## NO_{2} Diurnal Variations

The UARS satellite describes a non-sun-synchronous orbit.
If the orbit were sun-synchronous, then the observation of
each latitude in each orbit would correspond to two fixed local times
(one for ascending and one for descending orbits). This is not the
case and each day corresponds to a drift in local time of
some 20 minutes.

In some 30 days, then, ascending and descending node observations
will cover all local times. In practice, the geometry is not that
simple, but most local times will be sampled at most latitudes.
If all variation in the signal were diurnal, these monthly observations
would directly show the local time dependence of the NO_{2}
signal. In practice, this local time variation is combined with other
time variation of the signal and is fuzzed out some.

If we take the results of a Grose model run and evaluate the field
at longitude zero over the course of a month or so, we see the basic
results:

## GROSE Model NO_{2}

We can fly the UARS satellite through this data and
simulate a satellite dataset with its coverage reduced due to the realistic
orbit track. We can then Kalman filter
the ascending and descending node profiles separately as well as local
time. Then each day's mapping at a given latitude corresponds to a given
local time. We can test the ability of the sampled asynoptic data to
show local time variation just as we can with the true synoptic data above:

## GROSE Model NO_{2} Mapping

So, we see that at least in the Grose model case, the diurnal variation
of NO_{2} is easily recovered from asynoptic data such as we can
obtain from a satellite. We can then apply this to the CLAES instrument
aboard the UARS satellite:

## CLAES NO_{2} Mapping

We see qualitatively that the actual and model data agree well in their
diurnal signal.