Surface motion of mountain glaciers derived from satellite optical imagery

Abstract

A complete and detailed map of the ice-velocity field on mountain glaciers is obtained by cross-correlating SPOT5 optical images. This approach offers an alternative to SAR interferometry, because no present or planned RADAR satellite mission provides data with a temporal separation short enough to derive the displacements of glaciers. The methodology presented in this study does not require ground control points (GCPs). The key step is a precise relative orientation of the two images obtained by adjusting the stereo model of one “slave”' image assuming that the other “master” image is well georeferenced. It is performed with numerous precisely-located homologous points extracted automatically. The strong ablation occurring during summer time on the glaciers requires a correction to obtain unbiased displacements. The accuracy of our measurement is assessed based on a comparison with nearly simultaneous differential GPS surveys performed on two glaciers of the Mont Blanc area (Alps). If the images have similar incidence angles and correlate well, the accuracy is on the order of 0.5 m, or 1/5 of the pixel size. Similar results are also obtained without GCPs. An acceleration event, observed in early August for the Mer de Glace glacier, is interpreted in term of an increase in basal sliding. Our methodology, applied to SPOT5 images, can potentially be used to derive the displacements of the Earth's surface caused by landslides, earthquakes, and volcanoes.

Introduction

Accurate displacement measurements are needed to understand the dynamics of glaciers. Such measurements contribute to a better knowledge of the rheological parameters controlling the flow of glaciers. They are important to monitor icefalls, glacier surges (Fischer et al., 2003), and glacier hazards (Kääb et al., 2003). They can also detect ice-velocity changes caused by global warming (Rignot et al., 2002). Differential Global Positioning System (DGPS) ground surveys, synthetic aperture radar interferometry (InSAR), and optical image cross-correlation are the main ways to determine glacier displacements. The first two methods are the most accurate but present some severe limitations for the monitoring of mountain glaciers, i.e. all glaciers except large ice caps, ice fields, and the Greenland and Antarctic ice sheets. This study applies cross-correlation to well coregistered SPOT5 optical images to measure mountain glacier surface velocities.

Even with the advent of the DGPS, it remains difficult and time-consuming to perform regular ground-based surveys of glacier flow. Among the 86 regularly monitored glaciers with time series longer than 10 years (Braithwaite, 2002), only a few stakes on the flat parts can reasonably be surveyed, excluding icefalls and remote glaciers.

The 1990s brought some great improvements in the measurement of the surface motion of glaciers from satellite data. Two new techniques have been extensively investigated, especially on the rapid and large ice streams draining the Antarctic and Greenland ice sheets: InSAR and feature-tracking on optical images.

Combining two SAR images of the Rutford Ice Stream with short time separation (6 days), Goldstein et al. (1993) used InSAR to measure the flow of ice streams. Basic InSAR only measures the projection of the displacement vector onto the satellite line of sight. However, combining ascending and descending passes of the satellite and adding constraints on the ice flow yields all three components of the displacement vector (Joughin et al., 1998, Mohr et al., 1998). Recently, intensity-tracking and coherence-tracking, two cross-correlation techniques applied to SAR data, have been combined with InSAR to produce two-dimensional velocity fields (Gray et al., 2001, Strozzi et al., 2002).

If the correlation between the two radar images is good and the tropospheric, orbital, and topographic contributions can be modelled, the precision of InSAR is on the order of a centimeter (Massonnet & Feigl, 1998). However, additional errors may arise in resolving the phase ambiguity through unwrapping, especially in areas where the displacement gradient (i.e. strain) approaches the threshold of about 10⁻³ for ERS (Massonnet & Feigl, 1998). This condition often occurs in icefalls or marginal shear zones of glaciers (Goldstein et al., 1993).

On mountain glaciers, only a few studies (Mattar et al., 1998, Rabus & Fatland, 2000, Strozzi et al., 2002) have succeeded in measuring motion with InSAR. The steep topography, the strong tropospheric contribution, and the small size of the glaciers are obstacles to overcome. But the major problem is the time between two successive images. If it exceeds 1 or 3 days (Strozzi et al., 2002), the displacement gradient is larger than the threshold (10⁻³ for ERS), destroying the interferometric fringes. Furthermore, after a few days, the correlation is low due to rapid changes on the glacier surface. Only the ERS-1 ice phase (3-day orbital cycle) and the ERS-1 and ERS-2 Tandem Mission (1 day separating the passes of the satellites) can be used to derive velocity fields on glaciers. Consequently, no present or planned satellite mission can measure the motion of mountain glaciers using InSAR.

Repeated visible or near infrared images of the same area can be used to track the displacement of features such as crevasses or surficial debris moving with the ice (Lucchitta & Ferguson, 1986). Development of automatic feature-tracking algorithms has substantially increased the accuracy and the efficiency of this approach (Scambos et al., 1992).

The aim of our study is to demonstrate that high-resolution and accurate surface displacement maps can be routinely obtained on mountain glaciers using optical images. The goal is to provide an alternative to InSAR for the measurement of the glacier flow.

Correlation of optical images provides the two horizontal components of the displacement vector contrary to InSAR. Furthermore, the measurement is unambiguous: absolute displacements can be referenced to motionless areas which are always available for mountain glaciers. This approach can be applied to images with a large time separation. For some outlet glaciers of Greenland or Antarctic ice sheets, the persistence of the surface features permits velocity measurements from images separated by as much as 11 years (Berthier et al., 2003). Some velocity fields have also been derived from optical images separated by more than a year on mountain glaciers (e.g., Kääb, 2002). Previous studies on Antarctic ice streams (Scambos et al., 1992, Frezzotti et al., 1998) or mountain glaciers (Kääb, 2002) generally reached an accuracy of ±1 pixel. A smaller uncertainty and a methodology adapted to mountain glaciers are the focus of our study.

In the next section, we describe a procedure to extract displacements of the ground surface from two SPOT5 images. The images and the different data needed to apply and validate our methodology are presented for the Mont Blanc area in Section 3. Maps of the satellite-derived displacements and accuracy of our measurements are presented in Section 4. An acceleration event of the Mer de Glace glacier is also discussed before presenting conclusions.