Fire behaviour modelling has been based primarily on experiments involving
the measurement of a certain number of fires, where each variable is repres
ented by an average value per fire. The main objective of this study was to
examine if data collected from a microplot sampling design could be used t
o derive meaningful fire behaviour models. Three burns were conducted in lo
w shrubland of Erica umbellata Loefl., and Chamaespartium tridentatum (L.)
P. Gibbs in northeastern Portugal. Wind speed and aerial dead fuel moisture
content varied from 5 to 27 km/h and from 14 to 21%, respectively. Rate of
spread and flame length ranged from 0.3 to 14.1 m/min and from 0.2 to 3.1
m, respectively. Rate of fire spread could be described effectively in term
s of an empirical model with wind speed and fuel height as independent vari
ables. The coefficients that describe the effects of wind speed and fuel he
ight on fire propagation were consistent with published values for similar
fuel types. Flame length was strongly related to Byram's fireline intensity
. Microplot sampling is not free from methodological problems, which are di
scussed, but can be effectively used in field studies of fire behaviour.