Fourth Study Conference on BALTEX Scala Cinema Gudhjem

- 122 -
Interannual Changes in Heavy Precipitation in Europe from Station and
NWP Data
Olga Zolina 1 , Alicie Kapala 1 , Clemens Simmer 1 , Sergey Gulev 2
1 Meteorologisches Institut, Universitaet Bonn, 53121 auf dem Huegel 20, Bonn, Germany, olga.zolina@uni-bonn.de,
2 P.P.Shirshov Institute of Oceanology, Moscow, Russia, gul@sail.msk.ru,
1. Introduction
We present an analysis of the interannual to decadal
variability in heavy European precipitation using station
daily data as well as precipitation estimates from different
reanalyses. The main question to address is the robustness of
the variability patterns in different numerical weather
prediction (NWP) data sets and their comparability with
those derived from the station data.
2. Data
Station data consist of inputs from different collections
(Royal Netherlands Meteorological Institute, National
Climate Data Center, German Metoffice, Russian Metoffice)
and covers the 20 th century period. Altogether we assembled
more than 2000 stations, roughly 500 of which provide quite
a dense coverage of the last five decades. Reanalyses data
were taken from the four major reanalyses (NCEP/NCAR
Reanalysis versions 1 and 2, ECMWF Reanalyses ERA15
and ERA40). These data cover the periods from 15 to 55
years and overlap each other during the period 1979-1993.
Figure 1 shows spatial distribution and periods of
observations at European stations used in this study.
-20 -10 0 10 20 30 40 50 60
70
70
60
50
40
30
30
-20 -10 0 10 20 30 40 50 60
0 0 to - 30 30
30 - to 50 50
50 - to 100
100 - to 200 200years
Figure 1: Spatial distribution of European stations used in
this study with corresponding observational periods.
3. Methods
Methods of data preprocessing included the quality control,
homogenization and co-location of data from different
sources. Analysis of precipitation extremes is presented in
terms of the parameters of the probability density functions
as well as different percentiles of precipitation distributions.
Occurrences of heavy precipitation were quantified through
the scale and shape parameters of the gamma-distribution,
whose applicability has been tested using the k-s test. From
the distribution characteristics we derived the other
statistical parameters (percentiles, return values, etc.).
60
50
40
4. Results
The mean statistical characteristics exhibit patterns which
are qualitatively comparable to each other, but show quite
strong quantitative differences. The highest probability of
extreme and heavy precipitation (the largest shape
parameter) occurs in ERA40 while the lowest probability
is exhibited by ERA15. The analysis of linear trends in
ERA40 and NCEP1 for a 43-year period shows that
despite the similarity of the trends in mean seasonal
precipitation, trends in shape and scale parameters may
exhibit different signs. This holds for the winter shape
parameter in Eastern Europe and for the summer scale
parameter in the Alpine region. In Figure 2 we present, as
example, the winter trend patterns in the shape parameter
derived from ERA40 and NCEP1 reanalyses.
-20 -10 0 10 20 30 40 50 60
70
70
60
50
40
30
30
-20 -10 0 10 20 30 40 50 60
-20 -10 0 10 20 30 40 50 60
70
70
60
50
40
(A)
(B)
ERA40
JFM
NCEP1
JFM
30
30
-20 -10 0 10 20 30 40 50 60
Figure 2. Linear trends in the shape parameter (per 100
years) derived from ERA40(A) and NCEP1 (B) reanalyses
for winter for the period 1958-2001. Grey shading
indicates 95% significance (Student t-test).
Both NWP products show significantly positive trends
over Western Scandinavia, Northern United Kingdom and
Caucasus and significantly negative trends in Western
continental Europe for the shape parameter. At the same
time trends over the Central and Eastern European regions
exhibit drastic differences between ERA40 and NCEP1
reanalyses. ERA40 shows significantly positive trends
here, while NCEP1 diagnoses negative tendencies over the
last 4 decades.
Comparability of the interannual variability patterns was
analysed in terms of common EOFs. For both shape and
60
50
40
60
50
40