We present a dataset of daily resolution climatic time series that has been compiled for the European Climate Assessment (ECA). As of December 2001, this ECA dataset comprises 199 series of minimum, maximum and/or daily mean temperature and 195 series of daily precipitation amount observed at meteorological stations in Europe and the Middle East. Almost all series cover the standard normal period 1961-90, and about 50% extends back to at least 1925. Part of the dataset (90%) is made available for climate research on CDROM and through the Internet (at http://www.knmi.nl/samenw/eca).A comparison of the ECA dataset with existing gridded datasets, having monthly resolution, shows that correlation coefficients between ECA stations and nearest land grid boxes between 1946 and 1999 are higher than 0.8 for 93% of the temperature series and for 51% of the precipitation series. The overall trends in the ECA dataset are of comparable magnitude to those in the gridded datasets.The potential of the ECA dataset for climate studies is demonstrated in two examples. In the first example, it is shown that the winter (October-March) warming in Europe in the 1976-99 period is accompanied by a positive trend in the number of warm-spell days at most stations, but not by a negative trend in the number of cold-spell days. Instead, the number of cold-spell days increases over Europe. In the second example, it is shown for winter precipitation between 1946 and 1999 that positive trends in the mean amount per wet day prevail in areas that are getting drier and wetter.Because of its daily resolution, the ECA dataset enables a variety of empirical climate studies, including detailed analyses of changes in the occurrence of extremes in relation to changes in mean temperature and total precipitation.

No abstract

No abstract

No abstract

Three and four parameter distributions of events and intervals based on the Kummer and Tricomi functions are shown to be statistically tractable. The Kummer distributions cover events less as well as more clustered than the Poisson while the Tricomi distributions have two and three parameter negative binomials as special cases. This flexibility and the moment relations between the parameters make them attractive in cases where the Poisson distribution proves inadequate. Fitzgerald (2000) has shown that statistically interesting results involving Tricomi functions arise when the rate parameter of a Poisson process is randomised by an inverted beta or a Tricomi exponential variate. These results are now developed into a flexible method for treating series of events more clustered than the Poisson and the intervals between them. The method offered there for computing Tricomi functions by statistical means is improved and a method of generating Tricomi exponential variates is also given. Then, using truncated beta distributions as randomisers, waiting-time distributions based on Kummer functions are derived; a statistical method of computing the Kummer function emerges and leads to another accurate method of computing the Tricomi function. IntroductionMoment relations between the parameters of the distributions are readily obtained as are the probabilities of a given number of occurrences and the statistical properties of the inter-occurrence times.A hyper-Poisson distribution based on Kummer functions and having four parameters is also treated as it covers series of events whether more or less clustered than the Poisson. In a very similar vein, but having a negative binomial as a special case, a four-parameter process based on the Tricomi function can be found. Mathematically and statistically both are generally as tractable as the three-parameter randomised versions but moment solutions in the four-parameter case present problems.With all these processes it is shown that, when the distribution for the exceedances possesses an inverse, the cumulative distribution functions of the return periods have the same analytic form as that of the occurrences and the values can be estimated by a method such as bisection.