Open AccessComparative skill assessment of consensus and physically based tercile probability seasonal precipitation forecasts for B razil
Author(s) -
Coelho Caio A. S.
Publication year - 2013
Publication title -
meteorological applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.672
H-Index - 59
eISSN - 1469-8080
pISSN - 1350-4827
DOI - 10.1002/met.1407
Subject(s) - consensus forecast , forecast skill , reliability (semiconductor) , quantitative precipitation forecast , precipitation , environmental science , climatology , computer science , forecast verification , econometrics , quality (philosophy) , statistics , meteorology , mathematics , geography , power (physics) , physics , quantum mechanics , geology , philosophy , epistemology
This study aimed to perform a comparative skill assessment of consensus and physically based tercile probability seasonal precipitation forecasts for B razil produced during the last decade. Two fundamental forecast attributes have been examined: discrimination and reliability. The discrimination assessment revealed that forecast quality is seasonally dependent and that consensus and physically based forecasts are complementary. During spring and summer consensus forecasts were generally found to have better discrimination ability than physically based forecasts. During autumn and winter physically based forecasts were found to have better discrimination ability than consensus forecasts. However, discrimination is a necessary but not sufficient forecast skill attribute, and therefore only provides indication of potential forecast quality provided forecasts are reliable (i.e. well calibrated). The analysis of tendency diagrams has revealed that both consensus and physically based forecasts suffer from systematic errors (biases) for the three forecast categories. Both forecasts under‐forecasted the below‐normal category and over‐forecasted the above normal category. This over‐forecasting feature was stronger for physically based forecasts when compared to consensus forecasts. The normal category was more severely over‐forecast for consensus forecasts when compared to physically based forecasts. The assessment through the computation of the reliability component of the B rier Score has revealed that consensus forecasts are better calibrated than CPTEC / AGCM physically based forecasts. Copyright © 2013 Royal Meteorological Society