An EO-1 Hyperion scene was used to identify and map hydrothermally altered

    rocks and a Precambrian metamorphic sequence at and around the Alid volcanic

    dome, at the northern Danakil Depression, Eritrea. Mapping was coupled with

    laboratory analyses, including reflectance measurements, X-ray diffraction, and

    petrographic examination of selected rock samples. Thematic maps were

    compiled from the dataset, which was carefully pre-processed to evaluate and

    to correct interferences in the data. Despite the difficulties, lithological mapping

    using narrow spectral bands proved possible. A spectral signature attributed to

    ammonium was detected in the laboratory measurements of hydrothermally

    altered rocks from Alid. This was expressed as spectral absorption clues in the

    atmospherically corrected cube, at the known hydrothermally altered areas. The

    existence of ammonium in hydrothermally altered rocks within the Alid dome

    has been confirmed by previous studies. Spectral information of endmembers

    mineralogy found in the area (e.g. dolomite) enables a surface mineral map to be

    produced that stands in good agreement with the known geology along the

    overpass. These maps are the first hyperspectral overview of the surface

    mineralogy in this arid terrain and may be used as a base for future studies of

    remote areas such as the Danakil.

    1. Introduction

    The Hyperion imaging spectrometer, mounted on the NASA Earth Observing 1

    (EO-1) platform, is the first high spectral resolution imaging spectrometer that

    routinely acquires data from orbit (Pearlman et al. 2003, Ungar et al. 2003). During

    the time passed since its launch in November 2000, major work has been dedicated

    to checking its performances and capabilities (e.g. Cudahy et al. 2002, Green et al.

    2003, Pearlman et al. 2003, Thome et al. 2003). Researchers have evaluated its

    capabilities for studies of vegetation (Asner and Heidebrecht 2003, Datt et al. 2003,

    Goodenough et al. 2003), marine sciences (Kruse 2003) and geological purposes

    (Cudahy et al. 2001, Hubbard et al. 2003, Kruse et al. 2003). Studies that compared

    *Corresponding author. Present address: University of Southern California,Department of Earth Sciences, 3651 Trousdale Pkwy, Los-Angeles, CA 90089-0740,USA. Email: [email protected]

    International Journal of Remote Sensing

    Vol. 29, No. 13, 10 July 2008, 39113936

    International Journal of Remote SensingISSN 0143-1161 print/ISSN 1366-5901 online # 2008 Taylor & Francis

    http://www.tandf.co.uk/journalsDOI: 10.1080/01431160701874587

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    the performances of the Hyperion and spaceborne multispectral sensors and

    between the Hyperion and the airborne AVIRIS (Airborne Visible/Infrared Imaging

    Spectrometer) imaging spectrometer concluded that the Hyperion has a low signal-

    to-noise ratio (SNR), relative to other airborne sensors such as AVIRIS or HyMAP

    (Datt et al. 2003, Hubbard et al. 2003, Kruse et al. 2003). The low SNR is probably

    the major significant disadvantage of the sensor performance. Furthermore, it was

    shown that the Hyperion sensor has severe problems arising from the line-curvature

    effect, which eventually prevent adequate atmospheric rectification and accordingly

    precise thematic evaluation (Goodenough et al. 2003, Neville et al. 2003).

    Nevertheless, additional geoscience applications of the Hyperion data are needed

    for a better understanding of the advances and limitations of imaging spectroscopy

    as a routine research tool.

    In this study we used a single Hyperion scene acquired over east Eritrea to identify

    and map hydrothermally altered rocks and the Neoproterozoic basement within and

    around the Alid volcanic dome, located at the northern termination of the Danakil

    Depression. Geo-referenced ASTER scenes (Abrams et al. 2002) were used to

    examine the area on a broader scale and as a reference for rectifying the Hyperion

    scene. Our goal was to study the feasibility of the Hyperion data, calibrated and

    coupled with a comprehensive field study and laboratory studies of an arid and well-

    exposed area, as a tool for use in locating natural resources.

    The northern Danakil Depression and the Alid volcanic dome are of scientific and

    economic interest (Souriot and Brun, 1992, Drury et al. 1994, Beyth 1996, Ghebreab

    1998, Lowenstern et al. 1997, 1998). This area has never been studied by any

    hyperspectral sensor and only rarely studied on the ground. Therefore the

    opportunity of examining orbital hyperspectral data is significantly important in

    order to demonstrate that hyperspectral data is capable of precisely mapping remote

    areas.

    Important ground targets in this work were hydrothermally altered rocks within

    the Alid dome for their significant economical potential. Previous studies of Alids

    hydrothermal hot springs and fumaroles (Beyth 1996, Duffield et al. 1997) assessed

    its geothermal and epithermal gold potential. Both studies reported the existence of

    ammonium sulphate minerals. Ammonium-bearing minerals are associated with

    hydrothermally altered rocks, hydrocarbons in oil shale, and coal-bearing

    formations (Krohn et al. 1993), which are found to occur in several economically

    important geological environments (Krohn and Altaner 1987, Baugh et al. 1998).

    Yet, the role and importance of geologic nitrogen is conspicuously absent from most

    reviews of global nitrogen dynamics (Holloway and Dahlgren 2002). Knowledge of

    the origin and mineralogical relations of ammonium minerals in known

    hydrothermal systems is critical for the proper interpretation of remote sensing

    data and for testing possible links to mineralization (Krohn et al. 1993).

    2. Geological setting

    The Danakil Depression is the terminus of an embryonic spreading axis the

    northern part of the Afar triangle, which formed between the African plate on the

    west and the Danakil microplate on the east (figure 1). According to palaeomagnetic

    data, the Danakil microplate rotated 10u anti-clockwise during the past 4 millionyears (Souriot and Brun 1992, Ghebreab 1998). The Alid volcanic centre is located

    along the axis of the NNWSSE trending Danakil depression, informally known as

    the Alid graben (Duffield et al., 1997). The graben is topographically and

    3912 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    structurally limited on its western side by east-dipping normal faults which form the

    Red Sea escarpment (Drury et al. 1994, Duffield et al. 1997, Beyth et al. 2003).

    Neoproterozoic basement rocks, related to the East African Orogen (Drury et al.

    1994, Ghebreab, 1999a,b; Beyth et al. 2003, Ghebreab et al. 2005) on the western

    side of the graben rise up to 2500 m. The eastern topographic and structural

    boundary is marked by a steep and abrupt 300 m high west-dipping normal fault

    escarpment. Basement rocks are locally exposed at the base of this scarp and are

    unconformably overlain by a post-Miocene sequence of intercalated sedimentary

    and volcanic deposits (Duffield et al. 1997, Lowenstern et al. 1997).

    The Alid Dome, which rises about 900 m above sea level was interpreted by

    Duffield et al. (1997), Lowenstern et al. (1997) and Lowenstern et al. (1998) to be a

    structural dome rather than a classical volcano. They suggested that a shallow

    felsicgranitic magma body which lies about 25 km beneath the top of the

    mountain is responsible for a 1 km uplift of the overburden rocks. The volcanic

    sequences in the Alid include basalts, pumiceous rhyolites (some with xenoliths of

    basement rocks and of older rhyolite and basalt) and a sedimentary sequence with

    pillow basalt, submarine debris flow and shallow marine sediments (see Alid series

    in figure 1). Calculated 40Argon/39Argon ages of the volcanic succession near Alid

    range from 1.2 million years to 14 500 years (Duffield et al. 1997). Younger basalt

    Figure 1. The working area. (a) The Alid-area map is based on an ASTER scene and afterAbbate et al. (2004), Beyth (1996), Duffield et al. (1997) and Gersman et al. (2006). (b)Regional tectonic setting, modified after Lowenstern et al. (1997). Dark grey is mostlyPrecambrian basement, pale grey is mostly Cenozoic volcanics and sediments. Straight linesrepresent major spreading axes, in which the Danakil depression extends northwest from theAfar depression. The small frame represents the location of (a).

    Hyperion mapping in the Danakil Depression, Eritrea 3913

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    flows are estimated to be Holocene in age. Outcrops of Neoproterozoic pelitic

    biotitekyaniteschist of the metamorphic Bizen Domain (Beyth et al. 2003,

    Ghebreab et al. 2005) are exposed on the eastern side of the dome beneath the

    volcanicsedimentary succession, as a result of the structural updoming (Beyth

    1996, Duffield et al. 1997, Lowenstern et al. 1997, 1998, Gersman et al. 2006). Much

    of the pyroclastics and lava flows within the Alid dome have been strongly altered

    by hydrothermal activity (Beyth 1996, Duffield et al. 1997, Lowenstern et al. 1997,

    1998, Gersman et al. 2006). Beyth (1996) and Duffield et al. (1997) reported

    kaolinite, gypsum, anhydrite, montmorillonite, illite and ammonium sulphate

    minerals (tschermigite NH4Al(SO4)2 12H2O, mascagnite (NH4)2(SO4), kokatite

    (NH4)2Ca(SO4)2 H2O) around hydrothermal fumaroles and hot springs within the

    Alid dome. The continuing extension of crustal spreading across the Alid graben is

    creating north/northwest-trending faults and fissures (Duffield et al. 1997). A set of

    eastwest fractures was documented by the authors in the recent basaltic cover north

    of the Alid dome.

    3. Methods

    The study combined analysis of hyperspectral data as acquired from orbit

    (Hyperion) followed by a comprehensive field study, and a laboratory spectral

    and mineralogical study of rock samples.

    3.1 Field work

    A field reconnaissance was carried out between 5 and 15 February 2005.

    Geographical locations were measured by a Magellan GPS (SportTrack colour)

    with an average accuracy ,7 m. Samples for laboratory studies included fresh andsurface-weathered sides of representative rocks. Samples of hydrothermally altered

    rocks were taken from two sites within the Alid caldera: fumaroles in Darere

    (14.877uN, 39.915uE) and hot springs in Illegedi (14.882uN, 39.926uE). Samples ofhomogeneous alluvial fans and playas for atmospheric calibration (Ben-Dor et al.

    1994, Clark et al. 2002, Crouvi 2002, Rowan and Mars 2003) were collected from

    measured homogeneous areas of 60 m660 m, representing a square of 4 pixels in animage of 30 m pixel size.

    3.2 Laboratory work

    3.2.1 Mineralogy. The mineralogy of fine grained samples was studied using the

    X-ray diffraction (XRD) technique for bulk mineralogy (e.g. Warren 1990). The

    samples included hydrothermally altered rocks and alluvial/aeolian deposits for the

    atmospheric calibration. The XRD analyses were carried out on bulk powder using

    a Philips diffractometer PW 1830/3710/3020 of the Geological Survey of Israel.

    3.2.2 Petrography. The petrography of high grade pelitic schists and orthogneiss,

    basaltic/andesitic dykes, low-grade metavolcanic phyllite, mylonite, graphite schist

    and amphibolite gneiss were studied under an optical polarized microscope.

    3.2.3 Reflectance spectral analysis. Spectral measurements were carried out using

    the Analytical Spectral Devices Inc. (ASD) field spectrometer (FieldSpec pro FR)

    with a spectral range of 3502500 nm. The spectral resolution is 3 nm at 700 nm and

    10 nm at 1400/2500 nm. Sampling interval is 1.4 nm at 3501050 nm and 2 nm at

    10002500 nm (http://www.asdi.com). The measurements were performed at the

    3914 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    Remote Sensing laboratory in Tel-Aviv University using a contact probe and a

    built-in illumination source (A 4.5 W HalogenKrypton lamp with AC to DC

    current converter).

    The reflectance spectra were calibrated against a Halon plate (Weidner and Hsia

    1981). Black current calibration and white reference measurements were performed

    approximately every 30 min and the stability of the spectrometer was examined

    against a standard sample (dolomite).

    Each rock sample was measured in duplicate on both its fresh and weathered

    sides. Each calibration sample composed of unconsolidated sand/pebbles was mixed

    between every duplicate recording. The spectra of each calibration sample were

    averaged to one representative spectrum. The collected spectra were run through the

    continuum removal (CR) technique (Clark and Roush 1984) in order to recognize

    and allocate diagnostic spectral features.

    3.3 Hyperion data processing

    Very close to the field trip we ordered the acquisition of a single Hyperion scene that

    took place on 2 February 2005. Processing the scene was carried out using the

    ENvironment for Visualizing Images (ENVI 2003) software packages, version 4.0

    and other accessory codes to re-process the data.

    The Hyperion sensor provides continuous spectral coverage of 242 bands, in a

    10 nm (average) sampling interval over the reflected spectrum from 400 to 2500 nm.

    The instrument consists of two detectors. The VNIR (VIS +NIR) detector covers aspectral range of 4001000 nm in 70 channels and the SWIR detector covers the

    range of 9002500 nm in 172 channels (Ungar et al. 2003). There are 21 channels in

    the overlapping spectral range. Each scene is collected as a narrow strip, covering a

    ground area of approximately 7.7 km in the across-track direction, and 75 km (in

    our case) in the along-track direction in a pushbroom configuration, from an

    altitude of 705 km (Beck 2003, Ungar et al. 2003). The ground sample distance (pixel

    size) is 30 m for all bands.

    3.3.1 General. The processing of the Hyperion data is schematically summarized

    in figure 2. Pre-processing steps included the removal of overlapping and inactive

    bands (17, 5878 and 225242, after Barry 2001), and of the first and last samples

    of the scene. The radiometric calibration to convert raw DN to W m22 mm21 sr21

    units was done after Beck (2003). In the end this pre-processing step enables

    determining cube dimensions of 196 spectral bands, 254 columns and 2500 lines.

    Several processes were performed removing the atmosphere to filter out some data

    noises, and estimate the quality of the data as a spectral database.

    3.3.2 SNR estimation. SNR was estimated from the corrected radiometric image.

    Three homogeneous targets were selected for this purpose (figure 3) and the SNR

    was estimated according to Green et al. (2003) and Kruse et al. (2003). The SNR

    values are 90 : 1 in the VIS range, 60 : 1 at SWIR-I (,1000 to ,1600 nm (excludingthe water vapour signal at ,1400 nm)) and about 35 : 1 in SWIR-II (,20002400 nm). Kruse et al. (2003) reported SNR values of about 25 : 1 for the SWIR-II

    range, which is in good agreement with our data.

    3.3.3 Atmospheric correction. A two-stage atmospheric correction was applied to

    the scene, separated by a removal of vertical lines (de-striping) and residual

    noise stages. The two stages were radiative-modelled atmospheric correction and a

    Hyperion mapping in the Danakil Depression, Eritrea 3915

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    fine-tuning atmospheric correction using the empirical line (EL) technique. Noise

    reduction steps were performed between the two atmospheric correction stages.

    3.3.4 Radiative-modelled atmospheric correction. The principal atmospheric cor-

    rection was carried out by using the Atmospheric CORrection Now (ACORN)

    software that is based on licensed MODTRAN technology to determine the

    atmospheric parameters (Kruse et al. 2003; ACORNTM 5.0, 2004). A few correction

    sessions were carried out with different input values of estimated water vapour,

    atmospheric visibility, average altitude and artefact suppression type (ACORNTM

    5.0, 2004). The Thompson Ramo Woolridge (TRW) curvature function of the

    Hyperion (Pearlman et al. 2003) was integrated in the ACORN correction.

    3.3.5 Noise reduction. Line curvature estimation and correction: line curvature

    (smile effect), which significantly exists in pushbroom systems, refers to an across-

    track shift from a centre wavelength, which is due to the change of dispersion angle

    with field position (Goodenough et al. 2003, Neville et al. 2003). For the VNIR

    bands, the shifts range between 2.6 and 3.5 nm whereas for the SWIR bands the

    shifts are less than 1 nm (Goodenough et al. 2003). The VNIR variations are about

    30% with respect to the 10 nm full width at half maximum (FWHM) of the Hyperion

    Figure 2. The methodology of processing and interpreting the Hyperion data.

