A SEVIRI rainfall estimation technique based on Artificial Neural Networks (ANN) has recently been developed at the University of Ferrara. The algorithm makes use of the multispectral capabilities and the increased resolution of the new sensor on board the Meteosat Second Generation satellites. The present version of the algorithm provides rain maps in five rainrate levels, at 5 km of spatial resolution and 15 minutes of time resolution. It is tested and validated over U.K. area and for summer and winter season by using the Met Office Nimrod radar precipitation maps as ground truth. Performance are evaluated by calculating the Equitable Threat Score (ETS) and BIAS for rain – no rain classification. The algorithm performances strongly depend upon the characteristics of the training dataset, such as season, climatic regime and latitude: to export the technique on a given target area is thus highly recommended to train the ANN with data from that area. In the frame of an EU Community initiative programme INTERREG III B ARCHIMED the focus of the project RiskMed are the Central and Western Mediterranean areas, where a reliable, high-resolution, ground radar based precipitation dataset, large enough to train the ANN, is not available. Nevertheless, almost the half of the domain is covered by the TRMM-PR overpasses. The TRMM product 2A25, in this first approach, is therefore used to assess the possibility to re-calibrate the U.K. trained algorithm by using a limited PR-based data set. TRMM-PR data collected over six months are used to train the ANN over the areas for two different seasons (winter and summer), and the performances are discussed.

On the use of a SEVIRI-based statistical rainfall classification technique calibrated with TRMM-PR over southern Mediterranean

PORCU', Federico;PRODI, Franco
2007

Abstract

A SEVIRI rainfall estimation technique based on Artificial Neural Networks (ANN) has recently been developed at the University of Ferrara. The algorithm makes use of the multispectral capabilities and the increased resolution of the new sensor on board the Meteosat Second Generation satellites. The present version of the algorithm provides rain maps in five rainrate levels, at 5 km of spatial resolution and 15 minutes of time resolution. It is tested and validated over U.K. area and for summer and winter season by using the Met Office Nimrod radar precipitation maps as ground truth. Performance are evaluated by calculating the Equitable Threat Score (ETS) and BIAS for rain – no rain classification. The algorithm performances strongly depend upon the characteristics of the training dataset, such as season, climatic regime and latitude: to export the technique on a given target area is thus highly recommended to train the ANN with data from that area. In the frame of an EU Community initiative programme INTERREG III B ARCHIMED the focus of the project RiskMed are the Central and Western Mediterranean areas, where a reliable, high-resolution, ground radar based precipitation dataset, large enough to train the ANN, is not available. Nevertheless, almost the half of the domain is covered by the TRMM-PR overpasses. The TRMM product 2A25, in this first approach, is therefore used to assess the possibility to re-calibrate the U.K. trained algorithm by using a limited PR-based data set. TRMM-PR data collected over six months are used to train the ANN over the areas for two different seasons (winter and summer), and the performances are discussed.
9789291100798
telerilevamento; precipitazioni; TRMM
File in questo prodotto:
Non ci sono file associati a questo prodotto.

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: http://hdl.handle.net/11392/533081
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact