TESI SPECIALISTICHE
Durata: tipicamente 6 mesi.
La tesi può essere scritta in inglese.
Si consiglia l'uso del sistema di scrittura LaTeX, scaricabile per MacOSX e Windows.
A seconda dell'argomento relatori, saranno i professori Domenico Solimini, Paolo Ferrazzoli, Giovanni Schiavon, Leila Guerriero, Fabio Del Frate.
Lo studente è pregato di prendere contatto con il tutor che compare cliccando sul titolo.

Design and validation of Neural Network algorithms for direct monitoring of Tropospheric Ozone from space
TECHNICAL DETAILS: In the last few decades interest has been growing in the monitoring of air quality from satellite platform. The ozone present in trosphere is a direct greenhouse gas and is involved in a number of pollution phenomena (fig. 1). Satellites allow a global and constant observation of the atmosphere. The recent UV/VIS satellite sensors are particularly suitable for tropospheric sounding. However, the present techniques to invert the satellite signal to obtain Tropospheric Ozone information are based on methodologies that need huge computational efforts (fig. 2). This thesis work aims at the development and subsequent validation of Neural Networks (NNs) algorithms for the retrieval of Tropospheric Ozone from space. NNs can compute the information required in near real time and with a little computational effort. At the moment some preliminar experiments have been made in the frame of our laboratory, and the encouraging results have been presented at international conferences [1], [2].

FRAMEWORK: This thesis work falls in an emerging sector of rEarth observation and geoinformation. The air quality issue has great environmental and social importance and a number of European projects plan to monitor and evaluate our cities' air quality. Moreover, the enhanced spatial resolution and the near-daily global coverage of the recently developed sensors (e.g. NASA-Aura OMI (fig.3) and ESA-Envisat SCIAMACHY), for the first time in the satellite history, allow an efficient observation of the local and short range air pollution phenomena from space. In this frame, Tor Vergata participates in the calibration/validation programme of the OMI sensor.

REQUIRED EXPERIENCE: No previous knowledge of atmospheric chemistry and physics is expected from the candidate. A medium-good knowledge of English language would be appreciated.

OFFERED BENEFITS: The selected candidate will learn the basic issues of atmospheric chemistry and physics, and the state-of-the-art in related fields. He/she will learn advanced methodologies in remote sensing, working in a stimulating international context. He/she will acquire medium-high IDL programming skills.

 

REFERENCES:
[1] P. Sellitto, A. Burini, F. Del Frate, D. Solimini and S. Casadio "Dedicated Neural Networks algorithms for direct estimation of tropospheric ozone from satellite measurements", Proceedings of IGARSS 2007, Barcelona, Spain, preprints available.
[2] F. Del Frate, P. Sellitto and D. Solimini "Design of Neural Network algorithms for the retrieval of tropospheric ozone from satellite data", Proceedings of Envisat Symposium 2007, Montreux, Switzerland, preprints available.
torna alla home