Authors: Nazario D. Ramirez-Beltran; Joan M. Castro; Jorge González
Addresses: Department of Industrial Engineering, University of Puerto Rico, P.O. Box 9030, Mayagüez, PR 00681, USA ' Department of Civil Engineering, University of Puerto Rico, P.O. Box 9040, Mayagüez, PR 00681, USA ' Department of Mechanical Engineering, The City College of New York, 140 the Convent Ave., New York, NY 10031, USA
Abstract: A nowcasting algorithm is introduced in this work. The algorithm uses radar rainfall data to estimate the rainfall field by first predicting the most likely rain areas and then predicting the expected rainfall rate intensity in each rain pixel. Radar data of three tropical storms were used to validate the proposed approach. Prediction errors show that the proposed algorithm is a potentially useful tool for prediction of spatial rainfall variability for short time intervals. The outputs from the rainfall algorithm can be assimilated by a numerical hydrological/inundation mapping model to predict flash-flood events. Although radar reflectivity was used to illustrate the performance of the algorithm, an extension to satellite rainfall data is a straightforward implementation.
Keywords: nowcasting; spatial distribution; temporal distribution; rainfall rate; forecasting; radar; motion vector; rain pixels; precipitation rate; tropical storms; hydrological mapping; inundation mapping; modelling; flooding prediction; flash floods.
International Journal of Water, 2015 Vol.9 No.4, pp.315 - 333
Received: 22 Mar 2014
Accepted: 18 Aug 2014
Published online: 02 Oct 2015 *