Global tropospheric ozone column retrievals from OMI data by means of neural networks

摘要

In this paper, a new neural network (NN) algorithm to retrieve thetropospheric ozone column from Ozone Monitoring Instrument (OMI) Level 1bdata is presented. Such an algorithm further develops previous studies in orderto improve the following: (i) the geographical coverage of the NN, by extending itstraining set to ozonesonde data from midlatitudes, tropics and poles; (ii)the definition of the output product, by using tropopause pressureinformation from reanalysis data; and (iii) the retrieval accuracy, by usingancillary data (NCEP tropopause pressure and temperature profile, monthlymean tropospheric ozone column from a satellite climatology) to betterconstrain the tropospheric ozone retrievals from OMI radiances. The resultsindicate that the algorithm is able to retrieve the tropospheric ozone columnwith a root mean square error (RMSE) of about 5–6 DU in all the latitudebands. The design of the new NN algorithm is extensively discussed,validation results against independent ozone soundings andchemistry/transport model (CTM) simulations are shown, and othercharacteristics of the algorithm – i.e., its capability to detectnon-climatological tropospheric ozone situations and its sensitivity to thetropopause pressure – are discussed.