Examination of the physical background underlying the ERS response of fores
t and analysis of time series of ERS data indicates that the greater tempor
al stability of forest compared with many other types of land cover present
s a means of mapping forest area. The processing chain necessary to make su
ch area estimations involves reconstruction of an optimal estimate of the b
ackscattering coefficient at each pixel using temporal and spatial filterin
g so that classification rules derived from large scale averaging are appli
cable. The rationale behind the filtering strategy and the level of averagi
ng needed is explained in terms of the observed multitemporal behavior of f
orest and nonforest areas. Much of this analysis is generic and applicable
to a wide range of situation in which significant information is carried by
multitemporal features of the data. The choice of decision rules is based
on the forest observations, with the added requirement for robustness. The
performance of a classifier based only on change is assessed on a range of
test sites in the U.K., Finland, and Poland. Error sources in this classifi
er are identified, and the possibility of improving performance by includin
g radiometric information in the mapping strategy is discussed. Brief discu
ssions of how the classification is affected by the addition of coherence a
nd how the processing chain would need to be modified for other forms of sa
tellite data are included.