Prévision quantitative des précipitations journalières par une technique de recherche de journées antérieures analogues : optimisation du critère d'analogie

摘要: For quantitative precipitation forecasting, model output statistics (MOS) are very useful. We have tried to improve a MOS based on selection of analogue and past situations. Our study concerns the criterion: its form, what variables introduce (predictors), among those available (700 1 000 hPa geopotential fields) their format (principal components or PCs, gridded data, etc.). Initially, an Euclidean distance using predictors as PCs was used. Its replacement by Teweles-Wobus score, with known in grid format, gave significantly improved results.