We present a general approach for estimating systematic and random err
ors in eddy correlation fluxes and flux gradients measured by aircraft
in the convective boundary layer as a function of the length of the f
light leg, or of the cutoff wavelength of a highpass filter. The estim
ates are obtained from empirical expressions for various length scales
in the convective boundary layer and they are experimentally verified
using data from the First ISLSCP (International Satellite Land Surfac
e Climatology Experiment) Field Experiment (FIFE), the Air Mass Transf
ormation Experiment (AMTEX), and the Electra Radome Experiment (ELDOME
). We show that the systematic flux and flux gradient errors can be im
portant if fluxes are calculated from a set of several short flight le
gs or if the vertical velocity and scalar time series are high-pass fi
ltered. While the systematic error of the flux is usually negative, th
at of the flux gradient can change sign. For example, for temperature
flux divergence the systematic error changes from negative to positive
about a quarter of the way up in the convective boundary layer.