    3916 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    instrument (ACORNTM 5.0, 2004) and cannot be ignored as they may alter the pixel

    spectral response and reduce the total classification accuracies.

    Datt et al. (2003) and Goodenough et al. (2003) reported that the smile effect has

    a visual expression, as a vertical brightness gradient in the first eigenband of the

    radiometric corrected Hyperion cube. To estimate the line curvature effect, we

    applied the minimum noise fraction (MNF) transformation (Green et al. 1988,

    Jensen 2005) to the raw data (DN) image and to the radiometric corrected image. In

    both of the images the brightness gradient was clearly evident in the VNIR range

    and was less significant in the SWIR range, where the across-track shifts are

    reported to be rather small (figure 4). The line curvature effect was partly corrected

    by the ACORN 5.0 radiative-model atmospheric correction using the TRW

    parameters measured prior the launch (ACORNTM 5.0, 2004). However, as the

    smile effect is a dynamic phenomenon and might change from one scene to another

    (K. Staenze, personal communication), the spectral calibration data may have

    changed from the provided TRW parameters. Goodenough et al. (2003) and

    Pearlman et al. (2003) reported that changes of about 1 nm are possible since the

    sensor launch. This leads to residual curvature effects remaining and calls for

    specific correction (see below).

    Vertical stripe removal (de-striping): a de-striping procedure was applied to the

    data in order to overcome a spatial problem which further reduced the remaining

    curvature effect. Vertical stripes are often seen in data acquired using pushbroom

    Figure 3. Locations of regions for SNR estimation (a) and their SNR values (b): fan 1 (1),fan 2 (2), and stream (3).

    Hyperion mapping in the Danakil Depression, Eritrea 3917

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    (line-array) technology and may be caused by factors such as detector nonlinearities,

    movement of the slit with respect to the focal plane, and temperature effects (Kruse

    et al. 2003). The correction of the vertical stripes is based on adjusting the brightness

    of each image column (in all bands) based on a calculated offset relative to the

    average detector response of the scene (Kruse 1988, Kruse et al. 2003). In this study

    the de-striping process followed the first atmospheric correction to minimize

    changes in the raw data. The procedure was done according to Kruse (1988) and

    Kruse et al. (2003). Asner and Heidebrecht (2003), Cudahy et al. (2002), Datt et al.

    (2003) and Goodenough et al. (2003) have also reported successful vertical de-

    striping results using similar methods. The de-striping process reduced the residual

    line-array effect to a negligible level (figures 4 and 5) and slightly improved the SNR,

    especially in the NIRSWIR range (figure 6).

    Residual noise removal: a MNF procedure (ENVI 2003, Green et al. 1988, Jensen

    2005) was used to reduce the residual noise after the de-striping process. According

    to Datt et al. (2003), it is best to handle the VNIR and SWIR data separately and

    combine them after the MNF filtering process, rather than use the entire spectral

    range. In this study the MNF filtering process (forward MNF, removing the noisy

    channels, backward MNF) yielded similar results when all the bands were treated as

    a whole, and when the VNIR and the SWIR bands were treated separately. The

    transition between the VNIR and the SWIR sensors is reflected in places as a

    spectral jump between bands 57 (932.41 nm) and 79 (932.64 nm) (figure 7).

    Misalignment of the SWIR and VNIR arrays, which was reported by

    Goodenough et al. (2003) exists in the MNF of the separate detectors but was

    not visible in the MNF result of all the 196 bands (whole range). Based on the lower

    jump between bands 57 and 79 (figure 7) of our data presented, we chose to work

    Figure 4. The removal of the line-array effect as expressed by the first MNF band of theVNIR range at three processing levels. Note the visual improvement between the raw data (a),the ACORN 5.0 smile corrected data (b) and the de-striped data (c).

    3918 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    with the MNF transformation of the entire spectral range (figure 8). The inverse

    MNF procedure included the first nine eigenbands based on the asymptotic curve

    behaviour obtained in eigenvalues higher than 9.

    Figure 5. Results of vertical stripes removal: band 216 (a) before and (b) after de-striping.

    Figure 6. The effect of de-striping on the SNR.

    Hyperion mapping in the Danakil Depression, Eritrea 3919

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    Figure 7. Comparison between results of residual noise reduction. Reflectance of the MNFtreatment of the entire spectral range (196 bands) and separate MNF of VNIR and SWIRranges (VNIR + SWIR). Gaps are due to removal of water vapour absorption bands (119132, 165190) and VNIRSWIR overlapping (bands 58, 79).

    Figure 8. Comparison between MNF eigenvalues of the radiomertically corrected, atmo-spherically corrected and de-striped data. The first 35 MNF bands of the total 196 areincluded.

    3920 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    3.3.6 Fine tuning atmospheric correction using the EL method. After the first

    treatment process, in which dead pixels were gently removed, curvature alignment

    was corrected, and the atmospheric attenuation was removed by ACORN, a fine

    tuning stage was applied to the data. This was done by generating meticulous

    calibration alignment using ground targets and the EL procedure (Ben-Dor et al.

    1994, Kruse 1998; Clark et al. 2002; ENVI 2003). For that purpose we examined

    three combinations of areas with different albedos: dark, semi-dark, semi-bright and

    bright (figure 9, table 1). The selected areas included recent basalt as the dark target

    (1 in figure 9); rhyolite pebbles (semi-dark target, 3 in figure 9); fine sand from a

    Figure 9. Locations of possible sites for empirical line (EL) correction: (1) basalt (sampleerr-86); (2) aeolian deposit (sample err-84); (3) rhyolite pebbles (sample err-83, this samplewas collected about 1500 m NNW of the ROI location); (4) fine sand (sample err-51c). Seetable 1 for the EL combinations and table 2 for compositions of samples err-84 and err-51c.

    Table 1. Components used for the empirical line (EL) corrections. Combination A yielded thebest result.

    EL correction

    Components for EL

    Fine Sand Rhyolite Pebbles Aeolian Deposit Recent Basalt

    A x xB x x xC x x x

    Hyperion mapping in the Danakil Depression, Eritrea 3921

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    fluvial plain (semi-bright target, 4 in figure 9); and an aeolian deposit (bright target,

    2 in figure 9). Selection of the best EL result was done after comparison between the

    Hyperion-corrected reflectance curves of eight known ground targets which served

    as validation sites and their corresponding laboratory-measured spectra. The

    comparison included visual examination and scoring of four spectral parameters

    (VNIR absorptions, SWIR absorptions, curve shape and total reflectance) and

    spectral correlation using the CORREL function of the EXCEL program. The

    correlations were applied to the VNIR range and the SWIR range separately, as well

    as to the entire spectral range, excluding the strong water absorption regions at 1400

    and 1900 nm.

    3.3.7 Classification. After setting up the atmospheric correction and receiving a

    confident ground reflectance cube, we moved forward for the thematic part. In this

    stage, we used two methods for end-member selection to classify the area. The first

    method followed the approach suggested by Kruse (1998) and Kruse et al. (2003),

    namely, the Pixel Purity Index (PPI). The basic concept is to reduce the data to

    manageable levels by finding those pixels in the image that can explain every other

    pixel in terms of a mixing model without a priori knowledge of the ground surface

    (Boardman 1993, 1998, Boardman et al. 1995, Kruse 1998). The spectrally purest

    (extreme) pixels in the image are allocated by repeatedly projecting n-dimensional

    scatter-plots onto a random unit vector. Given a user-threshold, those pixels that

    were recorded as extreme most of the times are defined as pure pixels (Kruse et al.

