Remote sensing based retrieval of snow cover properties

https://doi.org/10.1016/j.coldregions.2008.07.007Get rights and content

Abstract

In order to overcome the restrictions of conventional observation methods, novel remote monitoring techniques such as terrestrial laser scanning (TLS) and ground based interferometric synthetic aperture radar (GB SAR) are concurrently operated. Snow depth and snow water equivalent (SWE) or the snow mass on ground are some of the key parameters in the assessment of avalanche hazard, for snow, snow drift and avalanche modelling as well as model verification. While the TLS provides maps of the spatial snow depth distribution, the GB SAR can in principle be used to retrieve snow depth and SWE. Remote sensing results are compared to traditional field work, additionally advantages and limitations of the techniques are identified. Finally, the applicability of the remote sensing based retrieval of these snow cover properties for snow and snow avalanche applications is summarized.

Introduction

Traditional observation techniques (snow pits, probing, ultrasonic snow depth sensors) provide primary input data for fore- and nowcasting snow avalanche hazard in alpine regions. Because of inhospitable weather conditions and inaccessibility due to avalanche danger, in situ observations in avalanche terrain are rare. Remote monitoring techniques offer the possibility to retrieve important snow cover parameters such as snow depth and snow water equivalent (SWE) from a safe distance. During the last decade substantial progress has been made in the development of physically based models (snow, snow drift and avalanches) and avalanche fore- and nowcasting tools (e.g. Bartelt and Lehning, 2002, Sampl and Zwinger, 2004). High resolution retrieval of snow cover properties is needed as model input, for model optimization and verification (Sailer et al., 2008) and the forecast of avalanche danger.

Objectives of the study carried out in the framework of the GALAHAD project (Advanced Remote Monitoring Techniques for Glaciers, Avalanches and Landslides Hazard Mitigation) are (i) the definition of the requirements for improved remote monitoring of snow depth and SWE, (ii) the validation of the remote monitoring observations and (iii) the evaluation of the fulfillment of the requirements. GALAHAD is a European Union funded research project focused on the development of advanced and innovative remote monitoring techniques, namely GB SAR (ground based synthetic aperture radar) interferometry and TLS (terrestrial laser scanning).

TLS is used in several applications such as scanning architecture (Pfeifer and Rottensteiner, 2001), topography (digital elevation models), landslides (Rowlands et al., 2003) and the derivation and interpretation of geomorphologic structure (Deline et al., 2004, Conforti et al., 2005). The use of airborne laser scanners is gaining importance in glaciological applications, in particular for the generation of glacier surface models (Baltsavias et al., 2001) and measurements of ablation and accumulation of snow and ice at an annual time scale (Lippert et al., 2006, Geist et al., 2003, Geist et al., 2005). However, adopting the means of laser scanning for snow and avalanche research is rare, up to now only few projects have been carried out. The TLS technology has been used for snowpack measurements within the SAMPLE project (Snow avalanche monitoring and prognosis by Laser equipment; Moser et al., 2001). Prokop et al. (in press) used TLS for the determination of the spatial snow depth distribution on slopes. The authors reported a mean deviation between TLS and tachymetry (reference measurements) of 4.5 cm with a standard deviation of 2 cm up to a distance of 300 m. Snow depth mapping in a forested area has been carried out with airborne laser scanning (Deems and Painter, 2006).

The estimation of snow parameters can benefit from microwave remote sensing based on passive (radiometry) and active (scatterometry, SAR) radar techniques. Different algorithms have been developed during the last years for the retrieval of SWE, e.g. Shi and Dozier (2000) for a radar algorithm. Dry snow layers at longer wavelengths (L to C band) can be considered almost transparent with a moderate volume scattering depending on the frequency. For dry snow conditions the penetration depth for the C band amounts to about 20 m. In contrast to the TLS, the main contribution to the backscattered signal stems from the snow/ground interface. When targeting wet snow, attenuation occurs due to the presence of liquid water, thus the interaction becomes more complex and the penetration depth reduces dramatically to a few centimetres. Higher frequencies show an increased sensitivity to dry snow properties, but have a limited ability of penetrating a wet snow cover. In the last years the capability of mapping snow cover by means of SAR images from satellite has been widely investigated. In particular C band data have been suggested for the classification and the discrimination of bare surface and snow covered area (e.g. Bernier and Fortin, 1998). In order to retrieve the SWE of dry snow cover from C band interferometric data available from spaceborne platforms, a retrieving approach has been investigated and applied to ERS interferometric data by Guneriussen et al. (2001). Changes of the snow properties between two consecutive interferometric SAR images cause changes of the interferometric phase. Several further studies demonstrated the capability of spaceborne and airborne SAR systems for the retrieval of dry snow properties (Koskinen, 2001, Rott et al., 2004). A similar approach has been applied for a ground based interferometer by Martinez-Vazquez and Fortuny-Guasch (2006) and by Luzi et al. (2007). Avalanche tracks appear as zones of high degradation of the coherence, in interferometric phase maps they show up as areas of random noise. In addition an algorithm to retrieve the depth of dry snow was developed and validated (Martinez-Vazquez and Fortuny-Guasch, 2006).

