|
Titel |
Photolysis rates for the POLYPHEMUS/DLR air quality model |
VerfasserIn |
Christoph Bergemann, Julian Meyer-Arnek |
Konferenz |
EGU General Assembly 2011
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 13 (2011) |
Datensatznummer |
250054243
|
|
|
|
Zusammenfassung |
Air pollution modelling with chemistry transport models requires accurate computation of
photolysis rates. These rates depend on various atmospheric factors. In this study we present
a novel approach to the computation of photolysis rates and analyze the importance of
different atmospheric parameters.
Given information on atmospheric composition, photolysis rates can in theory be
computed using radiative transfer models. These computations however tend to
be too computationally expensive for today’s hardware especially in operational
forecasting.
We demonstrate the usage of neural networks for the computation of photolysis
rates. We apply this method to the computation of clear-sky rates of photochemical
reactions from the RACM mechanism. In order to demonstrate the flexibility of the
approach we include dependence of photolyis rates on atmospheric columns of
ozone and NO2. This method is very general and can in principle be extended to
include other atmospheric variables or photochemical reactions from various chemical
mechanisms.
We present the results of training the neural network using sample data generated by the
radiative transfer model libRadtran. Once trained the neural network is computationally
efficient and can thus be included in operational air pollution modelling. We demonstrate this
approach by plugging the model into the POLYPHEMUS/DLR air quality model. The model
produces air quality simulations at regional and continental scale. In this study we assess the
impact of our method compared to photolysis data based on climatological ozone columns,
allowing us to estimate the impact of different atmospheric parameters on simulation results. |
|
|
|
|