Sonic anemometer observations were made 10 m above ground level for a perio
d of 1 yr. From these data, Eulerian autocorrelation functions were compute
d for the horizontal and vertical wind velocity fluctuations for low wind s
peeds. Although the autocorrelation function for the vertical velocity comp
onent exhibited the well-known exponential form, the function for the horiz
ontal components of the wind vector showed a negative loop for all stabilit
y classes at low wind speeds. This result might be an effect of low-frequen
cy meandering of the flow. Observations of the standard deviations of the v
ertical wind component confirmed the proportionality with the friction velo
city, though with a slightly lower constant of proportionality than has bee
n found by other authors. A Lagrangian dispersion model (LDM) with random t
ime steps and a negative intercorrelation parameter rho (u,nu) for the hori
zontal wind components was used to take the first of the above-mentioned fi
ndings into account. In a simple test case, it could be shown that using a
negative tail in the autocorrelation function for the horizontal wind fluct
uations in an LDM results in larger plume spreads as if the usual exponenti
al law were used. This model characteristic is in agreement with enhanced d
ispersion in low-wind situations as found by different authors earlier. Bec
ause the model reduces to the Langevin equation for rho (u,nu) = 0.9, it ha
s the advantage that it can be used for all wind speeds by simply adjusting
the intercorrelation parameter. Last, the model was tested against field e
xperiment data gathered by the Idaho National Engineering Laboratory during
stable, low-wind conditions. The results with the new method for these exp
eriments are very promising in comparison with methods used by other author
s earlier.