ASTROBIOLOGICAL CONSIDERATIONS FOR THE SELECTION OF THE GEOLOGICAL FILTERS ON THE EXOMARS PANCAM INSTRUMENT Claire R Cousins1,2., Andrew D Griffiths1,3., Ian A Crawford1,2., Bryan J Prosser4., Michael C Storrie-Lombardi5., Lottie E Davis1,6., Matthew Gunn7., Andrew J Coates1,3., Adrian P Jones1,6., John M Ward8. 1 Centre for Planetary Sciences at UCL/Birkbeck, Gower Street, London, WC1E 6BT. 2 Department of Earth and Planetary Sciences, Birkbeck College, London, WC1E 7HX 3 Mullard Space Science Laboratory, University College London, Holmbury St. Mary, Dorking, Surrey, RH5 6NT, UK. 4 Department of Computer Science, Queen Mary College, University of London, Mile End Road, London, E1 4NS. 5 Kinohi Institute, Pasadena, California, USA. 6 Department of Earth Sciences, University College London, Gower Street, London, WC1E 6BT. 7 Institute of Mathematics & Physics, Aberystwyth University, Penglais, Aberystwyth, UK, SY23 3BZ. 8 Department of Structural and Molecular Biology, University College London, Gower Street, London, WC1E 6BT. Corresponding author: Claire Cousins Dept. Earth Sciences, University College London, Gower Street, London, WC1E 6BT, Telephone: 02076792577, E-mail: c.cousins@ucl.ac.uk 1 Running title: Selecting geological filters for the ExoMars PanCam ABSTRACT The Panoramic Camera (PanCam) instrument will provide VIS-NIR multispectral imaging of the ExoMars rover’s surroundings to identify regions of interest within the nearby terrain. This multispectral capability is dependant upon the 12 pre-selected ‘geological’ filters that are integrated into two wide angle cameras. First devised by the Imager for Mars Pathfinder team to detect iron oxides, this baseline filter set has remained largely unchanged for subsequent missions (MER, Beagle2, Phoenix), despite the advancing knowledge of the mineralogical diversity on Mars. Therefore, the geological filters for the ExoMars PanCam will be re-designed to accommodate the astrobiology focus of ExoMars, where hydrated mineral terrains (evidence of past liquid water) will be priority targets. Here, we conduct an initial investigation into new filter wavelengths for the ExoMars PanCam, and present results from testing these on Martian analogue rocks. Two new filter sets were devised: one with filters spaced every 50nm (“F1-12”), and one utilising a novel filter selection method based upon hydrated mineral reflectance spectra (“F2-12”). These new filter sets, and the Beagle2 filter set (currently the baseline for the ExoMars PanCam), were tested on their ability to identify hydrated minerals and biosignatures present in Martian analogue rocks. The filter sets were found, with varying degrees of ability, to detect the spectral features of minerals jarosite, opaline silica, alunite, nontronite, and siderite present in these rock samples. Fossilised biomat structures and small (<2mm) mineralogical heterogeneities present in silica sinters however remained undetected with all the filter sets. Both the new filter sets 2 outperformed the B2 filters, with F2-12 detecting the most spectral features produced by hydrated minerals, and providing the best discrimination between samples. Future work involving more extensive testing on Martian analogue samples exhibiting a wider range of mineralogies would be the next step in carefully evaluating the new filter sets. 3 1. INTRODUCTION 1.1 The ExoMars PanCam instrument The ExoMars rover is the currently proposed Aurora flagship mission. It aims, primarily, to find evidence for past or present life on Mars via a drill that can reach depths of 2m into the subsurface. The rover will be equipped with a multi-purpose Panoramic Camera (PanCam) instrument, which will provide wide angle multispectral stereoscopic panoramic images of the rover’s surroundings (for a detailed description see Griffiths et al. 2006). As currently implemented, multispectral imaging works by utilising a set of filters, set at specific wavelengths and bandwidths, to record reflectance images from the field of view. One filter yields one reflectance value, and the spread of filters produces a spectral profile at each image pixel. The more filters available, the more detailed the spectra will be. Two of the main scientific objectives of the PanCam that will utilise this multispectral imaging capability are firstly the acquisition of geological information proximal to the rover (e.g. Smith et al. 1997a; Bell et al. 2004; Morris et al. 2004; Farrand et al. 2007), and secondly identification of the most suitable locations for further astrobiological investigation. It is noted that the PanCam is not designed to be a lifedetection instrument in itself, but provides the initial step in identifying suitable lithologies for further multi-level analysis with the aim to detect evidence of biological activity (Vago 2005). This site identification will be heavily dependant upon the surrounding geological terrain and mineralogy identified by the PanCam, and as a result, the PanCam needs to be carefully designed to achieve this goal. As the PanCam multispectral imaging capability is, in part, dependant upon a suitable set of ‘geological’ filters, the pre-selected wavelengths of these individual filters needs to be optimised to fit 4 the objectives of the ExoMars mission. The geological filter wavelengths for the PanCam are required to fall in the 440 – 1000nm range (the detection limits of the CCD detector), with 12 filter spaces available. A notional set of filter wavelengths has already been allocated for PanCam, however this filter set is inherited from Beagle 2, which in turn was adopted from Mars Pathfinder. Filters for the Imager for Mars Pathfinder (IMP) device were selected with two main objectives in mind. Firstly, to identify ferric oxides and oxyhydroxides, and secondly to determine silicate mineralogy present, particularly pyroxenes (Smith et al. 1997b). This filter set was largely adopted for the MER PanCam, again focusing on the detection of iron-bearing silicates, iron oxides and oxyhydroxides, and also to provide a direct comparison to the IMP results (Bell et al. 2003). ExoMars has a distinct astrobiological focus, and may encounter an extensive range of hydrated mineral-bearing lithologies. As a consequence, the notional filter set for the PanCam needs to be revised. The ability to detect regions or outcrops of interest remotely is a fundamental aspect of planetary rover exploration. However, whilst multispectral imaging at optical to nearinfrared wavelengths (400 – 1000nm) can reveal a lot about mineralogy and chemistry, the majority of distinguishing spectral features occur in the infrared. As a result, the combination of reflectance spectra with other instrumental analysis can enhance the ability of the PanCam in identifying mineralogy and lithology. The MicrOmega instrument can provide additional detailed spectral information in the 0.9 – 2.6µm region (Leroi et al. 2009), helping to confirm many of the mineralogical features identified with PanCam. Additionally, the Raman instrument will be able to provide mineral 5 identification of target samples (Wiens et al. 2005; Bazalgette et al. 2007; Sharma et al. 2007; Sharma et al. 2009), and will provide vital data to complement the PanCam’s multispectral capabilities. These instruments are currently envisaged to work on the micro-scale on drill samples, and as such are not remote surveying tools, potentially leaving PanCam as the sole means of remote target selection. However, in future missions a remote Raman system could provide an excellent way to support PanCam data. Here, the reflectance spectra from publically available spectral libraries were used to formulate prospective new geology filter sets for the ExoMars PanCam. These libraries included: the United States Geological Survey spectral library (USGS; Clark et al. 2007; 1993), the JPL spectral library (Baldridge et al. 2009), and RELAB (spectra acquired by Bruce Fegley, Carle Pieters, Edward Cloutis, Jack Mustard, Janice Bishop, Michelle Goryniuk, and Phoebe Hauff, with the NASA RELAB facility at Brown University) library. Specifically, spectra of hydrated minerals already identified on Mars were used. These alternative filter sets, plus the Beagle2 (B2) filter set (which is currently the baseline for PanCam) were tested on their ability to successfully detect hydrated minerals present in untreated Martian analogue rocks from Iceland. The benefit of using raw samples instead of homogenised, size-specific particulate fractions, has been highlighted by Harloff & Arnold (2001). Previous to that, Yon & Pieters (1988) noted that the most appropriate analogue to outcropping rock on a planetary surface is not a powdered sample but a rough bulk sample surface. As such, natural geological samples are used for this work, to best simulate the nature of geological outcrops that may be encountered 6 during the ExoMars mission. Multi-instrument and spectral analysis of analogue or astrobiologically-relevant material has previously shown the value of ground-truthing studies (e.g. Bishop et al. 2004a; Bishop et al. 2004b; Pullan et al. 2008). Contextual information for the samples was also gathered on mineralogy, chemistry, and textural/morphological features using Raman spectroscopy, X-ray Diffraction (XRD), Scanning Electron Microscopy (SEM), and Energy Dispersive X-Ray Spectrometer (EDS) analysis. 