You must have a javascript-enabled browser and javacript and stylesheets must be enabled to use some of the functions on this site.


Storm Forest Damage Mapping Based on VHR INSAR Data

Herve Yesou(1) and Cyrille Maire(2)

(1) SERTIT, Parc d'innovation, BP 10413 67412 ILLKIRCH CEDEX, France
(2) DLR, Oberpfaffenhofen, D-82234 Wessling, Germany


The devastating storms, that struck continental Western Europe in December 1999, destroyed the equivalent of several years of normal forest exploitation in France. Remote sensing data, more particularly SAR sensors with their “all-time-weather” capability and a very good potential for change detection within forested areas, seem to be useful for these type of applications. This paper is focused on the assessment of Very High Resolution images (VHR), INSAR data.

An advantage of interferometric systems is their capacity to generate topographic or altitude information with a metric relative accuracy. Up to now, airborne data are used to simulate at least partially future very high resolution space systems. The short revisit period of the next acquisition systems is the condition to preserve an important temporal coherence, and consequently very accurate DEM, even over forest areas. Within the framework of the Haguenau CNES project, it was possible to test the operational potential of such systems for forest damage mapping and forest exploitation monitoring. Over the Haguenau test site VHR INSAR data, X band, with sub-metric resolution, had been acquired in May 2000 using the airborne AeS-1 system Derived products such as a coherence, intensity images and a VHR DEM, were then generated by Aerosensing GmbH [Intermap].

The hypothesis was the following: the INSAR DEM represented the top canopy after the storm. So a pre storm top canopy was then generated based on a HR topographic DEM enhanced through the integration of an important network of differential GPS measurements providing accurate soil level altimetric information. Then, at a parcel level, based on the use of the species database and of tree growth models provided by the National Forest Office, a top canopy height was generated. The altimetric accuracy of the two DEMs was within 3 meters which is very satisfactory in a forestry environment. So, larger changes observed between the two DEMs would represent logging due to normal exploitation as the last update of the forestry database was done in 98, and windfall due to the December 99 storm. This produced a damage map showing the location of very narrow affected areas, small spots was then compared with the reference data of French Forestry Service. The calculated quality parameters of this mapping are very good, with a detection rate of 92% associated with low rates of false detection and of none detection, respectively 12% and 9%. The problematic parcels correspond in fact to young coniferous stands and to stands with two levels of trees where the higher, coniferous trees fell and the lower deciduous resisted the wind and kept upright. The analysis of amplitude data would then allow to correctly identify affected areas within these problematic stands. This highlight the synergetic use of coherence and amplitude SAR data thank to the very high resolution.

In conclusion, the obtained results show that forest monitoring at a local or regional scale will be a possible application for future INSAR systems such as Skymed Cosmo, Terra SAR X or the Interferometric Cartwheel.


Full paper

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry