Probabilistic hydrological forecasts forced with short-range numerical weather prediction ensembles for PREVENIR project.

LAGOS, MARINA; KAZIMIERSKI, LEANDRO; LÓPEZ, SEBASTIÁN; KAKINUMA, DAIKI; USHIYAMA, TOMOKI; MALDONADO, PAULA; DILLON, MARIA EUGENIA; AMEMIYA, ARATA; CASARETTO, GIMENA; RE, MARIANO; GARCIA, MARCELO; RUIZ, JUAN; GARCIA SKABAR, YANINA; MIYOSHI, TAKEMASA

virtual

Conferencia; 2nd WMO/WWRP Weather & Society Conference; 2024

Since 2022 Argentina and Japan are cooperating in the PREVENIR project, with the aim ofdeveloping an early warning system centered on the impacts caused by heavy rainfall andurban floods. We focus on two different argentinian regions: the Sarandí-Santo Domingo(SSD) Basin, with urban and flat characteristics, and the Suquía River Basin (SUQ), locatedin Córdoba Province's mountainous region. Therefore, two distinct hydrological models areused: a dual-drainage system, semidistributed model based on the Storm WaterManagement Model (SWMM) for SSD and the distributed model Rainfall-Runoff-Inundationmodel (RRI) for SUQ.One of the objectives of the project is to improve the hydrological forecasts using highresolution precipitation forecasts. We develop a 2-km resolution data assimilation systemwith the Local Ensemble Transform Kalman Filter (LETKF) to assimilate observation datafrom an automated weather station network and C-band doppler weather radars.Independent LETKF systems have been carried out with the models Weather Research andForecasting (WRF) and Scalable Computing for Advanced Library and Environment (SCALE), for both basins, obtaining 20-member 6-h ensemble forecasts. These systemswere used to produce analysis and forecasts on two selected flash flood case studies, one inSUQ (14th Dec 2018) and the other one in SSD (11th Oct 2019).In this work, the hydrological models were forced with the ensemble forecasted precipitation,to get an ensemble of hydrographs of flow discharge and water level. A verification wascarried out with in-situ observations of water level sensors in both basins (two in the mainstreams of SSD and two in the main rivers of SUQ). It was found that the assimilatedobservations in the numerical weather prediction systems produce an overall positive impactin the hydrological predictions. A preliminary evaluation of the quantification of theuncertainty in the hydrological forecasts will be presented at the conference.