1.2 Hydrated Minerals Hydrated mineral terrains on Mars are of particular interest to the ExoMars mission as they are essential in establishing the history of liquid water on the planet, which in turn is directly related to the search for past or present Martian life. Minerals with OH or H2O as part of their chemical structure require aqueous conditions to form, but do not necessarily need liquid water to remain stable in their current environment after formation (Bishop 2005). It is this knowledge that has driven the need to explore terrains rich in hydrated minerals, with the aim of identifying evidence for past habitable environments on Mars (Murchie et al. 2009). The results of the hyper-spectrometers OMEGA and CRISM currently orbiting Mars have revealed a large diversity in surface mineralogy and lithology (e.g. Bibring et al. 2005; Mustard et al. 2008; Poulet et al. 2007; Murchie et al. 2009; Poulet et al. 2009). Hydrated minerals on Mars have now been identified in distinct hydrated mineral terrains. Oldest are the phyllosilicates, which are found in early Noachian terrains (Poulet et al. 2005; Bibring et al. 2006; Mustard et al. 2008). These are predominantly Fe, Mg and Al – rich phyllosilicates (such as nontronite and 7 montmorillonite) and have been seen to occur as alternating lithologies (Loizeau et al. 2007). These phyllosilicates reveal a complex aqueous history of early Mars that may be indicative of long-term aqueous alteration of basaltic/igneous material in this area (Poulet et al. 2005; Bishop et al. 2008). Sulphates, notably gypsum, jarosite and kieserite, amongst others, have been identified in layered terrains on Mars (Klingelhofer et al. 2004; Bibring et al. 2005; Gendrin et al. 2005). These have been seen to occur in Late Noachian and early Hesperian terrains, and it has been suggested they indicate a change from neutral pH conditions to more acidic environments (Bibring et al. 2006). Sulphates are commonly formed through volcanic activity (such as alteration of volcanic rocks by acidic fumaroles) or evaporitic processes (Martinez-Frias & Amaral 2006). More recently discovered are deposits of opaline silica on Mars, both detected by CRISM and by the MER Spirit at Gusev Crater (Milliken et al. 2008; Squyres et al. 2008; Rice et al. 2010). Opaline silica commonly forms in hot spring systems, where the eruption of silica supersaturated hot spring fluids at the surface gradually produces siliceous sinters over time. Additionally, opaline silica can form via hydrothermal weathering of basaltic rock, stripping the mafic minerals away leaving behind a silica-rich crust (Kraft et al. 2003). Recently Grasby et al. (2009) have described silica deposits from cold springs believed to be unrelated to hydrothermal systems. Silica sinters are known to be excellent bio-preservers on Earth, and it has been noted previously that the discovery of hydrated silica on Mars could have important implications for the detection of Martian biosignatures that may be similar to those commonly found in hot spring sinters on Earth (Farmer & Des Marais 1999; Goryniuk et al. 2004). Likewise, the recent discovery of 8 possible carbonate deposits (Ehlmann et al. 2008; Palomba et al. 2009; Morris et al. 2010) suggest past environments on Mars were more habitable than previously thought. The aim of this work was to carry out the preliminary exploration of potential new geological filter wavelengths for the ExoMars PanCam, and present initial results from testing these new filters on Martian analogue rocks. Reflectance spectra from hydrated minerals, including gypsum, jarosite, alunite, calcite, montmorillonite, nontronite, and opaline silica, were used to formulate an alternative filter sets for the ExoMars PanCam, with the aim to detect these minerals within geological outcrops and deposits within the range of the rover. 2. MATERIALS AND METHODS 2.1 Hydrated mineral reflectance spectra Reflectance spectra between the region 440 – 1000nm of hydrated minerals from the USGS spectral library splib06a (Clark et al. 2007) and splib04 (Clark et al. 1993), JPL spectral library (Baldridge et al. 2009), and RELAB spectral library (Brown University, see above), were used to identify particular wavelengths that would be optimal for detecting the diagnostic spectral features of such minerals at the Martian surface. Minerals were chosen based on the identification of hydrated mineral groups on Mars, and are shown in Table 1. For filter selection (described below), the mineral spectra were required to have evenly spaced spectral points, and as such the published spectral data were re-sampled to produce data points every 10nm. It is noted that the mineral samples used to generate the library spectra are not always compositionally 100% pure, and often 9 have other minor mineral fractions present. As such, we merely take these powdered mineral reflectance spectra as an ideal spectrally pure ‘end member’. Another important factor is the variation between different sample spectra of the same mineral within the database. For example, the spectra for different samples of jarosite vary. Therefore, multiple sample spectra were utilised in the formulation of the new filter set to eliminate any bias towards one arbitrarily chosen sample. All mineral spectra used in the filter selection are given in Table 1, together with their associated mineral sample attributes, including impurities, and grainsize. 2.2 Martian analogue samples Martian analogue samples (Figure 1) were collected from Iceland. They include subglacial basaltic lavas, hydrothermally altered lavas and hot spring precipitates. The island of Iceland is a predominantly volcanic country, formed by the surface expression of the Mid Atlantic Ridge and an underlying mantle plume - the Icelandic Hot spot (Sigvaldason et al. 1974; Korenaga 2004). There are numerous examples of past (mostly Pleistocene) subglacial volcanic activity, and later Holocene lava flows, both of which commonly have undergone hydrothermal interaction. Geologically, Iceland bears many similarities to proposed Martian volcanism. It is dominated by basaltic volcanism that is a result of fissure eruptions and mantle plume activity. Many volcanic features identified on Mars are also seen in Iceland, notably widespread basaltic flows (Keszthelyi et al. 2004) and shield volcanoes. The terrestrial analogues of some features on Mars, such as pseudocraters (rootless cones), are found predominantly in Iceland (Fagents & Thordarson 2007), and several comparisons have been made between Martian and 10 Icelandic glaciovolcanism (Chapman & Smellie 2007). Therefore, Iceland is considered here as an ideal analogue for volcanic environments that have existed on Mars in the past. Additionally, as none of the Icelandic samples contained carbonate, a sample of siderite (from UCL Earth Sciences Geology Collections) was also used for testing the new filter wavelengths. 2.3 Multispectral imaging and analytical methods Martian analogue samples were imaged multispectrally at the Mullard Space Science Laboratory, UK. Samples were illuminated with a Solex solar lamp at an average distance of 60 cm, although this varied depending on the size of the sample. Still-capture imaging was carried out at a distance of 1 m (the minimum distance at which PanCam will be used) from the sample by a Foculus FO432SB camera (1.4Mpixles, 8- bits/pixel greyscale, 15° field of view lens, exposure time 1 ms to 65s). As with the ExoMars PanCam, this camera has a 1024 x 1024 pixel CCD which responds to wavelengths between 400 – 1000nm. The images taken at this distance typically had a spatial resolution of between 100 – 200 µm/pixel. The camera was interfaced with one of two CRI Varispec liquid crystal tuneable filters – one covering the visible (wavelength range of 400 – 730nm; bandpass of 20nm) and the other covering the near infrared (wavelength range of 650 – 1100nm; bandpass of 10nm). Images were taken with these filters in 10nm increments. Images were processed with ImageJ software (http://rsbweb.nih.gov/ij/). Relative reflectance spectra were calculated by dividing the brightness values averaged across a number of pixels for the sample/target of interest (T) by that measured from a white calibration target (C) in the background of each image. The size of the regions from 11 which spectra were extracted varied from 10 x 10 pixels to 200 x 200 pixels, depending upon the size of the feature in question. Mineralogical identification within the Martian analogue samples was achieved using Raman spectroscopy and X-Ray Diffraction (XRD) to confirm spot mineralogy (within the multispectral target area), and bulk mineralogy, respectively. Raman spectra were gathered using a Renishaw InVia Raman spectrometer coupled with a Leica microscope at University College London, using a 785nm laser, through either