On satellite observations of atmospheric composition and their interpretation

Abstract

Since more than 30 years satellites contribute significantly to our understanding of the composition of the atmosphere by performing global observations of atmospheric constituents from space. A recent addition to the series of Earth observing instruments is the Ozone Monitoring Instrument (OMI) that since October 2004 performs daily global measurements at high spatial resolution. The work presented in thesis focuses on spaceborne observations of NO2 and tropospheric aerosols, and the interpretation of the behavior of these constituents. NO2 plays an important role in the chemistry of the atmosphere due to its involvement in the catalytic destruction of ozone in the stratosphere, and by being a precursor of tropospheric ozone, linking NO2 to air quality and climate change. Aerosols also play an important role in chemistry and climate. For the DOAS-based retrieval of NO2 from OMI measurement data an accurate characterization of the OMI spectral slitfunction is essential. The spectral slitfunction was characterized with a novel method where the slitfunction for each wavelength and viewing angle was sampled by the spectrally narrow diffraction orders of an echelle grating, with wavelength increments 10 times smaller than the spectral resolution of OMI. The resulting parameterization of the spectral slitfunction is used in the retrieval of NO2 and other DOAS-based products from OMI. Tropospheric NO2 columns are retrieved from OMI measurements on an operational basis by the Dutch OMI NO2 (DOMINO) system. The DOMINO algorithm assimilates NO2 slant column in the TM4 chemistry transport model to estimate the stratospheric NO2 column. DOMINO data are available as a near-real time (within 3-4 hours after measurement) and as a consistent reprocessed offline dataset of collection 3, version 1.0.2. Based on the findings of validation studies involving DOMINO data, improvements to the DOMINO algorithm regarding surface albedo and a priori profile shape are identified . An extensive validation study shows that OMI stratospheric NO2 columns are consistent within 13% with ground-based observations from the SAOZ and NDACC network. The DOMINO product performs superior to the parallel existing Standard Product by capturing the dynamic variability of NO2 in the stratosphere, such as the daytime increase of stratospheric NO2 and the day-to-day variations in the NO2 field associated with the collapse of the Arctic Polar vortex. Analysis of the 5+ year OMI data record shows that OMI observes variations in stratospheric NO2 on a seasonal and multi-annual scale, e.g., the quasi biennial oscillation (QBO) and trends. The NO2 QBO signal exhibits a distinct interhemispheric asymmetry over the tropics, and is stronger over the Southern Hemisphere. There is good agreement between the Lauder data record and collocated OMI stratospheric NO2 observations, both showing a small increase of approximately +0.5% per decade for the timespan of the OMI mission (2004-2010). Observations from OMI and the spaceborne lidar CALIOP were used to characterize the around the world transport of an aerosol plume that was released by the intense Australian forest fires of December 2006. The plume crossed the Pacific in 5 days and completed the circumnavigation of the globe in 12 days. Estimates of the plume’s altitude from the OMI cloud retrieval algorithm indicate that the plume was injected into the tropopause region by pyro-convection, triggered by the combination of a passing cold front and the latent heat of the fires. The high altitude of the plume was confirmed by CALIOP that detected the plume at 11-15 km altitude as it passed over South America. Radiative transfer calculations indicate that the underestimation of the OMI plume height in comparison with CALIOP in a later stage of the plume’s transport is caused by photons scattered from lower-lying clouds that outshine the diluted plume. Simulations with TM4 agree best with OMI and CALIOP observations of the plume’s transport when a passive tracer is released at approximately 10 km altitude, to mimic the effect of pyro-convective lofting which is not simulated by the model.