No abstract

The interannual and decadal variability of seasonal precipitation (October through April) over Sardinia (Italy) is analysed for the years 1940–1995. It is shown that the area‐average seasonal rainfall total can vary greatly from one year to the next. The spatial correlation of Sardinia seasonal rainfall with that of neighbouring Mediterranean regions is generally high, though the correlation is less to the south of Sardinia, suggesting other teleconnection structures may operate there. Over a slightly shorter period (1946–1993), correlations between precipitation and large‐scale variables (sea level pressures (SLP), 500 hPa geopotential heights (Z500) and sea surface temperatures (SST)) are calculated and analysed. Clear connections between the North Atlantic Oscillation (NAO) and Sardinian precipitation are found and studied. Contrasting Atlantic storm tracks were found for the wet and dry seasons. In the wet years, the track is weaker over the North Atlantic (consistent with the NAO association), while over Europe there emerges a second minor storm track that enters the Mediterranean basin. In dry years this latter feature is less evident and is diverted to the lee of Sardinia, probably because of stronger intensity of the Azores High. Some apparent associations with other Northern Hemisphere remote teleconnections and large‐scale structures also emerge, including the North Pacific and across North Africa. Copyright © 2000 Royal Meteorological Society

A B S T R A C TIn this work, the main characteristics of COSMO-LEPS, the Limited-area Ensemble Prediction System developed in the framework of the COnsortium for Small-scale MOdelling, are presented. The present status of the system is shown with the description of the methodology and of the main upgrades which took place during its years of activity. The performance of COSMO-LEPS for the probabilistic prediction of precipitation is assessed in terms of both time-series and seasonal scores over a 7-yr period. A fixed number of stations are selected and observations are compared to short and early medium-range forecasts. Different verification indices are used to assess the skill of COSMO-LEPS and to identify the impact of system modifications on forecast skill. The different system upgrades are found to impact positively on COSMO-LEPS performance, with a gain of 2 d of predictability in the last 4 yr of operational forecasts. This holds when the skill of the system is assessed both for single events (e.g. precipitation surpassing a fixed threshold) and for multi-event situations. Scores for fixed forecast ranges but varying thresholds confirm increasingly better performance of the system. For a few seasons, the performance of COSMO-LEPS is also assessed in terms of probabilistic prediction of some upper-air variables. Then, the skill of COSMO-LEPS is compared to that of the global-ensemble system providing the boundaries, to identify the extent to which skill improvements may relate to those of the driving ensemble. Finally, the main streams of development for COSMO-LEPS system are discussed with future possible upgrades and methodology modifications.

No abstract

Abstract. Model Output Statistics (MOS) refers to a method of post-processing the direct outputs of numerical weather prediction (NWP) models in order to reduce the biases introduced by a coarse horizontal resolution. This technique is especially useful in orographically complex regions, where large differences can be found between the NWP elevation model and the true orography. This study carries out a comparison of linear and non-linear MOS methods, aimed at the prediction of minimum temperatures in a fruit-growing region of the Italian Alps, based on the output of two different NWPs (ECMWF T511-L60 and LAMI-3). Temperature, of course, is a particularly important NWP output; among other roles it drives the local frost forecast, which is of great interest to agriculture. The mechanisms of cold air drainage, a distinctive aspect of mountain environments, are often unsatisfactorily captured by global circulation models. The simplest post-processing technique applied in this work was a correction for the mean bias, assessed at individual model grid points. We also implemented a multivariate linear regression on the output at the grid points surrounding the target area, and two non-linear models based on machine learning techniques: Neural Networks and Random Forest. We compare the performance of all these techniques on four different NWP data sets. Downscaling the temperatures clearly improved the temperature forecasts with respect to the raw NWP output, and also with respect to the basic mean bias correction. Multivariate methods generally yielded better results, but the advantage of using non-linear algorithms was small if not negligible. RF, the best performing method, was implemented on ECMWF prognostic output at 06:00 UTC over the 9 grid points surrounding the target area. Mean absolute errors in the prediction of 2 m temperature at 06:00 UTC were approximately 1.2 • C, close to the natural variability inside the area itself.