Topographic correction of HiRISE and CaSSIS images: Validation and application to color observations of Martian albedo features

https://doi.org/10.1016/j.pss.2021.105198Get rights and content

Highlights

  • We present a methodology to apply the topographic correction to planetary images.

  • We apply our method to correct HiRISE and CaSSIS images from topographic shading.

  • We validate our method with published photometry of Martian Recurring Slope Lineae.

  • The corrected images allow to study the photometry of Martian albedo features.

  • Multi-band photometry of Recurring Slope Lineae hints towards a dry-dust origin.

Abstract

The topographic correction of satellite images has to be applied to both disentangle albedo features from illumination effects induced by local topography and perform spectrophotometric analyses of planetary surfaces. This work focuses on the modeling and removal of surface brightness variations induced by topography, referred as topographic shading, from high resolution images of Mars. Topographic shading can be modeled through functions of the surface illumination and observation angles called disk functions. We consider four disk functions that are widely used in planetary photometry: the Lambert, Lommel-Seeliger, Akimov and Minnaert disk functions. We test and evaluate their performances in removing topographic shading from High Resolution Imaging Science Experiment (HiRISE) and Colour and Surface Science Imaging System (CaSSIS) images. We here validate our method, moreover, we report scientific applications to single or multi-band datasets by analyzing topographically corrected HiRISE colour observations of Martian recurring slope lineae and dust devil tracks, as well as CaSSIS panchromatic observations.

Introduction

The apparent brightness of a planetary surface depends on the illumination and observation geometries and on the composition and texture of the reflecting material. The determination of a mathematical function that can accurately model the reflected radiation at given illumination and observation angles enables quantitative comparisons between photometric measurements of planetary surfaces, laboratory experiments and different spectrophotometric datasets. Such modeling allows to highlight and investigate differences in the scattering properties of the reflecting medium, discriminating albedo variations from illumination effects. If the model depends on physical properties of the regolith, such as composition, grain size and shape, porosity and roughness, it may also be used to indirectly probe such properties. The modeling and removal of brightness variations induced by topography, also referred as topographic shading, are therefore pivotal for the spectrophotometric study of planetary surfaces. For instance, this is required for the production of image mosaics and the analysis of the surface spectral reflectance. In this context, a variety of mathematical functions, or photometric models, have been proposed so far (McEwen, 1991; Hapke, 1993; Fairbairn, 2005, Schröder et al., 2013 and references therein). Some of these are simple combinations of functions of the illumination and observation angles and contain few or no parameters, as the Lambert, Lommel-Seelinger, Minnaert and Akimov models (Minnaert, 1941; Hapke, 1993; Fairbairn, 2005; Schröder et al., 2013). Other formulations consider shadow-hiding effects too and more advanced optical phenomena that depend on the phase angle (i.e. the angle between the observation and illumination directions). In particular, the Kaasalainen–Shkuratov (KS) model allows to couple any of the above simple functions with phase angle ones (Shkuratov et al., 2011; Velikodsky et al., 2011; Wu et al., 2013; Schröder et al., 2013). Finally, the Hapke photometric models (Hapke, 1993, 2002, 2008, 2012) possibly provides the most detailed and complete description of the scattering properties of a reflecting medium, but it implies a relatively complex mathematical expression.

In this work, we will test the performances of the Lambert, Lommel-Seeliger, Minnaert and Akimov photometric functions to model and remove topographic shading from High Resolution Imaging Science (HiRISE, McEwen et al., 2007) and Colour and Stereo Surface Imaging System (CaSSIS, Thomas et al., 2017) Martian images. The choice of using simple photometric models is motivated by the potentially large uncertainties and source of errors that may arise from using complex, multi-parameter models. Indeed, a single HiRISE or CaSSIS image is characterized by a relatively low coverage of illumination and observation angles, preventing an accurate and robust estimation of complex models parameters from a single image (Schmidt and Fernando, 2015). On the contrary, complex models are usually employed when multi-angular observations are available. For example, accurate estimates of the parameters of the Hapke model for Mars from orbit have been performed trough the multi-angular nature of Compact Reconnaissance Imaging Spectrometer for Mars (CRISM, Murchie et al., 2007) “emission phase function” (EPF) observations (Ceamanos et al., 2013; Fernando et al., 2013). Alternatively, they were also achieved by analyzing multiple (≥13) repeated low resolution observations (i.e. above 1 km/px) of the same locations taken with the “Observatoire pour la Minéralogie, l’Eau, les Glaces, et l’Activité” (OMEGA) hyperspectral imager (Vincendon et al., 2007).

While these approaches are very accurate in recovering surface reflectance, they require a high number of images of the same site with different observation geometries, and without significant surface changes between the images. For high-resolution (<5 ​m/px) datasets, this requirement may restrict their applicability to few very well monitored sites (i.e. landing locations), preventing the analysis of more common regions with a low number of observations. Complex models also require an accurate, radiative transfer based atmospheric correction to remove the contribution of aerosol to the observed reflectance, which is out of the scope of our manuscript. For these reasons, we opted for an initial, simplified treatment of the surface reflectance function, leaving a more detailed analysis for a future work. To validate our method, we compare the resulting topographically corrected (i.e. topographic-shading removed) reflectance with the relative albedo measurement of Martian Recurring Slope Lineae (RSL, McEwen et al., 2011) reported in Schaefer et al. (2019) in the RED HiRISE channel (Delamere et al., 2010). We also highlight its scientific benefits by applying it to HiRISE colour observations of albedo features, including RSL and dust-devil tracks. Dust devils tracks are dark or bright-toned lineaments left by the passage of Dust Devils (i.e. turbulent whirlwinds) that lift deposited dust and expose the underlying surface (Reiss et al., 2016). RSL are narrow (<5 ​m) low-albedo linear features, often sourcing from bedrock outcrops and lengthening hundreds of meters down warm Martian steep slopes (McEwen et al., 2011). The nature of RSL is still debated, although an increasing body of work suggest that they may be dry granular flows (Dundas et al., 2017; Schmidt et al., 2017; Schaefer et al., 2019; Munaretto et al., 2020; Gough et al., 2020) possibly of aeolian nature (Vincendon et al., 2019; Dundas, 2020). Both dust-devil tracks and RSL show a different albedo from nearby surfaces, hence representing an interesting scientific application of our method.

The only previous photometric analysis of RSL in all HiRISE filters highlighted that the ratios between RSL colours and adjacent areas are the same, within uncertainties (McEwen et al., 2011). By focusing on ratios between regions with the same illumination conditions, these measurements do not require a topographic correction: for this reason, it provides another benchmark to further validate our method. Nevertheless, the relatively small HiRISE colour swath implies a small probability of finding a common “normalisation” region with the same illumination and observation conditions as the features of interest, such as in our case RSL, dust-devil tracks and their nearby slopes. In most cases the absence of the topographic correction prevents the investigation and comparison of relative albedo of many different features in the HiRISE filters. For this reason, we extended our method to the blu-green (BG, from 400 to 600 ​nm) and near infrared (NIR, from 800 to 1000 ​nm) channels of HiRISE colour orthoimages depicting both RSL and dust-devil tracks. This allows us to investigate and compare the relative albedo of these features in multiple bands.

While absolute photometry would in principle permit a comparison of the RSL reflectances with wetting of dry Martian simulants (Pommerol et al., 2013) or dust fallout (Wells et al., 1984) laboratory measurements, it would require both a convolution of these datasets with the HiRISE bandpasses and a more accurate atmospheric correction (Ceamanos et al., 2013; Fernando et al., 2013; Vincendon et al., 2007), which are out the scope of this work. On the other hand, relative photometry analysis is often the preferred approach when considering HiRISE data (Daubar et al., 2016; Schaefer et al., 2019). Hence, we adopt this simpler approach and compare the relative albedo of dust devil tracks, which are well accepted dust-removed features (Klose et al., 2016), with RSL for all the HiRISE filters.

We apply our method to correct a CaSSIS panchromatic image of the Ascraeus Mons summit caldera complex. This location features multiple terraces, a dense fracture network and several faults (Byrne et al., 2012), resulting in a rich variety of slope orientations that allow to better showcase the potential of our methodology in removing topographic shading.

In section (2) we describe the photometric models tested in this work. In section (3) we describe our dataset and methodology. In section (4) we validate our methods and present few examples of scientific applications. In section (5) we discuss our results and finally in section (6) we draw our conclusions.

Section snippets

Photometric models

The reflectance (r) of a planetary surface illuminated by the Sun is defined as the ratio between the observed radiance (I, units: W m−2 ​μm−1 sr−1) and the normal solar irradiance (J, units: W m−2 ​μm−1) (Hapke, 1993; Schröder et al., 2013) and depends on the local incidence angle of sunlight (il), the local emission angle (el), the phase angle (α) and the wavelength. The local incidence angle is defined as the angle between the direction of the incoming sunlight and the normal to the surface.

Material and methods

We initially apply the topographic correction (i.e. modeling and removal of topographic shading) to HiRISE images of Tivat Crater (−45.9° N, 9.5° E), Mars. This location was selected because it is covered by multiple HiRISE images (see Table 1) and a digital terrain model (DTM) which is required for the application and validation of our method. All these datasets are publicly available through the Planetary Data System (PDS). In addition, orthorectified versions of the HiRISE images are also

Topographic correction of HiRISE colour images of Tivat crater

We apply the topographic correction to all filters of the HiRISE colour orthoimage ESP_021628_1335 of Tivat crater, obtaining the results in Figure (8). As we already considered the RED HiRISE filter in section (3.2), we now extend our analysis to the NIR and BG bands. The local incidence and emission angles maps were already computed in section (3.1) and are depicted in Figure (2). The original and corresponding corrected IRB images are shown in Figure (8a,b). Qualitatively, the corrected

