A Confidence Index Approach Based on ERA-40 Data for Numerical Short Range Forecasts

• Main Authors: Thomas Prenosil, Anna Barbara Herold, Frank Müller
• Format: Article
• Language: English
• Published: Borntraeger 2014-09-01
• Series: Meteorologische Zeitschrift
• Subjects: predictability; analogue method; short range forecasting; confidence indexing
• Online Access: http://dx.doi.org/10.1127/0941-2948/2014/0546

Critical weather related missions increasingly rely on highly automated numerical products, even if only limited computer capacities are available to generate them. This holds true especially for military tactical decision aids but also for civil requirements from firebrigades, the Red Cross or technical relief organizations. With respect to inherent atmospheric indeterminateness, a systematic quality control of numerical input turns out to become more and more essential for the users. As an economical alternative to the complex and expensive ensemble prediction method, the German Bundeswehr Geoinformation Centre has decided in favour of an analogue approach called similar synoptic situations (3s), which is based on ECMWF's ERA-40 re-analysis archive. Similarity is defined by a special distance measure for synoptic fields. The typical range of interest is 2500km×2500km$2500\,\text{km}\times2500\,\text{km}$ in space with approximately one degree of horizontal resolution and up to 36 hours of forecast time. Historical 12, 24 and 36 hours ERA-40 forecast qualities are merged by 3s into a confidence index, indicating current anomalies of numerical quality versus monthly means in special areas of interest. As the results from the ERA-40 archive are used without any statistical adaption, this assessment is exclusively valid for trouble-free synoptic model runs in the short range. For a better understanding of the estimated anomalies in numerical forecast quality, the involved synoptic conditions are classified by a well established weather type classification. The overall method has been verified from 45 years of ERA-40 data and 10 years of GME forecasts from the Deutscher Wetterdienst. The 3s technique is highly flexible all over the globe with the exception of the tropics, because the present version includes the geostrophic approximation. At present, 3s runs operationally within four geographic areas: (1) Central Europe, (2) Kosovo with the Eastern Mediterranean, (3) the Mediterranean with Northern Africa plus the Middle East and (4) Afghanistan. Its typical operational behaviour is demonstrated by means of case studies in which different global models over Central Europe were used for 1999, 2000 and 2013.