a 20x or 50x microscope lens. XRD analysis was conducted at Aberystwyth University using a Bruker D8 Advance XRD with a Vantec Super Speed detector. Sub-millimetre structural and elemental data were collected using a combined Scanning Electron Microscope (SEM) with Back Scatter Electron (BSE) detector, and Energy Dispersive X-ray Spectrometer (EDS) from samples to provide additional context for the interpretation of the observed reflectance spectra from the multispectral imaging data. For the SEM study, thin sections were carbon coated and analysed using a Jeol Scanning Electron Microscope (JSM6480LV) at University College London. For associated EDS analysis, an accelerating voltage of 15kV was used. 2.4 Selection of filter wavelengths for two possible new filter sets Two new alternative filter sets were devised. Reflectance spectra (from spectral databases, see section 2.1 above) of hydrated minerals nontronite, jarosite, montmorillonite, calcite, gypsum, opaline silica, and alunite were used to devise one of 12 these two new filter sets, based on the necessity to successfully identify hydrated minerals on Mars. The new filter sets were chosen as follows: Filter set F1-12: This filter set had 12 evenly spaced filters, spaced every 50nm (with the exception of the first two filters, which have a 60nm spacing) and was not biased towards any pre-determined set of minerals. Filter set F2-12: This filter set was generated statistically based upon hydrated mineral spectra. Twelve optimal wavelengths that most accurately reproduced a specific mineral spectrum were calculated for all the 70 selected hydrated minerals collectively (all normalised to the same starting value). This filter selection used a brute force approach to search through all the possible wavelength selections between 440 and 1000nm, using 10nm increments. Two of these 12 wavelengths were always fixed at 440 and 1000nm. The suitability of a wavelength selection was calculated firstly by interpolating the points between the 12 randomly chosen wavelengths to obtain a full set of data (i.e. a complete spectrum) for each individual mineral spectrum. Then, we calculated the absolute difference in reflectance (residual) between the actual spectrum of the mineral (Ra), and the spectrum estimated (Re) using the current wavelength selection at each respective 10nm spaced point (λ). The summation of these absolute distances is then used as the ‘error score’ for that particular wavelength selection and mineral (σm): m 13 1000 R ( ) R ( ) 440 a e where λ increments in steps of 10nm. The lower the error score, the better the wavelength selection, producing the optimal set of 12 wavelength points for each mineral. To find the best wavelength selection for the whole mineral set, the error scores for each mineral were summed into an overall error score (σ) for each combination of wavelengths assessed: M m m 1 This generated filter set ‘F2-12’. These two new filter sets, and the original Beagle2 (B2) filter wavelengths, were tested on Martian analogue samples to evaluate their suitability in the detection of hydrated minerals on Mars. 2.5 Testing filter sets on Martian analogue samples Multispectral data gathered for the Martian analogue samples were re-sampled to match that of the filter wavelengths in the proposed new filter sets, producing PanCam-style 12point spectra. In addition, the data were averaged within the range covered by the bandpass for each filter, of which the B2 filter bandpasses had already been previously assigned (Griffiths et al. 2006). These were used in the re-sampling of rock sample data averaged across the full-width-at-half-maximum (FWHM) values. For the new filter sets, estimated bandpasses were utilised, based on those assigned for the B2 filters. These are shown in Table 2. 14 The Martian analogue samples contained sulphate, phyllosilicate, and opaline silica minerals, and specific mineral species were confirmed with Raman spectroscopy and XRD (sample details in Table 4 and Figure 4). A sample of the Fe-carbonate siderite was also used, from which confirmatory Raman data was also acquired (Figure 4). Six Icelandic geological samples (Figure 1; plus one siderite sample) were used, and the spectra for each respective filter set for these samples are shown in Figure 5. Where relevant, spectra for different coloured regions (and therefore potentially different mineralogy) on the rock are shown, highlighting the spectral variation across a naturally heterogeneous geological sample. The filter sets were assessed firstly on whether or not they were able to capture spectral features that could potentially lead to the identification of a particular mineral species within the rocks, and secondly on their ability to distinguish between the different rock samples. Spectral parameters, such as those used to identify spectral variability within MER PanCam multispectral images (e.g. Farrand et al. 2006; 2008), were used, and are listed in Table 3. Different filter sets will have different filter wavelengths covering a particular spectral feature, such as the 440 – 700nm absorption edge, or a 900nm absorption band. Where there is not a filter centred at the specific band in question, the nearest filter is used, and so is representative of how spectral parameter data will appear with the different filter sets. Finally, multivariate analysis was employed to demonstrate the ability of the filter sets to identify and distinguish between clusters of similar mineral targets. 15 3. RESULTS 3.1 The new geological filter sets The new filter set F2-12 differs principally to the B2 filters in that it lacks a concentration of filters towards the NIR end of the spectrum, with a broader range of filters within the visible. The spectral region between 440-1000nm is dominated by Fe excitational and charge transfer bands (Bishop 2005), and therefore is particularly sensitive to ironbearing mineralogy (Farrand et al. 2008). As a result, distinctive spectral features are particularly evident in Fe-bearing hydrated minerals such as jarosite and nontronite, as well as silicates pyroxene and olivine and a range of iron oxides such as haematite and goethite. Ferric absorption bands exist at 480nm, 650nm and 950nm, whilst a ferrous absorption band also in the 950nm region can often obscure that produced by ferric iron (Stewart et al. 2006). Whilst nontronite and jarosite share these similar Fe-absorption bands, nontronite has a distinctive kink around 650nm that is not present in the jarosite spectrum, and it is features such as this that will be crucial in distinguishing PanCam spectral data. The filter centre wavelengths and their estimated bandpasses at FWHM are shown in Figure 2, together with the position of ferric and ferrous iron, and water/hydroxyl bands that are common to the hydrated minerals used in this study (also shown). It can be seen that the B2 filters miss a Fe3+ absorption band at ~500 and 650nm, and the jarosite and nontronite reflectance bands at 710nm and 580nm respectively. Overall, five of the B2 filters fall directly (their centre wavelength, 4 filters) or partially (coverage only by their bandpass, 1 filter) within these spectral bands. Similarly, six of the F1-12 filters cover the 16 spectral bands (five with their centre wavelength). In comparison, filter set F2-12 has eight filters falling within the spectral absorption/emission bands (six by their centre wavelength). As a simple example, Figure 3 shows two hydrated mineral spectra (nontronite and jarosite) focused on a particularly variable region, that demonstrates the effect different filter positions have on the observed spectrum. It can be seen that the B2 filter set does not fully recover key diagnostic spectral points for the mineral nontronite, where the Fe3+ absorption feature at 650nm falls between B2 filters 600nm and 670nm. Likewise, the B2 filters do not capture the reflectance peak at 710nm in jarosite, whilst filter set F1-12 partially misses it. Filter set F2-12 however closely follows the spectral features of both nontronite and jarosite within this wavelength region. 3.2 Identification of sulphates in Martian analogue samples Sample NAL is a subaerial basaltic lava that is covered in extensive sulphur-rich alteration products due to acid-fumarole weathering, producing distinctive red and yellow colour regions on the rock surface (Figure 1). EDS and BSE data show there to be different mineralogical deposits within lava vesicles, ranging from Si-rich, to Fe-Al-rich sulphate compositions (Figure 4, Table 4). Additionally, Raman spectroscopy shows the surface deposits to include natroalunite, natrojarosite, and haematite (Figure 4). The spectra of the different regions on this rock are consistent with the iron sulphate jarosite (KFe3(SO4)2(OH)6), and the aluminium sulphate alunite (KAl3(SO4)2(OH)6), for ‘Yellow’ and ‘Red’ regions respectively (Figure 5b). The ‘Yellow’ region produces a spectrum that has a steep ferric absorption edge between 440 – 700nm, followed by a broad absorption centred around 950nm, which is characteristic of Fe3+. The ‘Red’ region has spectral 17 features synonymous with those often seen for alunite, including a steep gradient between 440-700nm, and particularly an absorption feature at 950nm – 1000mn, which can be attributed to OH stretching vibrations that exist further into the NIR, the first band existing at 1010nm (Hunt et al. 1971). Pure alunite is typically white (Hunt et al. 1971), but in this case it exists as red deposits on the surface of the lava. This red colouration is due to the partial replacement of Al3+ with Fe3+ (Hunt et al. 1971), which is represented by the broad absorption feature between 800 and 900nm. Although jarosite is also a hydrated sulphate, it doesn’t exhibit any OH bands until further into the infra-red spectrum (Clark 1990). All these spectral features for both the ‘Yellow’ and ‘Red’ regions were best captured with the filter set F2-12, whilst both the B2 and F1-12 filter sets miss the peak at 710nm in the ‘Yellow’ spectrum (Figure 5b). Sample NBO, like sample NAL, is an acid-weathered basaltic lava, with orange coloured deposits filling the vesicles that are surrounded by unaltered basalt. The spectrum of these ‘Orange’ deposits is similar to the ‘Yellow’ mineral deposit in sample NAL, exhibiting a main peak at 720nm, and absorption features at 480 and 950nm (Figure 5a). As with the NAL deposits, these features suggest the presence of jarosite, which is entirely consistent with the geological setting of this lava. This is confirmed with Raman analysis (Figure 4). Unlike the deposits in sample NAL, all filter sets are able to correctly reproduce this spectrum, although the B2 and F1-12 filter sets only partially capture the 710nm peak, which lies either between the 670 and 750nm filters (B2), or 650 and 700nm filters (F112). 18 3.3 Identification of phyllosilicates in Martian analogue samples Hyaloclastite sample HH is rich in palagonite, which exists as alteration rinds around basaltic glass clasts, as seen under BSE (Figure 4). The smectite composition of hyaloclastite is represented by the elemental composition of the palagonite matrix surrounding the basaltic glass fragments (Table 4). XRD data revealed the sample to largely consist of amorphous, non-crystalline material, typical of palagonite (Stroncik & Schmincke 2002). Despite the lack of crystalline material detected with XRD, the PanCam spectral features for this rock suggest a nontronite component, with characteristic ferric or ferrous iron absorption features at 470, 680, and 950nm. The absorption features observed at ~460nm and ~950nm can be attributed to the presence of Fe3+, whilst the shifting of the absorption from 650nm (more typical for nontronite) to 680nm is indicative of a change from Fe3+ to Fe2+ (Stewart et al. 2006). Of the three filter sets, this spectrum is best reproduced by sets B2 and F1-12, in terms of the number of spectral features covered (Table 5). The new filter set F2-12 misses absorption features at 680nm and 800nm. However, whilst the B2 filters only miss one spectral feature (the kink at 680nm), this absorption is a key diagnostic feature of the nontronite spectrum (Figure 5d). 3.4 Identification of opal and carbonate in Martian analogue samples Silica-rich hot spring deposits are more ambiguous regarding the multispectral detection of mineral species present. Sample KH is a hyaloclastite with an opaline silica crust (1 – 4mm thick) on the surface, as confirmed by Raman spectroscopy and EDS (Table 4, Figure 4). XRD analysis shows this crust to be largely amorphous with minor amounts of 19 calcite. Additionally, this surface crust is inhabited by chasmolithic lichen communities. The spectrum of this white crust is consistent with that of opal-a, exhibiting a smooth arcing spectrum, although the hydration feature at 950nm is not present here (Figure 5f). The additional presence of the lichens has the effect of creating a shallow, broad absorption feature at 670nm, indicative of Chlorophyll a. As would be expected, this 670nm feature is prominent in the ‘Lichen’ spectrum, along with a steep H2O absorption between 900 – 1000nm. All the filter sets recovered these spectral features, from both the ‘White’ and ‘Lichen’ regions (Table 5f). Samples GY1 and GY2 are both hot spring silica sinters, again dominated by opaline silica. GY1 is characterised by several different coloured layered regions, ranging from black, red, yellow, and white (Figure 1b, Figure 4f), with these components ranging from 1 – 10mm in size. The spectrum of the ‘White’ region has a generally flat morphology, with a small absorption at 950nm, comparable to the spectral features of opal-CT, where the 950nm absorption feature can be attributed to OH-. This 950nm absorption is captured by all the filter sets (Figure 5c). The spectra of the other coloured regions of sample GY1 however are less indicative of its true mineralogy. Raman analysis of the ‘Red’ and ‘Yellow’ sinter components identifies these regions as haematite and sulphur (Figure 4) respectively, yet this is not revealed by the multispectral imaging. These mineral components are 1 – 2mm in size, and are therefore well within the limits of the test camera resolution (100 – 200 µm/pixel). Figure 6 shows the 12-point spectra for these two coloured regions, using filter set F2-12. It can be seen that the features are more consistent with those for the sulphate alunite, as seen in sample NAL (Figure 5b). These 20 two spectra exhibit differences in features that suggest perhaps a minor influence of their true mineralogy, such as the flatter profile of ‘GY1 Yellow’ between 600 – 830nm (sulphur has a flat profile between 500 – 1000nm), and the absorption between 440 560nm which is a feature typically seen for haematite. However, these features are by no means diagnostic, and highlight potential problems of remotely identifying small mineralogical targets within a heterogeneous rock sample. Like GY1, sample GY2 also formed as part of a silica sinter, but it exhibits a different spectral profile to that of GY1. GY2 is a silicified biomat with a dominantly Si composition (Table 4) and opal-a mineralogy (Figure 4), yet this is not represented within the spectral features (Figure 5c). As with the small sulphur and haematite mineral deposits in GY1, these spectral features are suggestive of alunite, despite the silica-rich composition of the rock. This alunite component can be seen to exist as a fine particulate coating on the surface of the sinter deposit (discernable from hand specimen), therefore preventing the silica mineralogy from influencing the observed spectrum. Likewise, the biological origin of this deposit appears to have no impact on the observed spectra, even though morphological microfossils can be easily identified as a significant structural component (Figure 7). As with sample KH, all filter sets captured the key spectral features of the GY2 spectrum (Table 5). Finally, iron-rich carbonate siderite (FeCO3) shows clear differences between the three filter sets. In particular, the B2 filters miss a significant peak at ~720nm (Figure 5e), a feature that is captured with both F1-12 and F2-12 filter sets. Siderite, along with magnesite, is one of the carbonates believed to exist on Mars (Morris et al. 2010; Bridges 21 et al. 2001), and indeed is one of the phases found in some Martian meteorites (Romanek et al. 1994; Bridges & Grady 2000). Considering the importance of carbonate minerals to astrobiology (e.g. Murchie et al. 2009), it is imperative PanCam is able to correctly identify them. 3.5 Spectral parameter representation of Martian analogue samples In addition to PanCam spectral data being used to identify putative mineral assemblages, this information also needs to successfully distinguish between different targets. For this, spectral parameter plots were used to group Martian analogue spectra into spectral groups based on particular spectral features, such as the steepness of an absorption slope, or the depth of an absorption band (listed in Table 3). These plots are often used to define spectral variability within Martian rocks, and therefore potential lithological variability (Farrand et al. 2006; 2008). Likewise, such parameters can be used to identify the distribution of a particular mineralogical feature across a Martian scene (Rice et al. 2010). Therefore, the selection of particular filter wavelengths may affect the ability of PanCam to correctly distinguish different lithologies/mineral species. Figure 8 plots ferric iron content (steepness of the 440 – 700nm slope) against water content (920 – 1000nm slope). The mineral spectral data used for the selection of filter set F2-12 (Table 1) are also plotted for comparison, and it can be clearly seen that a decrease in the ferric iron absorption slope corresponds with an increasing water absorption slope (Figure 8a). This trend is also seen in the sulphate-containing rock samples, with jarosite-rich samples NAL Yellow and NBO following this trend down to 22 alunite-containing samples NAL Red and GY2 (Figure 8b). There is little difference between samples KH White, GY1 White, and HH, which cluster in a similar region to opaline