Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
193<br />
On the validation of RCMs in terms of reproducing the annual cycle<br />
Tomas Halenka, Petr Skalak and Michal Belda<br />
Department of Meteorology and Environment Protection, Charles University; tomas.halenka@mff.cuni.cz<br />
1. Climate models and annual cycle<br />
There are many aspects of the validation of climate models.<br />
In addition to standard statistical characteristics a more indepth<br />
analysis of annual cycle performance can provide<br />
more information on ability of the models to reproduce<br />
properly the physical processes which strongly affect the<br />
behavior of climate parameters during the year. Global<br />
Circulation Models (GCMs) can reproduce climate features<br />
on large scales, but their accuracy decreases when<br />
proceeding from continental to regional and local scales<br />
because of the lack of resolution and thus on the regional<br />
scale they are very often rather poor in reproducing the<br />
annual cycle. The more detail analysis of 15 RCMs used in<br />
EC FP6 IP ENSEMBLES in ERA 40 driven experiment on<br />
25 km resolution for the period of 1961-2000 in different<br />
PRUDENCE regions presents the comparison of the models<br />
and their validation in terms of annual cycle reproduction.<br />
While for the temperature the performance of the models is<br />
mostly very good and quite consistent, there are some<br />
models with rather significant problems in some regions in<br />
reproducing annual cycle of precipitation.<br />
2. ENSEMBLES RCMs assessment in terms of<br />
annual cycle<br />
The quality of reproducing the annual course in the model<br />
simulations is believed to be good indication of quality of<br />
different atmospheric processes description which affects<br />
the overall model performance. However, to compare the<br />
annual course patterns with those based on observational<br />
data is not simple and unambiguous task, for the purpose of<br />
the use in the weighting methods in ENSEMBLES project<br />
some objective method is required. The problem of the<br />
annual course analysis is that we cannot use just one simple<br />
characteristic as bias, basically at least three patters are of<br />
importance, i.e. in addition to the bias it is the amplitude and<br />
shift in period, either as a whole or partly in some seasons.<br />
Taylor (2001) presented so called Taylor diagram, which is<br />
used for presenting the data in terms of RMS error, standard<br />
deviation and correlation and show just the analysis of<br />
models simulations with emphasis on annual cycle.<br />
Actually, although these characteristics are not completely<br />
independent, we can see the analogue between the bias and<br />
RMS error, as well as in between amplitude and standard<br />
deviation, correlation being good representation of the shifts<br />
in the annual course. For purpose of the weighting we<br />
propose to use the score index<br />
where R 0 is the maximum correlation attainable. For<br />
simplicity we used maximum value R 0 =1. Parameter R is the<br />
correlation coefficient with respect to the observation dataset<br />
used for comparison, m is standard deviation of the<br />
model results m normalized by standard deviation of the<br />
observational dataset used for validation . We have made<br />
calculations both for first degree formulation as above and<br />
fourth degree one discussed in Taylor (2001), which reads<br />
Additionally, we have tested the second degree<br />
formulation as well, basically the higher degree penalizes<br />
the lower correlation, for computations in this analysis<br />
seems to be usefull to use the 4 th order formulation for<br />
temperature to resolve the differences between otherwise<br />
very high scores, for precipitation the scores are not so<br />
high anyway and the first degree is enough to resolve the<br />
performance of the individual models. All the calculations<br />
were performed for individual PRUDENCE regions, quite<br />
significant differences appears between the models for<br />
some regions. To compare these score annual cycles for<br />
temperature and precipitation are presented in Fig. 1 and<br />
2, respectively, using ENSEMBLES gridded data (E-obs,<br />
Haylock et al., 2008). Original CRU databases are shown<br />
for comparison as well, although for precipitation there are<br />
some differences between the climatologies the resulting<br />
scores are not so significantly affected.<br />
It should be mentioned that this method is validating the<br />
general performance of the models anyway, for<br />
temperature, where the variance of the complete monthly<br />
data is usually nearly of order higher than interannual<br />
variance (year by year) of data for individual months, this<br />
performance well correspond with the performance of<br />
annual cycle. The results for precipitation are not so clear,<br />
the variance of all the time series is basically the similar as<br />
the interannual variability of individual monthly data so<br />
that the overall performance and quality of annual cycle<br />
cannot be well resolved by this method. Proper choice of<br />
R 0 for precipitation could be of great importance for<br />
further application of the method.<br />
Annual courses of temperature as well as precipitation will<br />
be presented for individual models and regions. Taylor’s<br />
diagrams will show the models validation against E-Obs.<br />
Acknowledgements<br />
This work is supported in framework of EC FP6 IP<br />
ENSEMBLES. Some contributions to the tasks are<br />
supported from local sources as well under Research Plan<br />
of MSMT, No. MSM 0021620860. We acknowledge the<br />
E-Obs dataset from the EU-FP6 project ENSEMBLES<br />
(http://www.ensembles-eu.org) and the data providers in<br />
the ECA&D project (http://eca.knmi.nl).<br />
References<br />
Haylock, M.R., N. Hofstra, A.M.G. Klein Tank, E.J. Klok,<br />
P.D. Jones, M. New, A European daily high-resolution<br />
gridded dataset of surface temperature and<br />
precipitation. J. Geophys. Res (Atmospheres), 113,<br />
D20119, doi:10.1029/2008JD10201, 2008.<br />
Taylor, Karl E., Summarizing multiple aspects of model<br />
performance in single diagram, J. Geophys. Res., 106,<br />
D7, 7183--7192, 2001.