A nonlinear regression model in the time and space domain for radar rainfall nowcasting. (2015)

• Record Type: Journal Article
• Authors: Ramirez-Beltran, Nazario D.; Torres-Molina, Luz; Castro, Joan M.; Cruz-Pol, Sandra; Colom-UstÁ¡riz, José G.; Hosannah, Nathan
• Is Part Of: International journal of hydrology science and technology. Volume 5:Number 3(2015)
• Page Start: 208
• Page End: 232
• Publication Date: 2015
• Subjects: rainfall nowcasting -- rain cell centroid advection -- radar images -- extrapolation -- nonlinear regression modelling -- time domain -- spatial domain -- high resolution -- rainfall rate distribution -- rainfall rate intensity; Hydrology -- Periodicals; Water resources development -- Periodicals; Hydrology; 553.705
• Languages: English
• ISSNs: 2042-7808

Abstract

The introduced algorithm uses high spatial and temporal resolution radar data to predict the evolving rainfall rate distribution. The most likely future rainfall areas are estimated by tracking rain cell centroid advection in consecutive radar images. A nonlinear regression model varying in the time and space domain is proposed to predict the intensity of rainfall rate. It is assumed that the current radar reflectivity is a function of the previous reflectivity observed in surrounding areas with its centre on the location of a predicted pixel. It is also assumed that the ratio of reflectivity of a given pixel to the reflectivity of the convective core is a relevant predictor for rainfall estimation. The algorithm was validated against five rainfall events. The hit rate, false alarm ratio, and the Heike Skill Score were: 0.64, 0.27, and 0.61, respectively. The root mean squared error exhibits an average of 0.03 mm/hr.