Mediterranean Forecasting System: forecast and analysis assessment through skill scores

摘要

This paper describes the first evaluation of thequality of the forecast and analyses produced at the basinscale by the Mediterranean ocean Forecasting System (MFS)(http://gnoo.bo.ingv.it/mfs). The system produces short-termocean forecasts for the following ten days. Analyses are producedweekly using a daily assimilation cycle. The analysesare compared with independent data from buoys, whereavailable, and with the assimilated data before the data areinserted. In this work we have considered 53 ten days forecastsproduced from 16 August 2005 to 15 August 2006.The forecast skill is evaluated by means of root meansquare error (rmse) differences, bias and anomaly correlationsat different depths for temperature and salinity, computingdifferences between forecast and analysis, analysisand persistence and forecast and persistence. The Skill Score(SS) is defined as the ratio of the rmse of the difference betweenanalysis and forecast and the rmse of the differencebetween analysis and persistence. The SS shows that at 5 and30m the forecast is always better than the persistence, but at300m it can be worse than persistence for the first days ofthe forecast. This result may be related to flow adjustmentsintroduced by the data assimilation scheme. The monthlyvariability of SS shows that when the system variability ishigh, the values of SS are higher, therefore the forecast hashigher skill than persistence.We give evidence that the error growth in the surface layersis controlled by the atmospheric forcing inaccuracies, while at depth the forecast error can be interpreted as due to thedata insertion procedure. The data, both in situ and satellite,are not homogeneously distributed in the basin; therefore, thequality of the analyses may be different in different areas ofthe basin.