The capability of some empirical algorithms to estimate surface rain-rate at mid-latitude basin scale from the Special Sensor Microwave Imager (SSM/I) data is analyzed. We propose three retrieval techniques based on a multivariate regression, a Bayesian maximum a posteriori inversion and on an artificial feed-forward Neural Network. Three algorithms available in literature are also included as benchmarks. The training data set is derived from coincident SSM/I images and half hourly rain-rate data obtained from a rain-gauge network, placed along the river Tiber basin in Central Italy, during 9 years (from 1992 to 2000). The work points out that an algorithm based on regression or Neural Network is a good estimator of low precipitation, while it tends to underestimate high rain rates. The best results have been achieved with the Bayesian method.
Empirical algorithms to retrieve surface rain-rate from Special Sensor Microwave Imager over a mid-latitude basin / Pulvirenti, Luca; Pierdicca, Nazzareno; Paolo, Castracane; D'Auria, Giovanni; Ciotti, ; Marzano, FRANK SILVIO; Basili,. - 3:(2002), pp. 1872-1874. (Intervento presentato al convegno IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2002)/24th Canadian Symposium on Remote Sensing tenutosi a TORONTO, CANADA nel JUN 24-28, 2002) [10.1109/igarss.2002.1026283].
Empirical algorithms to retrieve surface rain-rate from Special Sensor Microwave Imager over a mid-latitude basin
PULVIRENTI, Luca;PIERDICCA, Nazzareno;D'AURIA, Giovanni;MARZANO, FRANK SILVIO;
2002
Abstract
The capability of some empirical algorithms to estimate surface rain-rate at mid-latitude basin scale from the Special Sensor Microwave Imager (SSM/I) data is analyzed. We propose three retrieval techniques based on a multivariate regression, a Bayesian maximum a posteriori inversion and on an artificial feed-forward Neural Network. Three algorithms available in literature are also included as benchmarks. The training data set is derived from coincident SSM/I images and half hourly rain-rate data obtained from a rain-gauge network, placed along the river Tiber basin in Central Italy, during 9 years (from 1992 to 2000). The work points out that an algorithm based on regression or Neural Network is a good estimator of low precipitation, while it tends to underestimate high rain rates. The best results have been achieved with the Bayesian method.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
