Evaluation of a global ensemble flood prediction system in Peru

Konstantinos Bischiniotis*, Bart van den Hurk, Ervin Zsoter, Erin Coughlan de Perez, Manolis Grillakis, Jeroen C.J.H. Aerts

*Corresponding author for this work

Research output: Contribution to JournalArticleAcademicpeer-review


Flood early warning systems play a more substantial role in risk mitigation than ever before. Hydrological forecasts, which are an essential part of these systems, are used to trigger action against floods around the world. This research presents an evaluation framework, where the skills of the Global Flood Awareness System (GloFAS) are assessed in Peru for the years 2009–2015. Simulated GloFAS discharges are compared against observed ones for 10 river gauges. Forecasts skills are assessed from two perspectives: (i) by calculating verification scores at every river section against simulated discharges and (ii) by comparing the flood signals against reported events. On average, river sections with higher discharges and larger upstream areas perform better. Raw forecasts provide correct flood signals for 82% of the reported floods, but exhibit low verification scores. Post-processing of raw forecasts improves most verification scores, but reduces the percentage of the correctly forecasted reported events to 65%.

Original languageEnglish
Pages (from-to)1171-1189
Number of pages19
JournalHydrological Sciences Journal
Issue number10
Publication statusPublished - 27 Jul 2019


  • bias-correction
  • early warning
  • ensemble streamflow predictions
  • flood
  • forecast
  • risk


Dive into the research topics of 'Evaluation of a global ensemble flood prediction system in Peru'. Together they form a unique fingerprint.

Cite this