ERS INSAR data have proved to be of interest for forest applications. The i
nterferometric coherence was found to be related to various land uses and f
orest types, while in some special cases (e.g., flat terrain) the interfero
metric phase has been linked to the forest height. This paper reports an in
vestigation on the information content oft he interferometric coherence ove
r a hilly terrain supporting various land use types and large pine plantati
ons. The approach includes the rise of a Geographic Information System and
multitemporal data to analyze the coherence behavior as a function of fores
t-type forest parameters and environmental factors such as meteorological a
nd topographic effects. Coherence appears to be efficient to discriminate b
etween forest types. However, topography and environmental conditions stron
gly affect the coherence and its estimation, pointing out the need for reje
ction of strong slopes areas (>15 degrees) and the sensitivity to local met
eorological/seasonal effects. Based on these observations, forest classific
ation results are presented. Forest/nonforest discrimination is very effici
ent (accuracy >90%) using one-clay interval acquisition. More detailed clas
sification with discrimination between forest themes gives also good result
s. Then, we investigate the indirect link between coherence and forest para
meters. The coherence is sensitive to the forest growth stage, making fores
t parameter retrieval possible using a simple straight-line model. Finally,
the importance of wind upon temporal decorrelation is addressed and a semi
empirical correction is proposed. (C) Elsevier Science Inc., 2000.