silica (Figure 8d), consistent with their mineralogy except for HH, the presence of which is most likely explained by its amorphous composition. Iron-rich carbonate siderite plots with a much steeper 440 – 700nm ferric absorption slope, distinct from the rest of the samples within this plot. Lastly, these four spectral parameters (Table 3) were used together to distinguish between these samples using discriminant analysis (DA), and to cluster the rock sample targets using unsupervised hierarchical cluster analysis (HCA). The DA (Figure 9a) shows the ability of the three filter sets at distinguishing between the samples. Filter set F2-12 provides by far the best discrimination between the groups, whilst the B2 filter set performs worst with all three groups clustering close together. Likewise, in both B2 and F1-12 filter sets, Group 1 and Group 2 overlap, demonstrating poor discrimination between the rock targets within these groups. Conversely F2-12 filters clearly discriminate well between all three groups, demonstrating the benefit of optimising filter wavelengths to putative mineral targets. Overall, the filter testing on untreated, heterogeneous rock samples shows some variation between the filter sets, in terms of both ability to detect potentially diagnostic spectral features, and also to clearly discriminate between rock targets using spectral parameter values. Of the three filter sets tested, both F1-12 and F2-12 performed better than the baseline Beagle 2 filters, with filter set F2-12 detecting the most spectral features (Table 23 5), and also most clearly discriminating between rock samples (Figure 9). For comparison, this filter set is given alongside the ‘Geology’ filter wavelengths for other past and present Mars rovers/landers, including those for MER and currently proposed for the Mars Science Laboratory (MSL; Table 6). In particular, the MSL MastCam filter set has significantly fewer filters within the visible (<750nm). Where this wavelength range is covered by 8 filters by filter set F2-12, it is covered by only 4 filters in the MastCam. More extensive testing on a Martian analogue samples exhibiting a significantly wider range of mineralogies would be the next step in identifying the value in specific wavelength selections, or whether an entirely evenly spaced filter set would suffice. Indeed, such work not only has relevance to the ExoMars mission, but also to other astrobiologically focused missions, such as MSL. 4. DISCUSSION & CONCLUSIONS This study was conducted with two aims: firstly to produce and test alternative geological filter sets to replace the Beagle2 filters currently assigned to ExoMars, and secondly to present initial results from the testing of these possible filter sets on Martian analogue rock samples. It is apparent that a concentration of filters towards the NIR end of the spectrum, as with the B2 filter set, is not optimal for the ability of the PanCam to detect hydrated minerals as a component of heterogeneous rock samples. The first new filter set also explored - ‘F1-12’ - has filters spaced at regular intervals every 50nm. The benefit of this filter set is that the even spread of spectral points is not biased towards any particular group of minerals, and so theoretically makes the PanCam equally suited to detecting any given mineral or lithology it encounters. However, whilst much is still unknown as 24 regards to the lithology and mineralogy of the Martian surface, ExoMars will be focused particularly on regions where hydrated minerals are likely to predominate. The aim of positioning these points so that they favour certain mineral groups is to enhance the detection of these specific minerals in keeping with the mission objectives. The effectiveness of this can be seen in Figure 3. The filter set that was generated statistically (F2-12) based on a set of seven different hydrated minerals, proved most effective at identifying hydrated minerals in the Martian analogue rocks, suggesting this particular method of filter wavelength assignment is successful at selecting a suitable set of geological filters. As such, there is the potential to utilise this method using a much wider set of mineral spectral data before the PanCam filter wavelengths are finalised. Our knowledge of the mineralogical diversity of the Martian surface is increasing, and future PanCam geological filter selection and testing will need to incorporate a much wider range of minerals and analogue samples. Another factor is the mineralogy of the selected ExoMars landing site (still to be finalised). The availability of CRISM and OMEGA data allows the characterisation of the bulk mineralogy of a potential landing site, and this could be used to heavily influence the ExoMars PanCam filter wavelengths. For example, if such orbital spectrometer data showed there to be extensive and rover accessible clay-bearing geological units, the PanCam filters may be more useful if they were optimised to detect phyllosilicate minerals only. Such focused target selection would benefit from potentially more diagnostic PanCam spectra, but may significantly reduce the ability of PanCam to undertake more opportunistic science and any subsequent unexpected discoveries. 25 The need to expand on spectral ground-truthing data sets was recently highlighted by Poulet et al. (2009). As with natural outcrops that are likely to be encountered on Mars, these rock samples display large and small scale heterogeneities, only some which were clearly distinguished with PanCam-style 12-point spectra, as in sample NBO where the spectral data were clearly able to detect the ‘Orange’ jarosite component within the surrounding basalt lava (Figure 5a). Likewise, the lichens in sample KH were spectrally distinct from the surrounding silica crust. However, detecting heterogeneity was a problem for the silica sinter samples – deposits that would have a particular astrobiological relevance. Opaline silica deposits have the potential to provide information on past life on Mars, especially if such deposits formed through hot spring sinter development - a process commonly seen in volcanically active regions on Earth. There is a possibility for the discovery of morphological, mineralogical and chemical biosignatures within these precipitates (e.g. Goryniuk et al. 2004; Schulze-Makuch et al. 2007; Preston et al. 2008), but the reflectance spectra of the silica sinter samples from Geysir provide little evidence of their hot-spring origin, and therefore astrobiological potential. A similar example is provided by the small mineralogical variations within sample GY1, the spectra of which did not identify the presence of haematite or sulphur detected by Raman spectroscopy. Haematite has been previously documented to be an effective inorganic barrier to UV radiation (Clark 1998), and has also been shown to be present at the Martian surface by the MER Opportunity (Klingelhofer et al. 2004), and as such is one of the many minerals that could be expected to be found around the ExoMars landing site. 26 In conclusion, much work still needs to be done with regards to firstly choosing and testing an optimised geological filter set for the ExoMars PanCam instrument, and secondly providing essential ground-truthing using Martian analogue samples. Quantifying the limits of PanCam multispectral imaging in the remote detection of mineral targets is especially important, as is the effect of dust coverage on an already heterogeneous rock sample. Although the multispectral data acquired by PanCam is fairly crude in comparison to other planetary instrumentation, it plays a crucial role in the selection of specific targets for more detailed investigation, and as such should be developed to its full potential. ACKNOWLEDGEMENTS This work was jointly funded by an EPSRC studentship award obtained through a UCL strategic, interdisciplinary research initiative, the UK Science and Technology Facilities Council (STFC), who are the lead funding agency for the development of the ExoMars Panoramic Camera, and the Leverhulme Trust. Claire Cousins would also like to thank James Davy for assistance with SEM/EDS analysis, and Dr. Steve Firth for assistance with Raman analysis. Finally, all authors would like to acknowledge the reviewers for their time and suggestions which led to considerable improvements to the manuscript. REFERENCES 27 Baldridge, A. M., Hook, S. J., Grove, C. I. and G. Rivera (2009). The ASTER Spectral Library Version 2.0. In press Remote Sensing of Environment. Remote Sensing of Environment 113: 711-715. Bazalgette Courrèges-Lacoste, G., Ahlers, B., Rull Pérez, F. (2007) Combined Raman spectrometer laser-induced breakdown spectrometer for the next ESA mission to Mars. Spectrochim. Acta, Part A: Molecular and Biomolecular Spectroscopy 68: 1023–1028. Bell, J. F., Squyres, S. W., Arvidson, R. E., Arneson, H. M., Bass, D., Blaney, D., Cabrol, N., Calvin, W., Farmer, J., Farrand, W. H., Goetz, W., Golombek, M., Grant, J. A., Greeley, R., Guinness, E., Hayes, A. G., Hubbard, M. Y. H., Herkenhoff, K. E., Johnson, M. J., Johnson, J. R., Joseph, J., Kinch, K. M., Lemmon, M. T., Li, R., Madsen, M. B., Maki, J. N., Malin, M., McCartney, E., McLennan, S., McSween Jr., H. Y., Ming, D. W., Moersch, J. E., Morris, R. V., Noe Dobrea, E. Z., Parker, T. J., Proton, J., Rice Jr., J. W., Seelos, F., Soderblom, J., Soderblom, L. A., Sohl-Dickstein, J. N., Sullivan, R. J., Wolff, M. J., Wang, A. (2004) Pancam Multispectral Imaging Results from the Spirit Rover at Gusev Crater. Science 305: 800-806. Bell, J. F., Squyres, S. W., Herkenhoff, K. E., Maki, J. N., Arneson, H. M., Brown, D., Collins, S. A., Dingizian, A., Elliot, S. T., Hagerott, E. C., Hayes, A. G., Johnson, M. J., Johnson, J. R., Joseph, J., Kinch, K., Lemmon, M. T., Morris, R. V., Scherr, L., Schwochert, M., Shepard, M. K., Smith, G. H., Sohl-Dickstein, J. N., Sullivan, R. J., 28 Sullivan, W. T., Wadsworth, M. (2003). Mars Exploration Rover Athena Panoramic Camera (Pancam) investigation. Journal of Geophysical Research 108 (E12): 8063. Bibring, J.