    1997). The PPI and the n-dimensional visualizer tools enabled selecting the major

    end-members in the scene. In the second method, diagnostic absorption bands of

    defined ground targets were identified and mapped based on a priori knowledge of

    the composition and the laboratory-measured reflectance signature of these targets.

    The Spectral Angle Mapper (SAM) (Kruse et al. 1993, Jensen 2005) is an excellent

    tool for that purpose and thus was used for classifying the images pixels. The

    strategy was to perform a general mapping using the PPI, and then to conduct an

    intimate mapping procedure for specific targets using the previous knowledge.

    3.3.8 Geometric correction. The Hyperion image was geo-referenced relative

    to a rectified ASTER scene, using a first-degree polynomial of selected

    ground control points (GCP). Geo-referencing was carried out using the

    Environmental Systems Research Institute (ESRI) ArcView software, version 9.0

    (http://www.esri.com/).

    4. Results

    4.1 Mineralogical data

    4.1.1 XRD analysis. The modal compositions of the hydrothermally altered rock

    samples from Darere and Illegedi (figure 1) are similar and include quartz,

    potassium-feldspar, plagioclase, kaolinite and sulphates (table 2). The samples used

    for the EL atmospheric removal process included mainly quartz, potassium-feldspar

    and plagioclase, minor mica/illite and traces of chlorite and amphibole.

    4.1.2 Laboratory reflectance measurements. The reflectance of samples from

    Darere (all err-41 and dr-2 samples in figure 10 and table 2) exhibited strong

    absorption features of kaolinite and ferric iron and minor signatures of alunite.

    Other reflectance curves exhibited similarities to spectra of potassium-feldspar and

    opal (Clark et al. 1993, figure 10(b)). Two samples from Illegedi (samples err-42 and

    3922 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    Table 2. Mineralogical compositions of selected samples of hydrothermally altered rocks andof alluvial plains, as detected by the X-ray diffraction method. Samples err-37, err-39 and err-51 consists of fine arkosic sand from fluvial plains. Sample err-84 is from an aeolian deposit.

    Sample No.Hydrothermally altered, Alid Calibration samples

    Mineral dr-2 Err-41 Err-41b Err-42 Err-43 Err-69b Err-37 Err-39 Err-51c Err-84

    Quartz xx xx xxx ? ? xx xx xx xxxK-feldspar xx xx xx xx x xx xx ?Plagioclase xx xx x xxx xx xx xx xKaolinite xxx xMica/Illite tr tr x x x x xChlorite ? tr tr tr xAlunite xxJarosite xxGypsum* tr xAmphibole tr tr tr trTridymite x xxCristobalite x ?Calcite x

    *Including anhydrite. xxx: dominant; xx: major; x: minor; tr: traces.

    Figure 10. (a) Laboratory reflectance spectra of samples from Darere, Alid. Alunitesignatures are strongly masked by kaolinite. The shoulder at ,1770 nm (sample dr-2) impliesits existence. (b) Comparison with orthoclase and opal from Clark et al. (1993). Sample names(e.g. err-41, dr-2) indicate separate samples from the same location. See table 2 formineralogical compositions of selected samples.

    Hyperion mapping in the Danakil Depression, Eritrea 3923

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    err-43 in figure 11 and table 2) were analysed, showing in the first sample strong

    features of jarosite and gypsum, and opal features in the second sample, both

    confirmed by the XRD analysis. Continuum removal (CR) of the reflectance curves

    from 1500 to 1600 nm revealed an absorption feature at 1558 nm in four samples of

    hydrothermally altered rocks from Alid (figure 12): err-41c, err-41b-c (from Darere),

    err-42a, and err-42b (from Illegedi). This feature is attributed to the existence of

    ammonium in silicate minerals (Krohn and Altaner 1987, Baugh et al. 1998, Bishop

    et al. 2002). Absorption at 2254 nm (possibly due to iron-hydroxyl vibration) was

    found to be the most diagnostic spectral feature of the pelitic schist. More

    characteristic features are the absorptions of the combination vibration modes of

    Figure 11. Laboratory reflectance spectra of samples from Illegedi, Alid. See figure 10 andtable 2 for information of sample names and composition. The small absorption at 430 nm isdiagnostic to ferric iron and jarosite (Hunt and Ashley 1979, Cloutis et al. 2006). Spectra ofsample err-42 are similar to opal (Clark et al. 1993) in shape (also features at ,1130 and2250 nm).

    3924 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    magnesium-hydroxyl (Mg-OH) at 2324 nm10 nm and of the aluminium hydroxyl

    (Al-OH) at ,2200 nm in clay minerals.

    4.2 Processing of the hyperspectral data

    4.2.1 Atmospheric corrections. The most successful radiative-based correction was

    obtained using the 1.5pb mode (ACORNTM 5.0, 2004), which also partially

    corrected the line array effect (figure 4). However, residual atmospheric absorptions

    of oxygen and carbon dioxide were still apparent. To fine tune the ACORN results,we applied the EL approach. The best end-member combination for the EL

    correction comprised the fresh basalt and an aeolian deposit (table 1), which is

    dominated by quartz (sample err-84 in table 2). Nevertheless, the existence of calcite

    in sample err-84 (table 2) may have introduced a minor false absorption feature at

    ,2300 nm throughout the entire image.

    4.2.2 Classification. Unsupervised classification: end-members for classification

    were chosen using a Pixel Purity Index (PPI, Boardman et al. 1995) procedure

    with 15 000 iterations with a threshold value of 1.5. A total of 18 411 pixels were

    chosen in this process. Seventeen regions of interest (ROI) were finally chosen, in

    sizes that range from 19 to 82 pixels (17 10073 800 m2). Two classification

    sessions were carried out: the first used the entire spectral range of the Hyperion

    (426.822395.5 nm) excluding the water vapour regions (at ,1400 and ,1900 nm)and the second used the SWIR range only (2062.552395.5 nm). Both classifications

    successfully recognized the hydrothermally altered rocks and schist within the Alid

    Figure 12. (a) Reflectance curves of ammonium-bearing minerals and of samples ofhydrothermally altered rocks which might contain ammonium. (b) The continuum removal(CR) of the absorption feature at ,1560 nm. The minerals are from Clark et al. (1993) andGrove et al. (1992). Rock samples from Alid are marked as err-##.

    Hyperion mapping in the Danakil Depression, Eritrea 3925

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    dome, and identified kaolinite-rich areas in the southern part of the scene, which has

    a spectral signature similar to that of the altered rocks of the Alid dome (figures 13

    and 14). The kaolinite-rich rocks were not visited in the field and are suspected to be

    hydrothermally altered.

    Supervised classification: This classification was applied to the CR image limited

    to the spectral range between 1507 nm (band 136) and 1598 nm (band 145). The goal

    of this session was to detect the slight absorption at 1558 nm, which is attributed to

    the ammonium-related absorption, as seen in the laboratory spectral reflectance of

    the hydrothermally altered samples (figure 12). Band rationing of the CR Hyperion

    bands 140 and 141 (1548 nm/1558 nm) was performed prior to the classification.

    These bands represent the wavelength location of the ammonium-related absorption

    which was discovered in the laboratory-measured reflectance curves, as appears in

    the Hyperion spectral resolution (figure 12). The ratioing emphasized hydrother-

    mally altered rocks within Alid as well as three different rock units: black slate,

    schist rocks and kaolinite-rich rocks. Four different shapes of CR absorption

    features were chosen as end-members for the classification (figure 15). The end-

    members included schist rocks, kaolinite-rich rocks (neither was reached in the field

    trip), black-slate outcrops and hydrothermally altered rocks in the Alid dome. The

    classification result of the entire scene shows that the Alid-type end-member

    appears both in the kaolinite-rich and in the black-slate units (figures 14 and 15);

    each of these units is characterized by a different reflectance signature. The fourth

    end-member is the schist rocks, which differs in shape and depth from the other end-

    members.

    The Alid-type end-member is spatially associated with hydrothermally altered

    rocks within the Alid area (Duffield et al. 1997, Gersman et al. 2006, figure 14).

