The spatial variability of turbulence above a forest has been examined. Two
measurement towers were erected 800 m apart within a heterogeneous mixed f
orest located in the north east of the Netherlands. The measurements of u*/
u were analysed and subsequently used to test a surface layer model. The mo
del simulated the magnitude of the measurements reasonably well, but measur
ed trends were not always reproduced by the model. The variable (du/dz)/u d
id not adapt as quickly to the new surface as u*/u. This is in agreement wi
th Schmid (1994), and can be explained by a local decrease in mixing length
. It is recommended to adapt the mixing length near a surface transition to
improve the accuracy of surface layer models of heterogeneous landscapes.