In order to describe the production of nitrogen oxides (NOx) by lightning i
n chemistry-transport models the spatial and temporal distribution of this
NOx, source needs to be specified. Various meteorological model parameters
can be used as a proxy for specifying this distribution. In order to determ
ine the most suitable parameter as a proxy for lightning, we have correlate
d different meteorological quantities from the ECMWF model, such as convect
ive precipitation and cloud top height, with ground-based lightning observa
tions made during the EULINOX project. Convective precipitation gave the be
st correlation for summer conditions over Europe. Using the established rel
ationship between convective precipitation and lightning intensity, we have
tested different vertical distributions of the NOx, lightning source in th
e global chemistry-transport model TM3. The model simulated NO fields have
been compared with the NO observations made during the 1998 EULINOX and the
1997 POLI-NAT/SONEX campaigns. We found that a prescribed lightning NOx pr
ofile gave the best agreement with the observations. The observations cover
ed various conditions: typical background situations, inside thunderstorms,
and the outflow of thunderstorms of various ages. The model underestimated
the NO concentrations in cases of fresh lightning NOx, likely due to its i
nsufficient spatial and temporal resolution, but under most other circumsta
nces the new parameterisation performed reasonably well, and is a clear imp
rovement over the previously used parameterisation based on cloud top heigh
ts. (C) 2001 Elsevier Science Ltd. All rights reserved.