    Furthermore, rocks in the eastern escarpment of the Alid graben are classified both

    as hydrothermally altered and as ammonium-bearing (figure 14).

    5. Classification accuracy assessment

    A quantitative assessment of the mapping results was not conducted due to two

    main reasons: (1) No accurate, detailed spatial and spectral surveys were performed

    in the field (see Crouvi 2002, Levin 2002). This is also the reason why a wide-used

    error matrix for accuracy assessment was not established. (2) The mineralogical

    results of the XRD analysis estimated the bulk mineralogy of the rock samples and

    were not quantitative. On the basis of these estimations, the mineralogical

    distribution maps could be ranked on a scale of absent to dominant only.

    A visual assessment of the classification accuracy of the entire scene can be made

    from the map, integrating the major at-surface geological/lithological elements and

    the ground observations. This map consists of ground targets that were identified by

    the PPI procedure, some of which were visited in the field. Ground targets that were

    visited in the field but were not recognized by the PPI procedure are also included.

    The map in figure 13 classifies the end-members generated by the PPI procedure.

    The main end-members comprise the Neoproterozoic basement, Tertiary and

    Quaternary volcanic rocks, fluvial and aeolian deposits and hydrothermally

    altered rocks. Mapping with these end-members was carried out using the SAM

    procedure.

    Schist in the southwest corner of the map is classified as alluvial fan on the map,

    probably because of highly weathered surface. Dolomite patches intercalated

    between schistose areas are well mapped. Yet, the northern border of the dolomite

    3926 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    unit merges with an alluvial fan, which is dominated by dolomitic debris. The black

    slate is carefully classified towards south, but is also found in the western slope of

    Alid which most probably is an erroneous classification. Patches within the black

    slate domain are shaded areas, mistakenly classified as basalt. The identification of a

    Figure 13. Integrated thematic map using the Hyperion scene. The map units were choseneither from the targets identified by the Pixel Purity Index (PPI) procedure, or according tothe laboratory-measured spectrum of samples collected during the field trip. The map isoverlaid on a base image that used an RGB combination of bands 292011 (true colours).

    Hyperion mapping in the Danakil Depression, Eritrea 3927

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    basaltic cover around Alid, however, is correct. The topography has a major effect

    on the apparent reflectance (Richter 1998, Levin 2002, Feng et al. 2003, Riano et al.

    2003), hence the Alid slopes, which represent the areas rough terrain, are poorly

    classified. Yet, a basic distinction between acidic volcanic rocks (a mixture of

    rhyolite, ignimbrite and obsidian) and basaltic flows was obtained, as well as a

    distinction between the aeolian and alluvial covers. The classification procedure of

    the Hyperion did recognize hydrothermal sites within the Alid dome (figure 14): the

    sites of Illegedi, Ghinda and Humbebet, described by Duffield et al. (1997), were

    identified, whereas Abakri and Asaela (figure 14), both strongly shaded, were

    missed. The site of Darere, which was visited in the field trip, was not spectrally

    recognized, probably due to relatively dense vegetation that was present at the time

    of the overpass.

    A group of pixels northeast of Alid were classified as hydrothermally altered

    rocks (figure 14). The classified pixels are concentrated in a group and are not

    aligned on a residual vertical stripe, nor limited to shaded areas. This pattern

    suggests that the classification of those pixels is probably correct.

    The classification procedure identified outcrops of schist within Alid (figures 13

    and 14). The distribution of the schist outcrops on the map does not necessarily

    represent their exact location in the field, since this area is subject to strong

    Figure 14. (a) Map of hydrothermally altered areas within the Alid dome and on the easternfault escarpment. The classification of the surrounding volcanic rocks is not included; (b)mapping of the ammonium signal in the entire scene. There are distinct legends for (a) and (b).

    3928 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    topographic and shading effects. Yet, the existence of kyanite schist at Illegedi is

    known from the field trip (Gersman et al. 2006) and was reported by Duffield et al.

    (1997). The identification of rhyolites along the western border of the map, from

    Alid towards northeast, is most probably erroneous. In addition, the SWIR-based

    classification of alluvial and of aeolian plains yielded poor results. This area has

    remains of the vertical stripes a strong residual noise, especially in the SWIR range

    of the Hyperion sensor, which could be the cause of this poor classification result.

    6. Mapping of the hydrothermally altered rocks of Alid

    The mineralogical assemblage detected around fumaroles and hot springs resembles

    the advanced argillic alteration type (Hunt and Ashley 1979, Heald et al. 1987). This

    type of alteration contains kaolinite or pyrophyllite and alunite, together with

    possible quartz, diaspore, opal and potassium-mica (Hunt and Ashley 1979). The

    existence of ammonium sulphate minerals in Alid (tschermigite NH4Al(SO4)212H2O, mascagnite (NH4)2(SO4), kokatite (NH4)2Ca(SO4)2 H2O) was reported

    by Beyth (1996) and Duffield et al. (1997). Tschermigite and alum (KAl(SO4)2) are

    salts that are formed where ammonium- and sulphate-bearing waters react with

    crustal rocks at sub-boiling temperatures (up to 66uC in the thermal pools of Alid;Duffield et al. 1997) and oxidizing atmospheric conditions. Since the primary

    dissolved anion within the acidic thermal pools in Alid is sulphate (up to 1767 ppm

    in Illegedi) and the main dissolved cations are ammonium, calcium and potassium

    (Duffield et al. 1997, Lowenstern et al. 1997), salts containing these constituents are

    likely to precipitate as the solutions evaporate (Duffield et al. 1997). These

    conditions may be favourable for the incorporation of ammonium ions into alunite

    (Altaner et al. 1988, Krohn et al. 1993), aluminous clay minerals (Krohn and

    Altaner 1987, Sucha et al. 1998, Bishop et al. 2002, Holloway and Dahlgren 2002) or

    potassium-feldspar (Krohn and Altaner 1987, Krohn et al. 1993, Baugh et al. 1998,

    Holloway and Dahlgren 2002), depending on the redox conditions of the fluids

    (Krohn and Altaner 1987). Incorporation of ammonium into minerals occurs when

    an ammonium ions replace alkali cations (usually potassium, which is close in

    radius) in the crystal structure. Potassic and sericitic alteration, and advanced

    argillic alteration (with alunite) will create potassium substitution sites. Conversely,

    kaolinitization, where potassium-bearing minerals are removed (which is the case in

    the kaolinite-rich rocks of Alid), and silification will tend to prevent high

    ammonium values (Ridgway et al. 1990, Baugh et al., 1998).

    Mapping of ammonium in previous studies utilized different absorption features

    of the nitrogen-hydrogen (NH) bond representing the vibrational combination

    n1 + n3 where n1 and n3 are fundamental vibration modes of the ammonium ion(NH4

    + ) at 3040 cm21 and 3140 cm21, respectively (Krohn and Altaner 1987). Baugh

    et al. (1998) used the absorption features at 2120 nm to map buddingtonite, whereas

    Bishop et al. (2002) pointed out 1558 nm as a significant ammonium feature. The

    fact that the important ammonium-related feature at 2120 nm is absent from our

    laboratory spectra should make us very cautious in calling upon ammonium to

    explain the 1558 nm feature. Nevertheless, we could not find another mechanism to

    explain the observed absorption (Krohn et al. 1993; Baugh et al. 1998, Bishop et al.

    2002). Trace amounts of ammonium in minerals might not be detected by the XRD

    method: the XRD pattern of buddingtonite (NH4AlSi3O8), for example, can be

    overlapped by the potassium-feldspar pattern (Krohn et al. 1993, Baugh et al. 1998)

    and that may be the case in sample err-41 (major potassium-feldspar, table 2).

    Hyperion mapping in the Danakil Depression, Eritrea 3929

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    Sample err-42 (quartz dominant, table 2), contains trace amounts of mica/illite.

