Tree height estimation from multi-temporal ERS SAR interferometric phase
Maurizio Santoro(1) and Patrik Dammert(2)
(2) Ericsson Microwave Systems AB, Airborne Radar Division, 431 84 Mölndal, Sweden
ERS-1/2 SAR interferometry (InSAR) has been shown to provide results with high accuracy for forest mapping and stem volume retrieval
using "tandem" coherence. Another parameter of fundamental importance in forest inventory is tree height. Previous studies have shown
the potential of interferometric phase to retrieve tree height information (Hagberg et al. 1995, IEEE TGRS; Floury et al. 1996, Fringe 96).
At several test sites located in Europe (Sweden, Finland, France and Germany) we have analysed the possibilities and the limits of using
the ERS InSAR phase for tree height estimation. For each test site, between six and nine ERS-1/2 pairs acquired between 1995 and 1999
were available. This enabled a multi-temporal analysis of the dynamics of the InSAR phase signatures and the tree height estimates.
Because of the unstable nature of the canopy, the InSAR phase measurements show a strong noise component, devised by the low
coherence. A large number of interferometric pairs were furthermore affected by atmospheric artefacts. To remove this component, we
investigated several possible filtering techniques (Permanent Scatterers, grid of open areas and tropospheric delay estimation). Although
the phase was generally well corrected, the strong requirements set by each technique limit their application in particular in large
forested areas. In this case it more advisable to choose pairs free from artefacts. Typically, pairs acquired at night, under winter-stable
weather conditions, were found to be free from artefacts.
Retrieved interferometric tree height was between the ground level and slightly above the true height. In particular, the estimates from
pairs acquired at changing weather conditions (precipitation, snowmelt etc.) did not correlate with the true tree height. Pairs acquired
under dry-frozen conditions seemed to strongly underestimate the tree height. With a multi-temporal combination of single-pair
estimates, it was possible to decrease the retrieval error sensibly.
In order to take into account the effects of the interaction of the electromagnetic waves with the tree crowns, a compensation of the
interferometric tree height based on the Interferometric Water Cloud Model has been considered. Model parameters determined from
fitting observed coherence to stem volumes resulted in expected interferometric heights reasonably well in line with observations.
Nevertheless, the strong phase noise and the inhomogeneities within the forest stands resulted in strong uncertainties, thus affecting
the corrected tree heights, which showed only partial correlation with the true tree height.
Concluding, since InSAR phase measurements of forested areas are severely affected by several sources of error and uncertainty, the
usefulness of C-band repeat-pass SAR interferometry for tree height estimation is limited.
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,