Discussion

We tested and evaluated a method to correct satellite images from brightness differences induced by topographic shading. The aim of the topographic correction is to exploit the resulting corrected images to perform photometric studies of surface features. In this work, we focus on the topographic correction of images of the Martian surface obtained by the HiRISE and CaSSIS visible cameras. We highlight that this method could be applicable to any satellite image if a high-resolution DTM of the

Conclusions

We presented a method to correct the topographic shading in HiRISE and CaSSIS images. Our approach is similar to the photometric correction usually performed on global mosaics or spectroscopic datasets, but it takes advantage of high resolution digital terrain models to compute with high accuracy the local illumination conditions of each pixel of the images. It evaluates the best disk functions among the Lambert, Akimov, Minnaert and Lommel-Seelinger photometric models. We validated our method

CRediT authorship contribution statement

G. Munaretto: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft, Investigation, Writing – review & editing. M. Pajola: Conceptualization, Supervision, Writing – review & editing. A. Lucchetti: Conceptualization, Supervision, Writing – review & editing. C. Re: Software, Writing – review & editing. G. Cremonese: Conceptualization, Resources, Supervision, Funding acquisition, Writing – review & editing. E. Simioni: Software, Writing – review & editing. P.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We gratefully thank reviewer Dr. Mauro G. Spagnuolo and an anonymous reviewer for their thorough and insightful comments that greatly enhanced the quality of this article.

The study has been supported by the Italian Space Agency (ASI-INAF agreement no. 525 2017-03-17).

The authors wish to thank the spacecraft and instrument engineering teams for the successful completion of the instrument. CaSSIS is a project of the University of Bern and funded through the Swiss Space Office via ESA’s PRODEX

References (39)

  • M.B. Fairbairn

    Planetary photometry: the lommel-seeliger law

    ∖jrasc

    (2005)
  • J. Fernando et al.

    Surface reflectance of Mars observed by CRISM/MRO: 2. Estimation of surface photometric properties in gusev crater and meridiani planum

    J. Geophys. Res. E Planets

    (2013)
  • R.V. Gough et al.

    Changes in soil cohesion due to water vapor exchange: a proposed dry-flow trigger mechanism for recurring slope lineae on Mars

    Geophys. Res. Lett.

    (2020)
  • B. Hapke

    Theory of Reflectance and Emittance Spectroscopy

    (2012)
  • B. Hapke

    Bidirectional reflectance spectroscopy. 6. Effects of porosity

    Icarus

    (2008)
  • B. Hapke

    Bidirectional reflectance spectroscopy. 5. The coherent backscatter opposition effect and anisotropic scattering

    Icarus

    (2002)
  • B. Hapke

    Theory of Reflectance and Emittance Spectroscopy, Topics in Remote Sensing

    (1993)
  • M. Klose et al.

    Dust devil sediment transport: from Lab to field to global impact

    Space Sci. Rev

    (2016)
  • A.S. McEwen

    Photometric functions for photoclinometry and other applications

    Icarus

    (1991)
  • Cited by (9)

    • Assessing the spectrophotometric properties of Mercury's hollows through multiangular MESSENGER/MDIS observations

      2023, Icarus
      Citation Excerpt :

      Such a high number of fits prevented the use the grid-search approach for each of them. We instead opted for a numerical minimization approach, like in Munaretto et al. (2021), of the χ function as defined in Domingue et al. (2016) using the minimize routine provided by the SciPy library of the Python programming language and setting the initial parameter values to the ones in Domingue et al. (2016). We validated our approach by running 10,000 “test cases” simulations.

    • Seasonal southern circum-polar spots and araneiforms observed with the colour and stereo surace imaging system (CaSSIS)

      2022, Planetary and Space Science
      Citation Excerpt :

      The relative brightness ratio compares the reflectance of a dark area (i.e. dark core or dark ring) with a closeby bright area (usually the surrounding background terrain). This technique is widely used in Martian planetary photometry studies (Daubar et al., 2016; Schaefer et al., 2019; Munaretto et al., 2020, 2021, 2022) and allows to better compare the photometric properties of surface features imaged under different illumination and observation conditions without the need to assume any particular bidirectional reflectance distribution function (usually the Lambert one) for the surface that is implied for absolute photometry. Relative brightness ratios are then compared between various images to determine the evolution of the relative brightness over a period of time.

    • CaSSIS-based stereo products for Mars after three years in orbit

      2022, Planetary and Space Science
      Citation Excerpt :

      More specifically, CaSSIS DTMs and orthoimages have been already used to investigate the stratigraphy of the South Polar Layered Deposits (Becerra et al., 2019). In addition, they have been used to perform the topographic correction on CaSSIS orthoimages (i.e., modeling and removal of brightness differences induced by topography), allowing to obtain corrected images that will be key for future photometric analyses of Maritan surface features (Munaretto et al., 2021). The ability of CaSSIS to acquire along-track stereo, near-simultaneous stereo pairs avoids illumination differences between images and offers benefits for the DTMs and anaglyphs production.

    View all citing articles on Scopus
    View full text