Polarised bidirectional reflectance factor measurements from soil, stones, and snow

https://doi.org/10.1016/j.jqsrt.2009.04.008Get rights and content

Abstract

The bidirectional reflectance factor and degree of linear polarisation of selected snow, soil, and gravel types have been measured using Finnish geodetic institute field goniospectropolariphotometer (FIGIFIGO). It was observed that with all measured samples the degree of linear polarisation is weakest 030 backwards from nadir, turning 1–5% negative in the backward direction, and growing larger (5–50%) in the forward and Brewster directions. Polarisation was found to be inversely but non-linearly proportional to reflectance. In addition, a wavelength-dependent trend was found to exist in some data, but in general, the wavelength dependence was smooth. Dry old snow polarises clearly more than melting snow or new snow. White gravel polarises somewhat similarly to snow in visible region, and black gravel resembles snow in infrared.

The authors conclude that polarisation observations may be useful in land surface remote sensing, but that only far-forward angles (6070) contain a strong enough signal for easy interpretation. The polarisation spectrum brings little additional information to reflection spectrum, but in conditions where the reflectance spectrum cannot be accurately normalised or incident irradiation is not known, polarisation spectrum could still be usable. The directional signal is strong and may yield refractive index or microstructure, and must be understood in any application. Polarisation near nadir is low for all known samples, and can thus be safely ignored with most non-polarised imaging applications. However, for accurate atmospheric polarimetry, e.g. by polarisation and directionality of the earth's reflectances (POLDER) or aerosol polarimetry sensor (APS) sensors, it is recommended to take into account land surface polarisation.

Introduction

Polarisation observations have been coming to land surface remote sensing at least 20 years [1], [2], [3]. Despite many promising results [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], practical applications have remained few. The biggest polarisation project has been the polarisation and directionality of the earth's reflectances (POLDER) sensor [14]. Although it is primarily designed for atmospheric observations [15], [16], some land surface applications have also been developed [17], [18]. The airborne version of POLDER (OSIRIS) is opening new possibilities [19]. Another planned atmospheric polarisation sensor is the aerosol polarimetry sensor (APS) [20] in the NASA Glory mission. It is assumed, that in near future polarisation missions more suitable for land surfaces may be introduced.

However, before new missions can be seriously planned and proposed, polarisation from various targets should be much better understood and modelled. Even, when observing primarily atmosphere, the land surface background reflectance and polarisation signals must be well known and removed from the data. Polarisation measurements performed so far have been few, and limited by spectral and directional coverage. Large number of more systematic measurements are needed, before useful invertible models can be developed, and new applications created.

Polarisation of electromagnetic radiation waves, such as light, is defined by the direction of the electric field. Natural light is a mixture of a very large number of various wave packets. Sunlight and many other light sources contain equal amount of all polarisation states, thus said to be unpolarised. Scattering, reflection, or transmission through asymmetric media often polarise light. The most significant polarisation mechanisms are the Rayleigh scattering (resulting in up to 90% polarisation of blue sky) and Fresnel reflection (explaining the up to 100% polarisation of water surfaces). However, all scattering mechanisms polarise to some extent, although the more complicated ones usually less so. One discussed subtlety is the small negative polarisation near backward scattering angles caused by multiple scattering coherence effects [21], [22]. The polarisation properties depend mostly on the chemical composition (refractive index) and wavelength and subwavelength scale structures of the scatterers, and amount of multiple scattering.Conversely, these properties could be studied using polarisation, as is successfully done in aerosol research and astronomy.

The most convenient method for measuring the reflective properties of surfaces is to determine the bidirectional reflectance factor (BRF for short, denoted as R here), which is the ratio of the reflected light to an ideal Lambertian reflector under same incident irradiation. This factor is a function of four angles and wavelength, see Fig. 1. The reflected radiance can then be given as L=dμ0dφ0RL0, where L0 is the incident radiance (distribution) and integration goes over the hemisphere (μ0=cosι,φ0, see Fig. 1). In description of the polarisation the BRF is pictured as a 4×4 Muller matrix, and both reflected and incident light are described with the Stokes vector [I,Q,U,V].

Earlier measurements have already shown that land surfaces polarise weaker than atmosphere, and mostly in forward direction. Vegetation is discussed more in a preceding paper [23]. Goloub et al. [18] analysed POLDER data about snow and clouds and observed that snow polarised a few per cent in the forward direction, but did not show the rainbow arc feature present in clouds. Perovich [7] measured polarised reflectance from many ice and snow samples, observing very strong forward polarisation from ice and rather weak (<8%) polarisation from dry snow and half of that from melting snow. Bréon et al. [6] and Taixia and Yunsheng [10] measured polarisation from some bare soil samples. They too observed small negative polarisation in the backward directions and moderate positive polarisation in the forward direction that could be explained for the most part by Fresnel reflections.

In this paper we aim to provide spectrally- and directionally-resolved measurement data of selected land surface objects in order to quantify in detail the polarisation features and their usefulness. Puttonen et al. [24] have already reported some gonio-spectro-polarimetric reflection measurements from asphalt, sand, and concrete surfaces, and Suomalainen et al. [23] measured bidirectional polarisation factor from several pieces of vegetation. This paper provides more data from non-vegetated land surfaces.