    Drits et al. (1997) pointed out that XRD analyses of ammonium-bearing illite did

    not always indicate the presence of structural ammonium in the lattice. In this case

    reflectance spectroscopy seems to be more sensitive to the existence of ammonium

    than XRD (Krohn et al. 1993, Drits et al. 1997, Baugh et al. 1998). Krohn et al.

    (1993) claimed that minimum concentration for spectral detection of ammonium

    from whole-rock samples is approximately 0.06% of (NH4)2O by weight, depending

    upon the samples lithology. It is suggested, therefore, that the absorption at

    1558 nm is due to the existence of ammonium-bearing phyllosilicates, which were

    not detected by the XRD analysis.

    The detection of the feature at 1558 nm in the laboratory spectra has led us to seek

    an equivalent feature in the Hyperion dataset. As the SNR around 1558 nm is

    relatively high and no overlapping with other chromophors exists, this band was

    further studied and emphasized four major lithologies (figure 14(a), figure 15):

    hydrothermally altered areas in Alid (Alid type), kaolinite-rich areas, black slate/

    phyllite, and possibly pelitic schist, all maybe containing ammonium-bearing

    minerals (Krohn et al. 1993, Baugh et al. 1998, Holloway and Dahlgren 2002). The

    spatial association between the hydrothermally altered areas within Alid and the

    areas where the feature attributed to ammonium was detected (figure 14(b)), and the

    identification of the 1558 nm feature in a probably hydrothermally altered zone

    (kaolinite-rich type, figure 14(b)) support the possibility that the spectral absorption

    feature at 1558 nm may represent ammonium at the surface. The 1558 nm signal

    could have been distinguished from areas which did not display this feature

    (figure 16). It was not aligned on residual vertical stripes, nor restricted to shaded

    areas. It also did not originate from the reflectance curves used for the EL correction

    and it is not a carbon dioxide or an oxygen residual absorption. The laboratory

    spectral measurements and studies made by Beyth (1996) and Duffield et al. (1997)

    on the same area led us to suggest that the Hyperion was able to detect the 1558 nm

    Figure 15. (a) Continuum removal (CR) of the apparent reflectance image, in the range of15001600 nm (shaded areas). A laboratory-measured feature from a hydrothermally alteredrock (dashed line) is added for comparison. (b) Examples of apparent reflectance curves ofend-members from (a).

    3930 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    spectral feature around this specific Alid area. This finding together with other

    mapping units shows that despite the Hyperion limitations, it is useful for mapping

    mineralogical in remote areas. Furthermore, it stresses the need for hyperspectral

    data and technology in general.

    7. Summary and conclusions

    A careful processing of the Hyperion data followed well-known techniques and

    mapped the area well according to previous knowledge. A ground data reconnais-

    sance was performed along with careful laboratory analyses of the samples. The

    unsupervised classification approach separated between different types of rock

    groups, whereas the supervised classification pinpointed the unique spectral feature at

    around 1558 nm, arguably ammonium-related. This feature has a strong spatial

    association with kaolinite-rich, hydrothermally altered rocks. Absorption features in

    the SWIR-1 are significant because they are positioned in a high SNR spectral range

    that enables ammonium-bearing minerals to be detected quickly and non-

    destructively by a variety of airborne and ground-based sensors (Krohn et al. 1988,

    1993) and now, presumably, also by the spaceborne Hyperion sensor. The ability of

    the Hyperion to detect ammonium spectral signatures has not been previously

    reported. While the mapping of the ammonium-related feature is arguable, the

    suggested mineralogical mapping results demonstrate that the extraction of

    information from hyperspectral data is yet to be completed, and points out the

    significance of the smaller, indicative absorption features for this purpose. The need

    and use of hyperspectral technology for mapping remote areas and the potential

    success of such mapping are well demonstrated here.

    Figure 16. Continuum removal (CR) of the reflectance spectra used for the empirical linecorrection (err-84, err-86) and of a feature at 1558 nm.

    Hyperion mapping in the Danakil Depression, Eritrea 3931

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    Acknowledgements

    Without the support of Mr H. Goder, Israel ambassador to Eritrea, this projectcould not have been accomplished. We thank A. Mushkin, S. Peeri and Y. Golani

    for helping with the ASTER and Hyperion images, T. Medhin and staff geologists

    of the Department of Mines, Asmara, for sharing with us their profound knowledge

    during the field trip, A. Sandler for the XRD analysis and A. Kultunov for helping

    with the processing of the Hyperion data. Careful and thoughtful comments by Ed

    Cloutis and an anonymous reviewer helped to improve this manuscript and are

    greatly appreciated. This work was funded by the US Agency for International

    Development, Bureau for Economic Growth, Agriculture and Trade, ProjectNo. C23-001 and Award No. Ta-MOU-03-C23-001.

    ReferencesABBATE, E., WOLDEHAIMANOT, B., BRUNI, P., FALORNI, P., PAPINI, M., SAGRI, M.,

    GIRMAY, S. and TECLE, T.M., 2004, Geology of the Homo-bearing Pleistocene

    Dandiero Basin (Buya region, Eritrean Danakil Depression). Rivista Italiana di

    Paleontologia e Stratigrafia, 10, pp. 534.

    ABRAMS, M., HOOK, S. and RAMACHANDRAN, B., 2002, ASTER User Handbook. Available

    online at: http://asterweb.jpl.nasa.gov/documents.asp (accessed 21 January 2008).

    ACORNTM 5.0, 2004, Tutorial, ImSpec LLC, Advanced Imaging and Spectroscopy, ImSpec

    LLC.

    ALTANER, S.P., FITZPATRICK, J.J., KROHN, M.D., BETHKE, P.M., HAYBA, D.O., GOSS, J.A.

    and BROWN, Z.A., 1988, Ammonium in alunites. American Mineraogist, 73, pp.

    145152.

    ASNER, G.P. and HEIDEBRECHT, K.B., 2003, Imaging spectroscopy for desertification studies:

    comparing AVIRIS and EO-1 Hyperion in Argentina drylands. IEEE Transactions on

    Geoscience and Remote Sensing, 41, pp. 12831296.

    BARRY, P., 2001, EO-1/Hyperion science data users guide, level 1_B. TRW Space, Defense &

    Information Systems, No. HYP.TO.01.077, Rev. Public release L1_B.

    BAUGH, W.M., KRUSE, F.A. and ATKINSON, W.W., JR., 1998, Quantitative geochemical

    mapping of ammonium minerals in the Southern Cedar Mountains, Nevada, using

    the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS). Remote Sensing of

    Environment, 65, pp. 292308.

    BECK, R., 2003, EO-1 User guide, V.2.3. Available online at http://eo1.usgs.gov and http://

    eo1.usgs.gov/documents.php (accessed 21 January 2008). University of Cincinnati,

    Ohio.

    BEN-DOR, E., KRUSE, F.A., LEFKOFF, A.B. and BANIN, A., 1994, Comparison of three

    calibration techniques for utilization of GER 63-channel aircraft scanner data of

    Makhtesh Ramon, Negev, Israel. Photogrammetric Engineering and Remote Sensing,

    60, pp. 13391354.

    BEYTH, M., 1996, Preliminary assessment of the Alid geothermal field, Eritrea. Israel

    Geological Survey Current Research, 10, pp. 124128.

    BEYTH, M., AVIGAD, D., WETZEL, H.-U., MATHEWS, A. and BERHE, S.M., 2003, Crustal

    exhumation and indications for Snowball Earth in the East African orogen: north

    Ethiopia and east Eritrea. Precambrian Research, 123, pp. 187201.

    BISHOP, J.L., BANIN, A., MANCINELLI, R.L. and KLOVSTAD, M.R., 2002, Detection of soluble

    and fixed NH4+ in clay minerals by DTA and IR reflectance spectroscopy: a potential

    tool for planetary surface exploration. Planetary and Space Science, 50, pp. 1119.

