final_program_abstracts[1]

11 IMSC Session Program
New indices for monitoring changes in heatwaves and
extended cold spells
Friday - Parallel Session 7
Blair Trewin
National Climate Centre, Australian Bureau of Meteorology, Melbourne, Australia
Changes in temperature extremes have attracted considerable attention in recent years,
and numerous analyses have been published of changes in the occurrence of singleday
extremes. While the impact of multi-day extreme events, such as the European
heatwave of 2003, is well-known, there have been limited analyses of changes in such
events. A major contributor to this gap in analyses has been the lack of suitable,
widely-applicable indices which can be used to define multi-day heatwaves (or cold
spells), with most indices proposed either specific to a particular location, subject to
qualities which make them difficult to analyse (e.g. many years with zero or null
values), or both. The percentile-based approach widely used for indices of single-day
extremes breaks down for multi-day events as the probability of the number of
consecutive days above (below) a threshold is a product of the autocorrelation
structure of the daily time series as well as the frequency distribution of daily values.
A new suite of indices is proposed to address this question. These indices are defined
in each year, for a period of N days, as:
(a) the highest (lowest) temperature T such that there are at least N consecutive days
above (below) T in the year concerned;
(b) the mean temperature of the N-day period defined by (a) above; and
(c) the highest (lowest) value of the N-day running mean of temperature during the
year.
These indices are defined in every year which satisfy data completeness criteria, and
as they are defined in terms of °C, changes in their values can be readily compared
with changes in mean temperatures at the same location. They can be applied to daily
maximum, minimum or mean temperatures. An analysis of data from a range of
regions, including Australia, Europe, Canada and Antarctica, found that, in addition to
the absence of null values, the frequency distribution of these indices did not
significantly differ from the normal (Gaussian) at a substantial proportion of stations,
with only 7-15% of stations (depending on the index) showing skewness significant at
the 95% level, compared with 25-100% for other indices previously used in the
literature. The new indices also appear to be effective in identifying the most extreme
events such as the August 2003 heatwave in France. One drawback of the new indices
is that their definition is relatively complex and difficult to communicate to a lay
audience.
An analysis of trends over the period 1961-2007 for 5-day periods for index (a) found
strong positive trends in most cases. For Australian stations, trends ranged from
+0.05°C/decade for high maxima to +0.25°C/decade for low minima, while for
European stations trends ranged from +0.33°C/decade for high maxima to
+0.60°C/decade for low minima.
These indices may also be used to investigate the relationship between extended
heatwaves (or cold spells) and broad-scale drivers of climate variability. An
Abstracts 342