-P., Y. Langevin, J. F. Mustard, F. Poulet, R. Arvidson, A. Gendrin, B. Gondet, N. Mangold, P. Pinet, and F. Forget. (2006) Global mineralogical and aqueous Mars history derived from the OMEGA/Mars Express data. Science 312: 400-404. Bibring, J-P, Langevin, Y., Gendrin, A., Gondet, B., Poulet, F., Berthe, M., Soufflot, A., Arvidson, R., Mangold, N., Mustard, J., Drossart, P. and the OMEGA team. (2005) Mars Surface Diversity as Revealed by the OMEGA/Mars Express Observations. Science 307: 1576-1581. Bishop, J. L., Dobrea, E. Z. D., McKeown, N., K., Parente, M., Ehlmann, B. L., Michalski, J. R., Milliken, R. E., Poulet, D., Swayze, G. A., Mustard, J. F., Murchie, S. L., Bibring, J.-P. (2008) Phyllosilicate Diversity and Past Aqueous Activity Revealed at Mawrth Vallis, Mars. Science 321: 830 – 833. Bishop, J. L. (2005) Hydrated Minerals on Mars. In Water on Mars and Life, Tetsuya Tokano (ed.), Adv. Astrobiol. Biogeophys., pp. 65-96. Bishop, J. L., Murad, E., Lane, M., Mancinelli, R. L. (2004a) Multiple techniques for mineral identification on Mars: a study of hydrothermal rocks as potential analogues for astrobiology sites on Mars. Icarus 169: 311–323. 29 Bishop, J. L, Dyar, M. D., Lane, M. D. and Banfield, J. F. (2004b). Spectral identification of hydrated sulfates on Mars and comparison with acidic environments on Earth. International Journal of Astrobiology 3: 275–285. Bishop, J. L., H. Fröschl, and R. L. Mancinelli. (1998) Alteration processes in volcanic soils and identification of exobiologically important weathering products on Mars using remote sensing. Journal of Geophysical Research 103: 31457–31476. Bridges, J. C., Catling, D. C., Saxton, J. M., Swindle, T. D., Lyon, I. C., Grady, M. M. (2001) Alteration Assemblages in Martian Meteorites: Implications for Near-Surface Processes. Space Science Reviews 96: 365-392. Bridges, J. C., Grady, M. M. (2000) Evaporite mineral assemblages in the nakhlite (martian) meteorites. Earth and Planetary Science Letters 176: 267-279. Chapman, M. G. and Smellie, J. L. (2007) Mars interior layered deposits and terrestrial sub-ice volcanoes compared: observations and interpretations of similar geomorphic characteristics. In The Geology of Mars: Evidence from Earth-Based Analog, edited by M. G. Chapman. Published by Cambridge University Press, Cambridge, UK, 2007, pp.178. 30 Clark, R.N., Swayze, G.A., Wise, R., Livo, E., Hoefen, T., Kokaly, R., Sutley, S.J. (2007) USGS digital spectral library splib06a: U.S. Geological Survey, Digital Data Series 231. Clark, B. C. (1998) Surviving the limits to life at the surface of Mars. Journal of Geophysical Research Planets 103: 445 – 455. Clark, R.N., Swayze, G. A., Gallagher, A. G., King, T.V.V. and Calvin, W.M. (1993) The U. S. Geological Survey, Digital Spectral Library: Version 1: 0.2 to 3.0 microns, U.S. Geological Survey Open File Report, 93-592. Ehlmann, B. L., Mustard, J. F., Murchie, S. L., Poulet, F., Bishop, J. L., Brown, A. J., Calvin, W. M., Clark, R. N., Des Marais, D. J., Milliken, R. E., Roach, L. H., Roush, T. L., Swayze, G. A., Wray, J. J. (2008) Orbital Identification of Carbonate-Bearing Rocks on Mars. Science 322: 1928 – 1832. Evans, D. L., Adams, J. B. (1979) Comparison of Viking Lander multispectral images and laboratory reflectance spectra of terrestrial samples (1979). In: Lunar and Planetary Science Conference, 10th, Houston, Tex., March 19-23, 1979, Proceedings. Volume 2. (A80-23617 08-91) New York, Pergamon Press, Inc. pp. 1829-1834. Fagents, S. A. and Thordarson, T. (2007) Rootless volcanic cones in Iceland and on Mars. In The Geology of Mars: Evidence from Earth-Based Analog edited by M. G. Chapman Published by Cambridge University Press, Cambridge, UK, 2007, pp. 151. 31 Farmer, J. D. and Des Marais, D. J. (1999) Exploring for a record of ancient Martian life. Journal of Geophysical Research 104: 26977-26995. Farrand, W. H., Bell, J. F., Johnson, J. R., Arvidson, R. E., Crumpler, L. S., Hurowitz, J. A., Schroder, C. (2008) Rock spectral classes observed by the Spirit Rover’s Pancam on the Gusev Crater Plains and in the Columbia Hills. Journal of Geophysical Research 113: E12S38. Farrand, W. H., Bell, J. F., Johnson, J. R., Jolliff, B. L., Knoll, A. H., McLennan, S. M., Squyres, S. W., Calvin, W. M., Grotzinger, J. P., Morris, R. V., Soderblom, J., Thompson, S. D., Watters, W. A., Yen, A. S. (2007) Visible and near-infrared multispectral analysis of rocks at Meridiani Planum, Mars, by the Mars Exploration Rover Opportunity. Journal of Geophysical Research 112: E06S02. Farrand, W. H., J. F. Bell III, J. R. Johnson, S. W. Squyres, J. Soderblom, and D. W. Ming (2006) Spectral variability among rocks in visible and near-infrared multispectral Pancam data collected at Gusev crater: Examinations using spectral mixture analysis and related techniques, Journal of Geophysical Research 111: E02S15. Gendrin, A., Mangold, N., Bibring, J-P., Langevin, Y., Gondet, B., Poulet, F., Bonello, G., Quantin, C., Mustard, J., Arvidson, R., LeMouelic, S. (2005) Sulfates in Martian Layered Terrains: The OMEGA/Mars Express View. Science 307: 1587-1591. 32 Goryniuk, M. C., Rivard, B. A. and Jones, B. (2004) The reflectance spectra of opal-A (0.5–25 mm) from the Taupo Volcanic Zone: Spectra that may identify hydrothermal systems on planetary surfaces. Geophysical Research Letters, 31,L24701. Grasby, S. E., Bezys, R., Beauchamp, B. (2009) Silica Chimneys Formed by LowTemperature Brine Spring Discharge. Astrobiology 9: 931–941. Griffiths, A. D., Coates, A. J., Jaumann, R., Michaelis, H., Paar, G., Barnes, D., Josset, JL. and the PanCam Team. (2006) Context for the ESA ExoMars rover: the Panoramic Camera (PanCam) instrument. International Journal of Astrobiology 5: 269–275. Harloff, J., Arnold, G. (2001) Near-infrared reflectance spectroscopy of bulk analog materials for planetary crust. Planetary and Space Science 49: 191 – 211. Hunt, G. R., Salisbury, J. W. and Lenhoff, C. J. (1971). Visible and Near-infrared spectra of minerals and rocks: IV. Sulphides and sulphates. Modern Geology 3: 1-14. Keszthelyi, L., Thordarson, T., McEwen, A., Haack, H., Guilbaud, M., Self, S. and Rossi, M. J. (2004) Icelandic analogs to Martian flood lavas. Geochemistry Geophysics Geosystems 5: 1-32. 33 Klingelhofer, G., Morris, R. V., Bernhardt, B., Schroder, C., Rodionov, D. S., de Souza Jr., A., Yen, A., Gellert, R., Evlanov, E. N., Zubkov, B., Foh, J., Bonnes, U., Kankeleit, E., Gutlich, P., Ming, D. W., Renz, F., Wdowiak, T., Squyres, S. W., Arvidson, R. E. (2004) Jarosite and Hematite at Meridiani Planum from Opportunity’s Mossbauer Spectrometer. Science 306: 1740-1745. Korenaga, J. (2004). Mantle mixing and continental breakup magmatism. Earth and Planetary Science Letters 218: 463-473. Kraft, M. D., Michalski, J. R., Sharp, T. G. (2003) Effects of pure silica coatings on thermal emission spectra of basaltic rocks: Considerations for Martian surface mineralogy. Geophysical Research Letters 30 (24): 2288. Loizeau, D., Mangold, N., Poulet, F., Bibring, J.-P., Gendrin, A., Ansan, V., Gomez, C, Gondet, B., Langevin, Y., Masson, P. and Neukum, G. (2007) Phyllosilicates in the Mawrth Vallis region of Mars. Journal of Geophysical Research 112: E08S08. Leroi, V., Bibring, J-P., Berthe, M. (2009) Micromega/IR: Design and status of a nearinfrared spectral microscope for in situ analysis of Mars samples. Planetary and Space Science 57: 1068 – 1075. Malin, M. C., Caplinger M. A., Edgett, K. S., Ghaemi, F. T., Ravine, M. A., Schaffner, J. A., Baker, J. M., Bardis, J. D., DiBiase, D. R., Maki, J. N., Willson, R. G., Bell, J. F., 34 Dietrich, W. E., Edwards, L. J., Hallet, B., Herkenhoff, K. E., Heydari, E., Kah, L. C., Lemmon, M. T., Minitti, M. E., Olson, T. S., Parker, T. J., Rowland, S. K., Schieber, J., Sullivan, R. J., Sumner, D. Y., Thomas, P. C., Yingst, R. A. (2010) The Mars Science Laboratory (MSL) Mast-Mounted Cameras (Mastcams) Flight Instruments. 