    BOARDMAN, J.W., 1993, Automated spectral unmixing of AVIRIS data using convex

    geometry concepts. In Summaries, Fourth JPL Airborne Geoscience Workshop, JPL

    Publication 93-26, vol. 1 (Pasadena, CA: Jet Propulsion Laboratory), pp. 1114.

    BOARDMAN, J.W., 1998, Leveraging the high dimensionality of AVIRIS data for improved

    sub-pixel target unmixing and rejection of false positives: mixture tuned matched

    3932 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    filtering. In Summaries of the 7th Annual JPL Airborne Geoscience Workshop,

    Pasadena, CA (Pasadena, CA: Jet Propulsion Laboratory), p. 51.

    BOARDMAN, J.W., KRUSE, F.A. and GREEN, R.O., 1995, Mapping target signatures via partial

    unmixing of AVIRIS data. In Summaries, Fifth JPL Airborne Earth Science

    Workshop, JPL Publication 95-1, 1 (Pasadena, CA: Jet Propulsion Laboratory), pp.

    2326.

    CLARK, R.N. and ROUSH, T.L., 1984, Reflectance spectroscopy: quantitative analysis

    techniques for remote sensing applications. Journal of Geophysical Research, 89, pp.

    63296340.

    CLARK, R.N., SWAYZE, G.A., GALLAGHER, A.J., KING, T.V.V. and CALVIN, W.M., 1993, The

    US Geological Survey, Digital Spectral Library: Version 1: 0.2 to 3.0 microns. US

    Geological Survey Open File Report 93-592.

    CLARK, R.N., SWAYZE, G.A., LIVO, K.E., KOKALY, R.F., KING, T.V.V., DALTON, J.B.,

    VANCE, J.S., ROCKWELL, B.W., HOEFEN, T. and MCDOUGAL, R.R., 2002, Surface

    reflectance calibration of terrestrial imaging spectroscopy data: a tutorial using

    AVIRIS. In Summaries of the 11th Annual JPL Airborne Geoscience Workshop,

    Pasadena, CA (Pasadena, CA: Jet Propulsion Laboratory), pp. 3858.

    CLOUTIS, E.A., HAWTHORNE, F.C., MERTZMAN, S.A., KRENN, K., CRAIG, M.A.,

    MARCINO, D., METHOT, M., STRONG, J., MUSTARD, J.F., BLANEY, D.L., BELL, J.F.

    III. and VILAS, F., 2006, Detection and discrimination of sulfate minerals using

    reflectance spectroscopy. Icarus, 184, pp. 121157.

    CROUVI, O., 2002, Geomorphic mapping using hyperspectral remote sensing: the Wadi

    Raham alluvial fan as a case study. MSc Thesis, Institute of Earth Sciences, The

    Hebrew University, Jerusalem.

    CUDAHY, T.J., HEWSON, R.D., HUNTINGTON, J.F., QUIGLEY, M.A. and BARRY, P.S., 2001,

    The performance of the satellite-borne Hyperion Hyperspectral VNIR-SWIR imaging

    system for mineral mapping at Mount Fitton, South Australia. In Proceedings of the

    IEEE 2001 International Conference on Geoscience and Remote Sensing, Sydney,

    IEEE, New Jersey, pp. 913.

    CUDAHY, T.J., RODGER, A.R., BARRY, P.S., MASON, P., QUIGLEY, M.A., FOLKMAN, M.A.

    and PEARLMAN, J., 2002, Assessment of the stability of the Hyperion SWIR module

    for hyperspectral mineral mapping using multi-date images from Mount Fitton,

    Australia. In Proceedings of the IEEE 2002 International Conference on Geoscience and

    Remote Sensing, Toronto, IEEE, New Jersey, pp. 2428.

    DATT, B., MCVICAR, T.R., VAN NIEL, T.G., JUPP, D.L.B. and PEARLMAN, J.S., 2003,

    Preprocessing EO-1 Hyperion hyperspectral data to support the application of

    agricultural indexes. IEEE Transactions on Geoscience and Remote Sensing, 41, pp.

    12461259.

    DRITS, V.A., LINDGREEN, H. and SALYN, A.L., 1997, Determination of the content and

    distribution of fixed ammonium in illite-smectite by x-ray diffraction: application to

    North Sea illitesmectite. American Mineralogist, 82, pp. 7987.

    DRURY, S.A., KELLEY, S.P., BERHE, S.M., COLLIER, R.E.L. and ABRAHA, M., 1994,

    Structures related to Red Sea evolution in northern Eritrea. Tectonics, 13, pp.

    13711380.

    DUFFIELD, W.A., LEAKE, W., BULLEN, T.D., KAHSAI, G., CLYNNE, M.A., KIDANE, W.,

    FOURNIER, R.O., THEODORES, T., JANIK, C.J., LANPHERE, M.A., LOWENSTERN, J. and

    SMITH, J.G., 1997, Geothermal potential of the Alid Volcanic Center, Danakil

    Depression, Eritrea. US Geological Survey, USA, Ministry of Energy, Mines and

    Water Resources, Eritrea.

    ENVI, 2003, ENVI Users Guide, Research Systems, Inc., Boulder, CO, USA.

    FENG, J., RIVARD, B. and SANCHEZ-AZOFEIFA, A., 2003, The topographic normalization of

    hyperspectral data: implications for the selection of spectral end members and

    lithologic mapping. Remote Sensing of Environment, 85, pp. 221231.

    Hyperion mapping in the Danakil Depression, Eritrea 3933

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    GERSMAN, R., BEN-DOR, E., AVIGAD, D., BEYTH, M., ABRAHA, M. and KIBREAB, A., 2006,

    Hyperspectral remote sensing as a tool for geological exploration, Northern

    Danakil Depression, Eritrea. The Geological Survey of Israel, Report GSI/08/06,

    Jerusalem.

    GHEBREAB, W., 1998, Tectonics of the Red Sea region reassessed. Earth Science Reviews, 45,

    pp. 144.

    GHEBREAB, W., 1999a, Pan-African and Red Sea tectonics of Eastern Eritrea. Faculty of

    Science and Technology, Uppsala University, Uppsala.

    GHEBREAB, W., 1999b, Tectono-metamorphic history of Neoproterozoic rocks in eastern

    Eritrea. Precambrian Research, 98, pp. 83105.

    GHEBREAB, W., TALLBOT, C.J. and PAGE, L., 2005, Time constraints on exhumation of the

    East African Orogen from field observations and 40Ar/39Ar cooling ages of low-angle

    mylonites in Eritrea, NE Africa. Precambrian Research, 139, pp. 2041.

    GOODENOUGH, D.G., DYK, A., NIEMANN, K.O., PEARLMAN, J.S., CHEN,Hao, HAN, T.,

    MURDOCH, M. and WEST, C., 2003, Processing Hyperion and ALI for forest

    classification. IEEE Transactions on Geoscience and Remote Sensing, 41, pp.

    13211331.

    GREEN, A.A., BERMAN, M., SWITZER, P. and CRAIG, M.D., 1988, A transformation for

    ordering multispectral data in terms of image quality with implications for noise

    removal. IEEE Transactions on Geoscience and Remote Sensing, 26, pp. 6574.

    GREEN, R.O., PAVRI, B.E. and CHRIEN, T.G., 2003, On-orbit radiometric and spectral

    calibration characteristics of EO-1 Hyperion derived with an underflight of AVIRIS

    and in situ measurements at Salar de Arizaro, Argentina. IEEE Transactions on

    Geoscience and Remote Sensing, 41, pp. 11941203.

    GROVE, C.I., HOOK, S.J. and PAYLOR, E.D., 1992, Laboratory reflectance spectra for 160

    minerals 0.42.5 micrometers. JPL Publication 92-2, Jet Propulsion Laboratory,

    Pasadena, CA.

    HEALD, P., FOLEY, N. and HAYBA, D., 1987, Comparative anatomy of volcanic-hosted

    epithermal deposits: acid-sulfate and adularia-sericite types. Economic Geology, 82,

    pp. 126.