The requirements for improved remote monitoring of snow depth and SWE were defined as follows. The operational range of the instruments must cover the entire avalanche track, in order to determine the avalanche mass balance. Additionally the range must allow the installation of the sensors in a safe distance of the target, providing observation geometry suitable for the specific applications. Continuous observations with adequate spatial (less than 5 m) and temporal resolution (1 h to 1 day) are required for snow cover models, avalanche dynamics models, forecast of avalanche hazard and snow redistribution studies. The accuracy of the observations has to be high enough to resolve significant snowpack changes in order to observe the snowpack evolution during a storm (accumulation, wind influence and settlement), before and after avalanche events and lies at about 0.1 m (mean absolute error, MAE). The acceptable error of the SWE retrieval is about 20%. Observational and technical requirements are summarized in Table 1.

Section snippets

Test site

The GALAHAD project test site for snow and avalanche observations is located in the Wattener Lizum (province of Tyrol, Austria), a training centre of the Austrian Army (Fig. 1). The study area (Fig. 2) is equipped with four automatic weather stations (AWS). The Meteo Slope AWS is in the line of sight of the remote monitoring instruments in the central part of the target slope. Ultrasonic snow depth measurements deliver ground truth observations for the verification of the remote sensing data.

Technique

Laser scanning, based on the measurement of the time of flight of short laser pulses, is utilized to determine the distance to a target. The wavelengths of laser systems depend on the application and range from about 250 nm to 11,000 nm, corresponding to a range from UV to IR (Weitcamp, 2005). Multiple scattering of the light travelling through the snowpack restricts the penetration depths to no more than 0.5 m in blue wavelengths and only a few millimetres in the near infrared and infrared (

Technique

The snow mass distribution along an avalanche track (start mass and potential entrainment mass) is one of the key input parameters for avalanche dynamics simulations. The potential risks and high spatial variability characterize in situ SWE observations (snow pits). They are not suitable for the estimation of the snow mass distribution along an avalanche path. SWE (product of snow depth and the mean density of a snow column) is defined as the height of the equivalent water column [mm] or as

Conclusions

Two remote monitoring techniques were concurrently used for measuring snow depth and SWE. Whereas the TLS technique just provides information on the snow depth, GB SAR interferometry offers the opportunity to deduce SWE. Snow depth, derived from TLS, was validated with other observation techniques (tachymetry, snow stakes and ultrasonic measurements) in a validation area in up to 1 km distance to the instrument. Considering the measurement uncertainties of the applied techniques, the TLS

Acknowledgements

The GALAHAD project is funded by the European Union (Specific Targeted Research Project FP6-2004-Global-3, N. 018409). The authors would like to thank colonel Knoll, head of the military training camp in Wattener Lizum, and his team for the support and the two anonymous reviewers for the detailed and valuable comments.

References (24)

  • BarteltP. et al.

    A physical SNOWPACK model for the Swiss Avalanche Warning Services. Part 1: numerical model

    Cold Regions Science and Technology

    (2002)
  • BaltsaviasE.P. et al.

    Digital surface modelling by airborne laser scanning and digital photogrammetry for glacier monitoring

    Photogrammetric Record

    (2001)
  • BernierM. et al.

    The potential of time series C-Band SAR data to monitor dry and shallow snow cover

    IEEE Transactions on Geoscience and Remote Sensing

    (1998)
  • ConfortiC. et al.

    Terrestrial Scanning Lidar Technology applied to study the evolution of the ice-contact image lake (Mont Blanc, Italy)

  • DeemsJ. et al.

    Lidar measurement of snow depth: accuracy and error sources

  • DelineP. et al.

    Drainage of ice-contact Miage Lake (Mont Blanc Massif, Italy) in September 2004

    Geografia Fisica e Dinamica Quaternaria

    (2004)
  • DozierJ. et al.

    Multispectral and hyperspectral remote sensing of Alpine snow properties

    Annual Review of Earth and Planetary Sciences

    (2004)
  • GeistT. et al.

    Airborne laser scanning technology and its potential for applications in glaciology

    International Archives of Photogrammetry, Remote Sensing and Spatial Information Science

    (2003)
  • GeistT. et al.

    Investigation on intra-annual elevation changes using multitemporal airborne laser scanning data — case study Engabreen, Norway

    Annals of Glaciology

    (2005)
  • GuneriussenT. et al.

    InSAR for estimating changes in snow water equivalent of dry snow

    IEEE Transactions on Geoscience and Remote Sensing

    (2001)
  • JörgP. et al.

    Measuring snow depth with a terrestrial laser ranging system

  • KoskinenJ.T.

    Snow monitoring using interferometric TOPSAR data

  • Cited by (0)

    View full text