41st Lunar and Planetary Science Conference, held March 1-5, 2010 in The Woodlands, Texas. LPI Contribution No. 1533, p.1123. Martinez-Frias, J., Amaral, G., (2006) Astrobiological significance of minerals on Mars surface environment. Reviews in Environmental Science and Biotechnology 4: 219 – 231. Milliken, R. E., Swayze, G. G., Arvidson, R. E., Bishop, J. L., Clark, R. N., Ehlmann, B. L., Green, R. O., Grotzinger, J. P. (2008) Opaline silica in young deposits on Mars. Geology 36: 847 – 850. Morris, R. V., Squyres, S., Arvidson, R. E., Bell, J. F., Christensen, P. C., Gorevan, S., Herkenhoff, K., Klingelhofer, G., Rieder, R., Farrand, W., Ghosh, A., Glotch, T., Johnson, J. R., Lemmon, M., McSween, H. Y., Ming, D. W., Schroeder, C., de Souza, P., Wyatt, M., and the Athena Science Team (2004). A first look at the mineralogy and geochemistry of the MER-B landing site in Meridiani Planum. 35th Lunar and Planetary Science Conference, March 15-19, 2004, League City, Texas, abstract no.2179. Murchie, S. L., Mustard, J. F., Ehlmann, B. L., Milliken, R. E., Bishop, J. L., McKeown, N. K., Noe Dobrea, E. Z., Seelos, F. P., Buczkowski, D. L., Wiseman, S. M., Arvidson, 35 R. E., Wray, J. J., Swayze, G., Clark, R. N., Des Marais, D. J., McEwen, A. S., Bibring, J-P. (2009) A synthesis of Martian aqueous mineralogy after 1 Mars year of observations from the Mars Reconnaissance Orbiter. Journal of Geophysical Research 114: E00D06 Mustard, J. F., Murchie, S. L., Pelkey, S. M., Ehlmann, B. L., Milliken, R. E., Grant, J. A., Bibring, J.-P., Poulet, F., Bishop, J., Noe Dobrea, E. N., Roach, L., Seelos, F., Arvidson, R. E., Wiseman, S., Green, R., Hash, C., Humm, D., Malaret, E., McGovern, J. A., Seelos, K., Clancy, T., Clark, R., Marais, D. D., Izenberg, N., Knudson, A., Langevin, Y., Martin, T., McGuire, P., Morris, R., Robinson, M., Roush, T., Smith, M., Swayze, G., Taylor, H., Titus, T. and Wolff, M. (2008) Hydrated silicate minerals on Mars observed by the Mars Reconnaissance Orbiter CRISM instrument. Nature 454: 305-309. Palomba, E., Zinzi, A., Cloutis, E. A., D’Amore, M., Grassi, D., Maturilli, A. (2009) Evidence for Mg-rich carbonates on Mars from a 3.0 µm absorption feature. Icarus 203: 58-65. Poulet, F., Beaty, D. W., Bibring, J-P., Bish, D., Bishop, J. L., Dobrea, E. N., Mustard, J. F., Petit, S., Roach, L. H. (2009). Key Scientific Questions and Key Investigations from the First International Conference on Martian Phyllosilicates. Astrobiology 9: 257 – 267. Poulet, F., Gomez, C., Bibring, J.-P., Langevin, Y., Gondet, B., Pinet, P., Belluci, G., Mustard, J. (2007) Martian surface mineralogy from Observatoire pour la Mineralogie, 36 l’Eau, les Glaces et l’Activite on board the Mars Express spacecraft (OMEGA/MEx): Global mineral maps. Journal of Geophysical Research 112: E08S02. Poulet, F., J.-P. Bibring, J. F. Mustard, A. Gendrin, N. Mangold, Y. Langevin, R. E. Arvidson, B. Gondet, and C. Gomez (2005), Phyllosilicates on Mars and implications for early Martian climate, Nature 438: 623– 627. Preston, L. J., Benedix, G. K., Genge, M. J., Sephton, M. A. (2008) A multidisciplinary study of silica sinter deposits with applications to silica identification and detection of fossil life on Mars. Icarus 198: 331–350. Pullan, D., Westall, F., Hofmann, B. A., Parnell, J., Cockell, C. S., Edwards, H. G. M., Jorge Villar, S. E., Schroder, C., Cressey, G., Marinangeli, L., Richter, L. and Klingelhofer, G. (2008) Identification of Morphological Biosignatures in Martian Analogue Field Specimens Using In Situ Planetary Instrumentation. Astrobiology 8: 119156. Rice, M. S., Bell, J. F., Cloutis, E. A., Wang, A., Ruff, S. W. Craig, M. A., Bailey, D. R., Johnson, J. R., de Souza Jr., P. A., Farrand, W. H. (2010) Silica-rich deposits and hydrated minerals at Gusev Crater, Mars: Vis-NIR spectral characterization and regional mapping. Icarus 205: 375-395. 37 Romanek, C.S., Grady, M.M., Wright, I.P., Mittlefehldt, D.W., Socki, R.A., Pillinger, C.T., Gibson, E.K. (1994) Record of Fluid-rock interactions on Mars from the meteorite ALH 84001. Nature 372: 655-657. Schulze-Makuch, D, Dohm, J. M., Fan, C., Fairen, A. G., Rodrigues, J. A. P., Baker, V. R., Fink, W. (2007) Exploration of hydrothermal targets on Mars. Icarus 189: 308–324. Sharma, S. K., Misra, A. K., Lucey, P. G., Wiens, R. C., Clegg, S. M. (2007) Combined remote LIBS and Raman spectroscopy at 8.6m of sulfur-containing minerals, and minerals coated with hematite or covered with basaltic dust. Spectrochimica Acta Part A 68: 1036–1045. Sharma, S. K., Misra, A. K., Lucey, P. G., Lentz, R. C. F. (2009) A combined remote Raman and LIBS instrument for characterizing minerals with 532nm laser excitation. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 73: 468 – 476. Sigvaldason, G. E., Steinthorsson, S, Oskarsson, N. and Imsland, P. (1974) Compositional variation in recent Icelandic tholeiites and the Kverkfjoll hotspot. Nature 251: 579-582. Smith, P. H., Bell, J. F., Bridges, N. T., Britt, D. T., Gaddis, L., Greeley, R., Keller, H. U., Herkenhoff, K. E., Jaumann, R., Johnson, J. R., Kirk, R. L., Lemmon, M., Maki, J. N., Malin, M. C., Murchie, S. L., Oberst, J., Parker, T. J., Reid, R. J., Sablotny, R., 38 Soderblom, L. A., Stoker, C., Sullivan, R., Thomas, N., Tomasko, M. G., Ward, W., Wegryn, E. (1997a) Results from the Mars Pathfinder Camera. Science 278: 1758 – 1764. Smith, P. H., Tomasko, M. G., Britt, D., Crowe, D. G., Reid, R., Keller, H. U., Thomas, N., Gliem, F., Rueffer, P., Sullivan, R., Greely, R., Knudsen, J. M., Madsen, B., Gunnlaugsson, H. P., Hviid, S. F., Goetz, W., Soderblom, L. A. Gaddis, L., Kirk, R. (1997b) The imager for Mars Pathfinder experiment. Journal of Geophysical Research 102: 4003-4025. Squyres, S. W., Arvidson, R. E., Ruff, S., Gellert, R., Morris, R. V., Ming, D. W., Crumpler, L., Farmer, J. D., Des Marais, D. J., Yen, A., McLennan, S. M., Calvin, W., Bell, J. F., Clark, B. C., Wang, A., McCoy, T. J., Schmidt, M, E., de Souza, P. A. (2008) Detection of Si-rich Deposits on Mars. Science 320: 1063-1067. Stewart, L., Cloutis, E., Bishop, J., Craig, M., Kaletzke. L., and McCormack, K. (2006). Classification of iron bearing phyllosilicates based on ferric and ferrous iron absorption bands in the 400 – 1300nm region. 37th Lunar and Planetary Science Conference, LPI Contribution No. 2185. Stroncik, N. A., Schmincke, H-U. (2002) Palagonite – a review. International Journal of Earth Science 91: 680 – 697. 39 Treiman, A. H., Amundsen, H. E. F., Blake, D. F., Bunch, T. (2002) Hydrothermal origin for carbonate globules in Martian meteorite ALH84001: a terrestrial analogue from Spitsbergen (Norway). Earth and Planetary Science Letters 204: 323 – 332. Vago, J.L. (2005). ExoMars Science Management Plan, EXM-MS-PLESA-00002. Wiens, R. C., Sharma, S. K., Thompson, J., Misra, A. K, Lucey, P. G. (2005) Joint analyses by laser-induced breakdown spectroscopy (LIBS) and Raman spectroscopy at stand-off distances. Spectrochimica Acta Part A 61: 2324–2334. Yon, S.A., Pieters, C.M. (1988) Interactions of light with rough dielectric surfaces spectral reectance and polarimetric properties. Proceedings of the Lunar Planetary Science Conference 18th, pp. 581- 592. 40 TABLES Table 1. Mineral reflectance spectra used for the selection of geological filter wavelengths for filter set F2-12. Data on grain size and minor mineral impurities are given where information is available. Mineral group Phyllosilicates Sulphates 41 Mineral (sample ID) Nontronite (PS-6Ba) Nontronite (PS-6Bb) Nontronite (PS-6Da) Nontronite (PS-6Aa) Nontronite (PS-6Bc) Nontronite (C1CY20) Nontronite (NG-1a) Nontronite (NG-1b) Nontronite (Swa-1a) Nontronite (Swa-1b) Montmorillonite (PS-2B) Montmorillonite (PS-2D) Montmorillonite (CM26) Montmorillonite (CM27) Montmorillonite (SCa-2a) Montmorillonite (SCa-2b) Montmorillonite (STx-1) Montmorillonite (CM20) Montmorillonite (SAz-1) Montmorillonite (SWy-1) Jarosite (SO-7Aa) Jarosite (SO-7Ab) Jarosite (SO-7Ac) Jarosite (C1CY16) Jarosite (GDS101) Jarosite (GDS24) Jarosite (GDS98) Jarosite (GDS99) Jarosite (JR2501) Jarosite (SJ-1) Alunite (HS925) Alunite (SO-4Aa) Alunite (SO-4Ab) Alunite (SO-4Ac) Alunite (CBCC08) Alunite (CBCC09) Alunite (C1CY02) Alunite (GDS82) Alunite (GDS83) Alunite (GDS85) Gypsum (SO-2Ba) Gypsum (SO-2Bb) Gypsum (SO-2Bc) Gypsum (CBCC16) Gypsum (C1JB556) Grainsize (µm) 0 - 45 45 - 125 0 - 45 0 - 45 125 - 500 <500 <2 <2 0 - 45 0 - 45 0 - 45 45 - 125 125 - 500 <500 300µm 125-500 µm 45 - 125 0 - 45 25 - 75 25 - 75 <500 50 35 0 - 45 45 - 125 125 - 500 25 - 75 <63 Impurities Database Trace quartz Trace quartz Trace quartz Trace quartz Trace quartz Bentonite+quartz+feldspar+bassanite Trace quartz Trace quattz Trace quartz Trace quartz Quartz Quartz + feldspar - JPL JPL JPL JPL JPL RELAB USGS USGS USGS USGS JPL JPL USGS USGS USGS USGS USGS USGS USGS USGS JPL JPL JPL RELAB USGS USGS USGS USGS USGS USGS USGS JPL JPL JPL RELAB RELAB RELAB USGS USGS Pure Pure Pure Pure Trace quartz and clay Pure Pure Minor quartz - RELAB RELAB RELAB RELAB RELAB Opaline Silica Carbonate 42 Gypsum (C1SF03) Gypsum (C1CC36) Gypsum (C2PG03) Gypsum (HS333) Gypsum (SU2202) Opal (C1OP01) Opal (C1OP02) Opal (C1OP03) Opal (C1OP04) Opal (C1OP05) Opal (C1OP06) Opal (C1OP07) Opal (C1OP08) Opal (C1OP09) Opal (TM8896) Calcite (GDS304) Calcite (C-3Ab) Calcite (C-3Db) Calcite (C-3Ec) Calcite (C-3Eb) Calcite (C-3Ea) Calcite (CAL110) Calcite (CRB109) Calcite (CRB112) Calcite (CO2004) <45 <25 <250 242 133 <74 74 – 250 250 - 500 <74 74 – 250 250 - 500 <74 74 – 250 250 - 500 0 - 45 0 - 45 45 - 125 0 - 45 125 - 500 <45 <45 <45 480 Pure Pure Minor quartz RELAB RELAB RELAB USGS USGS RELAB RELAB RELAB RELAB RELAB RELAB RELAB RELAB RELAB USGS USGS JPL JPL JPL JPL JPL RELAB RELAB RELAB USGS Table 2. Filter centre wavelengths (λ) and estimated filter bandpasses at FWHM for B2 and the proposed filter sets. The ‘B2’ filter set is a duplicate of that from the Beagle2 PanCam (Griffiths et al. 2006); ‘F1-12’ has filters regularly spaced every 50nm; F2-12 was calculated statistically. All data are in nm. λ 440 480 530 600 670 750 800 860 900 930 965 1000 43 B2 Bandpass 22 28 32 21 17 18 20 34 42 32 29 28 λ 440 500 550 600 650 700 750 800 850 900 950 1000 F1-12 Bandpass 22 30 29 21 18 17 18 20 32 42 30 28 λ 440 470 510 560 590 650 710 750 820 890 960 1000 F2-12 Bandpass 22 26 30 27 21 18 17 18 27 38 30 28 Table 3. Spectral parameters used to assess the different filter sets (adapted from Farrand et al. 2008). All data are in nm; ‘R’ denotes reflectance. Parameter 440 – 700 Slope 920 – 1000 Slope 900 Band Depth 470 Band Depth 44 Filter Set B2 F1-12 F2-12 B2 F1-12 F2-12 B2 F1-12 F2-12 B2 F1-12 F2-12 Representative filter wavelengths 440 – 670 440 – 700 440 – 710 930 – 1000 950 – 1000 890 – 1000 900 900 890 480 500 470 Description (R670 – R440)/(670 – 440) (R700 – R440)/(670 – 440) (R710 – R440)/(710 – 440) (R1000 – R930)/(1000 – 930) (R1000 – R950)/(1000 – 950) (R1000 – R890)/(1000 – 890) 1 – (R900/[(0.429*R860)+(0.571*R930)]) 1 – (R900/[(0.500*R850)+(0.500*R950)]) 1 – (R890/[(0.500*R820)+(0.500*R960)]) 1 – (R480/[(0.556*R440)+(0.444*R530)]) 1 – (R500/[(0.455*R440)+(0.545*R550)]) 1 – (R470/[(0.571*R440)+(0.429*R510)]) Table 4. Mineralogical (XRD and Raman), compositional (EDS), and textural (BSE/SEM) properties of the Martian analogue samples used to test the filter sets. Associated data and images are given in Figure 4, also showing where EDS spot measurements were taken. NB: EDS composition details the main elements only (quantified as compound wt.%, normalised to 100%) – trace elements are not given. Error values represent one standard deviation (n=10). XRD Raman EDS BSE/SEM NBO Jarosite Jarosite Mineralinfilled vesicles NAL (Yellow) Jarosite Natrojarosite NAL (Red) Alunite/ Natroalunite Haematite 52.6 (±4.6) SiO2 23.2 (±3.8) Al2O3 17.6 (±7.2) FeO 2.0 (±0.8) TiO2 1.6 (±1.1) SO3 A) 98.9 (±0.2) SiO2 B) 86.5 (±1.2) FeO 4.7 (±0.4) SO3 4.2 (±0.7) Al2O3 C) 74.3 (±3) SiO2 16.3 (±2.7) FeO 6.6 (±0.5) Al2O3 GY1 - GY2 Opal Alunite Alunite Vesicular glass clasts, palagonite Nontronite? KH (White) Amorphous, minor calcite 78.2 (±10) SiO2 8.5 (±4.3) Na2O 98.8 (±0.3) SiO2 0.4 (±0.2) Al2O3 56.0 (±1.0) SiO2 16.2 (±0.9) Al2O3 13.9 (±1.8) FeO 7.3 (1.6) MgO 4.2 (±0.4) CaO 97.8 (±0.9) SiO2 1.0 (±0.4) CaO Amorphous silica matrix Microfossils HH Calcite Quartz Amorphous Haematite Sulphur Opal Amorphous silica overlying hyaloclastite Opal, Chlorophyll a 45 Montmorillonite? Opal Mineralinfilled vesicles Multispectral Imaging (440 – 1000nm) Jarosite Jarosite Alunite Table 5. Ability of the Beagle2 (B2), MER, and the two new filter sets to detect specific spectral features in the Martian analogue rock spectra, displayed in Figure 5. Detection is classed as either positive (+) or negative (-). Sample and description HH Hyaloclastite Colour/ Region Whole sample Yellow NAL Acid weathered basalt KH Silica encrusted hyaloclastite GY1 Geysirite GY2 Silcified biomat 46 Red White Lichen White Whole Siderite Whole NBO Acid weathered basalt Orange Feature (nm) B2 MER F1-12 F2-12 440 – 700 (Ferric absorption edge) 470 (Fe3+) 680 (Fe2+) 950 (Fe3+) 800 (Peak) 440 – 700 (Ferric absorption edge) 710 (Peak) 950 (Fe3+) 440 – 700 (Ferric absorption edge) 750 (Peak) 850 (Fe3+) 950 - 1000 (OH-) 670 (Chlorophyll) 800 (Peak) 670 (Chlorophyll) 900 – 1000 (H2O) 950 (OH-) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + - + + + + + + + + + + + + + + + + 700 (spectral plateau) 950 - 1000 (OH-) + + + + + + + + 470 (Fe3+) 640 (Fe3+) 720 (Peak) 950 (Fe3+) 440 – 700 (Ferric absorption edge) 480 (Fe3+) 600 (Start of Fe3+ absorption) 750 (Peak) 950 (Fe3+) POSITIVE NEGATIVE + + + + + + + + + + + + + + + + + + + + + + + + + + 24 4 + + 22 6 + + 26 2 + + 27 1 Table 6. ‘Geology’ Filter wavelengths (in nm) for comparable Mars rover and lander stereo cameras, given in chronological order. Viking[I] (1976 – 1982) Pathfinder[II] (1997) Beagle[III] (2003, Lost) 400 - 500 500 - 600 600 - 700 820 - 920 900 - 940 930 - 1100 443 479 530 599 671 752 801 858 897 931 966 1002 440 480 530 600 670 750 800 860 900 930 965 1000 [I] Evans & Adams (1979); Bell et al. (2003); [IV] 47 [II] Smith et al. (1997b); Malin et al. (2010); [VI] MER[IV] (2004 – Present ) 432 482 535 601 673 753 803 864 904 934 1009 [III] [VII] MSL[VI] (Launch 2011) 440 525 675 750 800 865 905 935 1035 Griffiths et al. (2006) This paper ‘F2-12’[VII] (ExoMars) (Launch 2018) 440 470 510 560 590 650 710 750 820 890 960 1000 FIGURE LEGENDS FIG 1. Martian analogue samples (a) GY2 – silicified biomat; (b) GY1 – silica sinter; (c) KH – subglacial hyaloclastite with opaline silica crust; (d) NAL – acid fumarole weathered lava; (e) HH – subglacial hyaloclastite; (f) NBO – acid weathered lava. Scale bar = 1cm. FIG 2. Plot of filters and estimated filter bandpasses for Beagle2 (B2) and new filter sets, together with vertical spectral bands for Fe (grey, solid line) and water (blue, dashed line), and also reflectance peaks for jarosite and nontronite (pink). Also shown are the mean spectra of each mineral species used to statistically calculate new filter selections (only the mean is shown here for clarity). For specific sample numbers, see Table 1. 48 FIG 3. Reflectance spectra of nontronite and jarosite (SWa-1.a and JR2501 respectively- see Table 1) showing the capability of different filters in identifying diagnostic spectral features. For both minerals, F2-12 provides a better fit to the original spectrum than F1-12. Both F1-12 and F212 are better than the B2 filters. Arrows highlight features that are missed by the filters. 49 FIG 4. Multispectral RGB composite image (scale bar=1cm), BSE image, and Raman spectroscopic data for Martian analogue samples. Boxes in the RGB composites represent multispectral targets used to acquire PanCam spectra, whilst red dots in the BSE images indicate where EDS measurements (given in Table 4) were made. (a) NBO, showing mineral deposits within lava vesicles, identified as jarosite by Raman; (b) NAL, showing ‘Yellow’ and ‘Red’ targets, with Raman data showing the presence of haematite and jarosite; (c) KH, showing ‘White’ and ‘Lichen’ targets, Raman identifies the white crust to be opaline silica; and the hyaloclastite – silica crust contact can clearly be seen under BSE; (d) HH hyaloclastite showing glass clasts surrounded by palagonite. In BSE images, ‘G’ = Glass clast; ‘GM’ = groundmass; ‘J’ = jarosite; ‘P’ = palagonite; ‘S’ = opaline silica; ‘V’ = vesicle. Associated EDS and XRD data is given in Table 4. 50 FIG 4 Continuted. (e) GY2, showing textured biomat surface, which has a dendritic structure under BSE. Raman data indicates the sample to be opaline silica; (f) GY1, showing variable coloured layered regions, with Raman showing the red and yellow regions to be haematite and sulphur respectively; (g) Siderite with confirmatory Raman data. 51 FIG 5. Martian analogue sample spectra modelled for the B2 and two new filter sets ‘F1-12’ and ‘F2-12’. (a) NBO; (b) NAL Yellow (top spectrum) and NAL Red (bottom spectrum); (c) GY1 (top spectrum) and GY2 (bottom spectrum); (d) HH; (e) Siderite; (f) KH White (top spectrum) and KH Lichen (bottom spectrum). Where the MER filters produce a different spectrum to that generated by the B2 filters (due to the absence of a filter at ~965nm), this is plotted on top of the B2 spectra (plots c and d) to highlight where spectral features are missed. 52 FIG 6. GY1 Red and Yellow PanCam spectra using the F2-12 filters, and reflectance spectra for haematite and sulphur (USGS spectral library reference WS161 and GDS94 respectively), all normalised to a starting point of ~0.1. FIG 7. BSE images of the GY2 silica sinter with preserved biological structures within the silica matrix, showing branching dendritic structures that extend up from the base of the sinter (a) and detailed cellular structures (b). 53 FIG 8. Spectral parameter plot of 920 – 1000nm slope vs. 440 – 700nm slope for all hydrated mineral data from Table 1, showing the relationship between the two parameters depicted by the dashed linear trendline (a); and the Martian analogue rock spectra represented by the three different filter sets for the sulphates (b); phyllosilicates (c); and for opaline silica and carbonate (d). 54 FIG 9. Discriminant analysis (a) and hierarchical cluster analysis (b) for the Martian analogue samples as plotted using the three different filter sets. In (a), the further the distance between groups, the better the discrimination between different samples. NB: colours are based on those from the F2-12 analysis. 55