Climate changes and atmospheric pollution are currently topical issues given their possible dramatic effects from the health, social and economical points of view. Assessing the causes and possible adaptation/mitigation strategies is a challenge in modern science. To understand and quantify the anthropic role in such changes is of a particular interest to depict future scenarios and to warn politicians about local and global intervention in emissions control. Ozone is one of the most important trace gases in the Earth's atmosphere. It is mainly present in the stratosphere, with only 10% in the troposphere. Despite its small amount, (2-7) 10ô 3 % in molar fraction, the solar radiation at wavelengths below 310 nm does not reach the Earth surface because of the large absorption cross sections characterizing ozone molecules at those wavelengths. Variations in the stratospheric ozone content may play a dramatic role in a possible increase of the surface UV radiation. The discovery of the Antarctic ozone hole, i.e. a considerable reduction of ozone in the polar stratosphere, was a dramatic evidence of the effects of anthropogenic emissions on the ozone layer. Human activity is likely responsible also for tropospheric ozone enhancements caused by the photochemistry associated to industrial emissions involving ozone precursors as the nitrogen dioxide. The effect of these variations at lower altitudes, with respect to background values, have been estimated to be the third largest source of the greenhouse effect. To support interpretation of the atmospheric phenomena, as well as interactions with the oceans and the ground, a constant and systematic monitoring of several atmospheric parameters, and with a good spatial coverage, is crucial. In this framework, global and systematic space-based observations of the atmospheric composition and its variations in time and space play a major role. Satellite measurements of atmospheric parameters has a proven and recognized effectiveness for such tasks. The advantage of atmospheric sounding performed from space, with respect to ground based techniques, lies in the very high number of available measurements per day and in the global coverage of the Earth, allowing for a detailed and continuous investigation of the atmospheric state. A number of different techniques are available, using different instruments, bands and viewing geometries. For all of them, a major problem is related to the intrinsically indirect nature of the measurements, as they result from the interaction between the electromagnetic radiation and the atmospheric constituents. The retrieval phase requires the solution of an inverse problem, which is never trivial and can be computationally very intensive, especially for this kind of nonlinear problems. A significant concurrent requirement is an adequate spatial resolution. Horizontal resolution is very hard to achieve by limb measurements, while it can be attained by nadir observations. Nadir measurements, however, can have poor vertical resolutions, and the inversion problem can be particularly computationally expensive. In this thesis we present novel approaches to the inversion of the nadir UV/VIS satellite Earth's radiance spectra for the retrieval of height resolved ozone information. The considered platforms are ESA EnviSat-SCIAMACHY and NASA-Aura OMI, which are particularly suited for these tasks owing to their combined high spectral and spatial resolutions. Both ozone concentration profiles and tropospheric ozone column are retrieved by means of NNs algorithms. NNs are made of interconnected elementary processing units, called neurons, and can learn from a training dataset; they were proven to be robust on systematic errors and calibration uncertainties on the input measurements vector, and they are likely to work better than OE with respect to cloudy scenarios or in presence of significant aerosols burdens. Once a net is trained it can perform retrievals in real-time.

Sellitto, P. (2009). Neural networks algorithms for the estimation of atmospheric ozone from Envisat-SCIAMACHY and Aura-OMI measurements.

Neural networks algorithms for the estimation of atmospheric ozone from Envisat-SCIAMACHY and Aura-OMI measurements

SELLITTO, PASQUALE
2009-03-04

Abstract

Climate changes and atmospheric pollution are currently topical issues given their possible dramatic effects from the health, social and economical points of view. Assessing the causes and possible adaptation/mitigation strategies is a challenge in modern science. To understand and quantify the anthropic role in such changes is of a particular interest to depict future scenarios and to warn politicians about local and global intervention in emissions control. Ozone is one of the most important trace gases in the Earth's atmosphere. It is mainly present in the stratosphere, with only 10% in the troposphere. Despite its small amount, (2-7) 10ô 3 % in molar fraction, the solar radiation at wavelengths below 310 nm does not reach the Earth surface because of the large absorption cross sections characterizing ozone molecules at those wavelengths. Variations in the stratospheric ozone content may play a dramatic role in a possible increase of the surface UV radiation. The discovery of the Antarctic ozone hole, i.e. a considerable reduction of ozone in the polar stratosphere, was a dramatic evidence of the effects of anthropogenic emissions on the ozone layer. Human activity is likely responsible also for tropospheric ozone enhancements caused by the photochemistry associated to industrial emissions involving ozone precursors as the nitrogen dioxide. The effect of these variations at lower altitudes, with respect to background values, have been estimated to be the third largest source of the greenhouse effect. To support interpretation of the atmospheric phenomena, as well as interactions with the oceans and the ground, a constant and systematic monitoring of several atmospheric parameters, and with a good spatial coverage, is crucial. In this framework, global and systematic space-based observations of the atmospheric composition and its variations in time and space play a major role. Satellite measurements of atmospheric parameters has a proven and recognized effectiveness for such tasks. The advantage of atmospheric sounding performed from space, with respect to ground based techniques, lies in the very high number of available measurements per day and in the global coverage of the Earth, allowing for a detailed and continuous investigation of the atmospheric state. A number of different techniques are available, using different instruments, bands and viewing geometries. For all of them, a major problem is related to the intrinsically indirect nature of the measurements, as they result from the interaction between the electromagnetic radiation and the atmospheric constituents. The retrieval phase requires the solution of an inverse problem, which is never trivial and can be computationally very intensive, especially for this kind of nonlinear problems. A significant concurrent requirement is an adequate spatial resolution. Horizontal resolution is very hard to achieve by limb measurements, while it can be attained by nadir observations. Nadir measurements, however, can have poor vertical resolutions, and the inversion problem can be particularly computationally expensive. In this thesis we present novel approaches to the inversion of the nadir UV/VIS satellite Earth's radiance spectra for the retrieval of height resolved ozone information. The considered platforms are ESA EnviSat-SCIAMACHY and NASA-Aura OMI, which are particularly suited for these tasks owing to their combined high spectral and spatial resolutions. Both ozone concentration profiles and tropospheric ozone column are retrieved by means of NNs algorithms. NNs are made of interconnected elementary processing units, called neurons, and can learn from a training dataset; they were proven to be robust on systematic errors and calibration uncertainties on the input measurements vector, and they are likely to work better than OE with respect to cloudy scenarios or in presence of significant aerosols burdens. Once a net is trained it can perform retrievals in real-time.
4-mar-2009
A.A. 2008/2009
Geoinformazione
21.
neural networks; ozone
satellite
Settore ING-INF/03 - TELECOMUNICAZIONI
English
Tesi di dottorato
Sellitto, P. (2009). Neural networks algorithms for the estimation of atmospheric ozone from Envisat-SCIAMACHY and Aura-OMI measurements.
File in questo prodotto:
File Dimensione Formato  
PhD_Thesis__SELLITTO.pdf

accesso aperto

Dimensione 5.1 MB
Formato Adobe PDF
5.1 MB Adobe PDF Visualizza/Apri
PhD_Thesis__SELLITTO.pdf

accesso aperto

Dimensione 5.09 MB
Formato Adobe PDF
5.09 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2108/819
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact