Comparison of sub-daily precipitation extremes in observed data and regional climate model simulations

0471615 - ÚFA 2017 GB eng A - Abstrakt
Kyselý, Jan
Comparison of sub-daily precipitation extremes in observed data and regional climate model simulations.
Programme and Abstracts: 10th International Conference on CFE and 9th International Conference of the ERCIM Working Group on CMStatistics. London: CFE and CMStatistics networks, 2016 - (Fernandez, B.; Rodriguez, G.). s. 110. ISBN 978-9963-2227-1-1.
[CFE-CMStatistics 2016. 09.12.2016-11.12.2016, Seville]
Institucionální podpora: RVO:68378289
Klíčová slova: regional climate models (RCMs) * precipitation extremes * reanalyses
Kód oboru RIV: DG - Vědy o atmosféře, meteorologie
http://cmstatistics.org/CMStatistics2016/docs/BoA%20CFE-CMStatistics%202016.pdf?20161110230105

The study compares characteristics of observed sub-daily precipitation extremes with those simulated by HadRM3 and RCA4 regional climate
models (RCMs) driven by reanalyses. The observed warm-season (May-September) maxima of short-duration (1 h, 2 h and 3 h) amounts show one
diurnal peak in the afternoon, which is simulated reasonably well by RCA4, although the peak occurs too early in the model. HadRM3 provides an
unrealistic diurnal cycle with a night-time peak and an afternoon minimum coinciding with the observed maximum for all three ensemble members,
which suggests that convection is not captured realistically. Distorted relationships of the diurnal cycles of hourly precipitation to daily maximum
temperature in HadRM3 further evidence that underlying physical mechanisms are misrepresented in this RCM. Goodness-of-fit tests indicate that
the Generalized Extreme Value (GEV) distribution is an applicable model for both observed and RCM-simulated precipitation maxima. However,
the RCMs are not able to capture the range of the shape parameter estimates of distributions of short-duration precipitation maxima realistically,
leading to either too many or too few grid boxes in which the shape parameter corresponds to a heavy tail. This means the distributions of maxima
of sub-daily amounts are distorted in the RCM-simulated data and do not match reality well.
Trvalý link: http://hdl.handle.net/11104/0268972