Section snippets

Instruments and measurement practices

The measurements were taken using the Finnish geodetic institute field goniospectrometer (FIGIFIGO), see Fig. 2. The FIGIFIGO system consists of a motor-driven moving arm that tilts up to ±90 from the vertical, variable fore optics in the high end of the arm, and an ASD FieldSpec Pro FR 350–2500 nm spectroradiometer. Accurate angles are read with an inclinometer and an electronic compass. The position is determined with GPS. The system can be mounted on a light sledge for snow measurement, on a

Campaigns and targets

The snow measurements come from two campaigns in Sodankylä, in North Finland, and one in Masalasee Table 1. Year 2007 measurements were taken in a semi open area in sunlight under clear sky. The snow was melting rapidly, and the snow surface was rather rough. During the next campaign in 2008, the sky was overcast most of the time, and all but one measurements are performed during nights using artificial illumination in a tight spot inside sparse forest. In the beginning the snow was several

Results

The results are shown in Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 8, Fig. 9, Fig. 10, Fig. 11. For comparison, the angle of incidence for the presented results is selected around 65 for snow and 55 for gravel and sand. For each target a six element figure of measurement results is given. An illustrative photograph is shown first, then in the left column the BRF cake in red channel (670±10nm) and the brightness spectrum in four angles in the principal plane. In the right column, the degree

Conclusions

We have measured the spectral and angular distribution of the linear polarisation (-Q/I) of reflected light from several bare-land surface types in more detail than has been done before. In total, our measurement data base now contains about 15 vegetation samples and 20 bare-land samples measured with polarisation, and BRF measurements without polarisation from over 200 samples. This study therefore confirmed the prior knowledge, that in general, land surfaces polarise weakly, and no dramatic

Author contribution

Jouni Peltoniemi supervised the work, analysed the data and wrote most of the paper, with contributions from everyone. Juha Suomalainen, Eetu Puttonen and Teemu Hakala performed the measurements. Teemu Hakala constructed the polarisation device with Juha Suomalainen.

Acknowledgements

The work has been supported by the Academy of Finland, the European Commission Interreg program project Norsen, and the Finnish Ministry of Agriculture and Forestry. Some measurements were made during the Snortex campaign. The authors would like to thank many of our Norsen and Snortex partners.

The authors declare that, at the time of submission, they have not been anyhow involved in the afore mentioned nor competing satellite missions.

References (30)

  • J.I. Peltoniemi

    Spectropolarised ray-tracing simulations in densely packed particulate medium

    J Quant Spectrosc Radiat Transf

    (2007)
  • W.G. Egan

    Photometry and polarization in remote sensing

    (1985)
  • D.A. Talmage et al.

    Remote sensing using partially polarized light

    Int J Remote Sensing

    (1986)
  • V.C. Vanderbilt et al.

    Specular, diffuse, and polarised light scattered by two wheat canopies

    Appl Opt

    (1985)
  • G. Rondeaux et al.

    Polarisation de la lumiere par les couverts végétaux: possibilités d’applications agronomiques

    Can J Remote Sensing

    (1990)
  • V.C. Vanderbilt et al.

    Polarization of light by vegetation

  • F.-M. Bréon et al.

    Polarized reflectance of bare soils and vegetation: measurements and models

    IEEE Trans Geosci Remote Sensing

    (1995)
  • D.K. Perovich

    Observation of polarisation of light from sea ice

    J Geophys Res

    (1998)
  • A.A. Kokhanovsky

    Polarization optics of random media

    (2003)
  • Elias TG, Cairns B, Chowdhary J. Surface optical properties measured by the airborne research scanning polarimeter...
  • W. Taixia et al.

    The bidirectional polarized reflectance model of soil

    IEEE Trans Geosci Remote Sensing

    (2005)
  • M. Shibayama et al.

    Estimating the leaf inclination angle of wheat canopies using reflected polarized light

    Plant Prod Sci

    (2007)
  • M. Shibayama et al.

    Testing polarisation measurements with adjusted view zenith angles in varying illumination conditions for detecting leaf orientation in wheat canopy

    Plant Prod Sci

    (2008)
  • S. Liang et al.

    Remote sensing signatures: measurements, modelling and applications

  • P. Deschamps et al.

    The POLDER mission: instrument characteristics and scientific objectives

    IEEE Trans Geosci Remote Sensing

    (1994)
  • Cited by (84)

    • Extension of the Hapke model to the spectral domain to characterize soil physical properties

      2022, Remote Sensing of Environment
      Citation Excerpt :

      Reportedly, soil reflectance generally decreases from the backscattering to the forward-scattering directions and typically shows a reflectance peak in the backward direction when the incidence and viewing directions coincide (i.e., the hotspot effect) (Hapke, 1981, 2002, 2012; Liang and Mishchenko, 1997). Soil anisotropy is very essential in remote sensing scenes, and its effect is even more important when data are collected at different solar-view angles (Peltoniemi et al., 2009). Additionally, soil is a complex medium, and many factors, such as the soil moisture content (SMC), affect its directional and spectral characteristics under natural conditions (Liang and Townshend, 1996a, 1996b; Jacquemoud, 1992).

    • Light scattering from volcanic-sand particles in deposited and aerosol form

      2019, Atmospheric Environment
      Citation Excerpt :

      As such, our experiment reproduces both of measurement scenarios relevant for remote-sensing observations of volcanic sand, i.e., in the atmosphere or deposited on an ice/snow surface. Previous related measurements mainly focus on either single-particles or deposited-particles (Muñoz et al., 2004, Dabrowska et al., 2015; Hadamcik, 2002; Sun et al., 2014; Peltoniemi et al., 2009, Wilkman et al., 2016). There are only few examples where both light-scattering scenarios are simultaneously studied (e.g., Shkuratov et al., 2004, Shkuratov et al., 2006; Mirvatte et al., 2011) and Icelandic volcanic sand is not encompassed in that work.

    View all citing articles on Scopus
    View full text