Einzelbeispiel satellitengestützter Erderkundung

Schneeflächen in den Alpen

ERS SAR multitemporal ERS SAR mit DGM Landsat 743


Snow cover information of the Central Austrian Alps

Snow coverage information is vital in for the management of water resources commonly used for electric energy production, water supply and irrigation. In a SAR image the signature of snow depends largely from its free water content.

Dry snow is rather transparent for microwaves, whilst melting snow attenuates the backscattering and interacts as a smooth surface. This results into a much lower backscatter in case of wet snow. The assessment of melting snow cover is the most important parameter for run-off models.

Unfortunately, where snow monitoring is more important, like in mountainous areas, SAR has a limited use due to the complex image geometry.

In order to obtain the best possible coverage necessary to measure the snow surface extension, precise map information together with satellite data from ascending and descending passes are required.

The figure shows a multitemporal SAR image of the Oetztal area in the Central Alps of Austria in ground range projection. Remarkable are the glacier tongue of the Hintereis Ferner in the centre and the Weisskugel (3736 msl) to the left.

No geometric nor radiometric corrections have been applied, hence only the slopes away from the SAR illumination are fully visible. The white part of the mountains represents the illuminated slopes with extreme foreshortening and layover.

The following acquisitions have been used:
- 1 June 1992 displayed in blue
- 6 July 1992 in green
- 10 August 1992 in red.

The surface including wet snow on 1 June is given as red and yellow-green, while on 6 July the wet snow surface is shown in magenta.
The very dark area correspond to the snow coverage on 10 August. Trihedral corner reflectors deployed on the high glacier plateau in July and August can be seen as 5 bright yellow spots. Bluish and white areas cannot be used since they fall into zones of strong foreshortening or layover.

Absolute surface measurements can only be done after referencing the image to a similarly projected digital terrain model. Radiometric corrections are also needed to achieve a homogeneous "colour saturation", necessary for digital image classification.

In this figure 2 , ascending (ENE-looking) and descending (WNW-looking) SAR passes have been combined and radiometrically and geometrically corrected on the basis of a digital terrain model. The image combination process selects the optimum local incidence angle to minimize the loss of information due to foreshortening and layover.

The layover area consists of 35.99 % in the images from the ascending pass and of 32.68 % in the images from the descending pass. After an optimised combination, only 6.65 % of the area of layover remains (displayed in black in the image).
The following data were used.

On 27 April, the whole imaged area was covered by snow. On the high glacier plateau strong backscattering is observed (blue) due to the dry firn which include ice lenses.
On 1 June 1992, the magenta and blue gives the extension of the wet snow area.
On 6 July, the remaining snow was wet and includes the area of cyan (greenish blue) and blue colour.

It results from this analysis that at all acquisition dates the area of wet snow could be determined, and in practice the extension can be accurately mapped.

Comparison with optical data

While the availability of optical images during the snow melting period is very much dependent from cloud coverage, their is a much higher reliability to access SAR images. In 1992 only 2 useful Landsat TM could be acquired, one on the 12 May 1992 and one on 29 June 1992 (partly clouded).

The later image 3 shows the situation one week before the SAR acquisition of 6 July 1992. Although the optical image is easier to analyse, since snow (light blue in the TM image) can be distinguished well from other features including clouds (white) by using TM channels 743, SAR data might serve equally good after further steps of image filtering.

The distribution of snow in the SAR image (cyan and blue) is in good agreement with the snow coverage in the optical image (light blue), considering also the time difference of the data acquisitions.

Despite a more complex data handling, once a dedicated processing chain is in place, data from SAR and optical sensors can equally contribute to a reliable and weather-independent snow monitoring system.

(Image processing and text by:
Thomas Nagler, Helmut Rott
Institute for Meteorology and Geophysics University of Innsbruck/Austria)


Quelle: ERS Spaceborne Radar Imagery, Copyright ESA 1996