    HOLLOWAY, J.M. and DAHLGREN, R.A., 2002, Nitrogen in rock: occurrence and

    biogeochemical implications. Global Biogeochemical Cycles, 16, pp. 11181135.

    HUBBARD, B.E., CROWLEY, J.K. and ZIMBELMAN, D.R., 2003, Comparative alteration

    mineral mapping using visible to shortwave infrared (0.42.4 mm) Hyperion, ALI, and

    ASTER imagery. IEEE Transactions on Geoscience and Remote Sensing, 41, pp.

    14011410.

    HUNT, G.R. and ASHLEY, R.P., 1979, Spectra of altered rocks in the visible and near infrared.

    Economic Geology, 74, pp. 16131629.

    JENSEN, J.R., 2005, Introductory Digital Image Processing. A Remote Sensing Perspective, 3rd

    edn (Upper Saddle River, NJ: Prentice Hall).

    KROHN, M.D. and ALTANER, S.P., 1987, Near-infrared detection of ammonium minerals.

    Geophysics, 52, pp. 924930.

    KROHN, M.D., ALTANER, S.P. and HAYBA, D.O., 1988, Distribution of ammonium minerals

    at Hg/Au-bearing hot springs deposits. In Proceedings of the Bulk-Mineable Precious

    Metal Deposits of the Western United States Symposium, Geological Society of

    Nevada, R.W. Schafer, J.J. Cooper, and P.G. Vikre (Eds), Geological Society of

    Nevaza, Reno, pp. 661680.

    KROHN, M.D., KENDALL, C., EVANS, J.R. and FRIES, T.L., 1993, Relations of ammonium

    minerals at several hydrothermal systems in the western U.S. Journal of Volcanology

    and Geothermal Research, 56, pp. 401413.

    KRUSE, F.A., 1988, Use of airborne imaging spectrometer data to map minerals associated

    with hydrothermally altered rocks in the Northern Grapevine Mountains, Nevada,

    and California. Remote Sensing of Environment, 24, pp. 3151.

    3934 R. Gersman et al.

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    KRUSE, F.A., 1998, Advances in hyperspectral remote sensing for geologic mapping and

    exploration. In Proceedings 9th Australasian Remote Sensing Conference, Sydney,

    Australia, http://www.hgimaging.com/FAK_Pubs.htm

    KRUSE, F.A., 2003, Preliminary results hyperspectral mapping of coral reef systems using

    EO-1 Hyperion, Buck Island and U.S. Virgin Islands. In Proceedings of the 12th JPL

    Airborne Geoscience Workshop, JPL Publication 04-6 (Pasadena, CA: Jet Propulsion

    Laboratory), pp. 157173.

    KRUSE, F.A., BOARDMAN, J.W. and HUNTIGTON, J.F., 2003, Comparison of airborne

    hyperspectral data and EO-1 Hyperion for mineral mapping. IEEE Transactions on

    Geoscience and Remote Sensing, 41, pp. 13881400.

    KRUSE, F.A., LEFKOFF, A.B., BOARDMAN, J.B., HEIDEBRECHT, K.B., SHAPIRO, A.T.,

    BARLOON, P.J. and GOETZ, A.F.H., 1993, The spectral image processing system

    (SIPS) interactive visualization and analysis of imaging spectrometer data. Remote

    Sensing of Environment, 44, pp. 145163.

    KRUSE, F.A., RICHARDSON, L.L. and AMBROSIA, V.G., 1997, Techniques developed for

    geologic analysis of hyperspectral data applied to near-shore hyperspectral ocean

    data. In Proceedings, ERIM 4th International Conference, Remote Sensing for Marine

    and Coastal Environments, vol. I (Ann Arbor, MI: Environmental Research Institute

    of Michigan), pp. I-233I-246.

    LEVIN, N., 2002, Quantitative mapping of the soil rubification process on the

    coastal sand dunes of Israel using an Airborne CASI hyperspectral sensor. The

    sand dunes f Ashdod as a case study. MA thesis, Department of Geography

    and Human Environment, Tel-Aviv University, Tel-Aviv, Israel [in

    Hebrew].

    LOWENSTERN, J.B., JANIK, C.J., TESFAI, T. and FOURNIER, R.O., 1997, Geochemical study of

    the Alid hydrothermal system, Danakil Depression, Eritrea. Stanford Geothermal

    Program Report SGP-TR-150, Stanford University, pp. 3744.

    LOWENSTERN, J.B., CLYNNE, M.A. and BULLEN, T.D., 1998, Comagmatic A-type granophyre

    and rhyolite from the Alid volcanic center, Eritrea, northeast Africa. Journal of

    Petrology, 38, pp. 17071721.

    NEVILLE, R.A., SUN, L. and STAENZ, K., 2003, Detection of spectral line curvature in imaging

    spectrometer data. In Algorithms and Technologies for Multispectral, Hyperspectral,

    and Ultraspectral Imagery IX, S.S. Shen, and P.E. Lewis (Eds). Proceedings of the

    SPIE, 5093, SPIE, Bellingham, WA, pp. 144154.

    PEARLMAN, J.S., BARRY, P.S., SEGAL, C.C., SHEPANSKI, J., BEISO, D. and CARMAN, S.L.,

    2003, Hyperion, a space-based imaging spectrometer. IEEE Transactions on

    Geoscience and Remote Sensing, 41, pp. 11601173.

    RIANO, D., CHUVIECO, E., SALAS, J. and AGUADO, I., 2003, Assessment of different

    topographic corrections in Landsat-TM data for mapping vegetation types. IEEE

    Transactions on Geoscience and Remote Sensing, 41, pp. 10561061.

    RICHTER, R., 1998, Correction of satellite imagery over mountainous terrain. Applied Optics,

    37, pp. 40044015.

    RIDGWAY, J., APPLETON, J.D. and LEVINSON, A.A., 1990, Ammonium geochemistry in

    mineral exploration a comparison of results from the American cordilleras and the

    southwest Pacific. Applied Geochemistry, 5, pp. 475489.

    ROWAN, L.C. and MARS, J.C., 2003, Lithologic mapping in the mountain pass, California

    area using Advanced Spaceborne Thermal Emission and Reflection Radiometer

    (ASTER) data. Remote Sensing of Environment, 84, pp. 350366.

    SOURIOT, T. and BRUN, J.P., 1992, Faulting and block rotation in the Afar triangle, East

    Africa: the Danakil crank-arm model. Geology, 20, pp. 911914.

    SUCHA V., ELSASS, F., EBERL, D.D., KUCHTA, L., MADEJOVA, J., GATES, W.P. and

    KOMADEL, P., 1998, Hydrothermal synthesis of ammonium illite. American

    Mineralogist, 83, pp. 5867.

    Hyperion mapping in the Danakil Depression, Eritrea 3935

  • Dow

    nloa

    ded

    By:

    [Ger

    sman

    , Ron

    en] A

    t: 21

    :27

    16 J

    une

    2008

    THOME, K.J., BIGGAR, S.F. and WISNIEWSKI, W., 2003, Cross comparison of EO-1 sensors

    and other Earth resources sensors to Landsat-7 ETM+ using Railroad Valley Playa.IEEE Transactions on Geoscience and Remote Sensing, 41, pp. 11801188.

    UNGAR, S.G., PEARLMAN, J.S., MENDENHALL, J.A. and REUTER, D., 2003, Overview of the

    Earth Observing One (EO-1) mission. IEEE Transactions on Geoscience and Remote

    Sensing, 41, pp. 11491159.

    WARREN, B.E., 1990, X-Ray Diffraction (New York: Dover).

    WEIDNER, V.R. and HSIA, J.J., 1981, Reflection properties of pressed polytetrafluoro-

    ethylene powder. Journal of the Optical Society of America, 71, pp. 856861.

    3936 Hyperion mapping in the Danakil Depression, Eritrea


Recommended