Spectral analysis of atmospheric composition: application to surface ozone model–measurement comparisons
Models of atmospheric composition play an essential role in our scientific understanding of atmospheric processes and in providing policy strategies to deal with societally relevant problems such as climate change, air quality, and ecosystem degradation. The fidelity of these models needs to be a...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2016-07-01
|
| Series: | Atmospheric Chemistry and Physics |
| Online Access: | https://www.atmos-chem-phys.net/16/8295/2016/acp-16-8295-2016.pdf |
| Summary: | Models of atmospheric composition play an essential role in our scientific
understanding of atmospheric processes and in providing policy strategies to
deal with societally relevant problems such as climate change, air quality,
and ecosystem degradation. The fidelity of these models needs to be assessed
against observations to ensure that errors in model formulations are found
and that model limitations are understood. A range of approaches are
necessary for these comparisons. Here, we apply a spectral analysis
methodology for this comparison. We use the Lomb–Scargle periodogram, a
method similar to a Fourier transform, but better suited to deal with the
gapped data sets typical of observational data. We apply this methodology to
long-term hourly ozone observations and the equivalent model (GEOS-Chem)
output. We show that the spectrally transformed observational data show a
distinct power spectrum with regimes indicative of meteorological processes
(weather, macroweather) and specific peaks observed at the daily and annual
timescales together with corresponding harmonic peaks at one-half, one-third, etc., of
these frequencies. Model output shows corresponding features. A comparison
between the amplitude and phase of these peaks introduces a new comparison
methodology between model and measurements. We focus on the amplitude and
phase of diurnal and seasonal cycles and present observational/model
comparisons and discuss model performance. We find large biases notably for
the seasonal cycle in the mid-latitude Northern Hemisphere where the
amplitudes are generally overestimated by up to 16 ppbv, and phases are too
late on the order of 1–5 months. This spectral methodology can be applied to
a range of model–measurement applications and is highly suitable for
Multimodel Intercomparison Projects (MIPs). |
|---|---|
| ISSN: | 1680-7316 1680-7324 |