Climate Sense

[

] 185

bserving

, P

redicting

and

rojecting

limate

onditions

so knowledge of R is not necessarily required. A basic

data quality control procedure that checks for outliers

(unusual values that might be the result of, for example,

a data transcription error) and does a basic homogene-

ity assessment is also built into the software. Though

the indices have simple definitions, their calculation is

not necessarily simple, particularly if one wants to avoid

inducing artificial inhomogeneities (jumps) in the index

time series

or bias in percentile-based indices due to

a lack of precision in observational data.

Innovative

approaches have been implemented in RClimDex to

ensure the indices it calculates are homogeneous.

Furthermore, to ensure that the index time series are

homogeneous, the climatic time series from which they

are calculated should also be homogeneous. However,

climatic time series may contain spurious (nonclimatic)

jumps and/or gradual shifts due to changes in station

location, environment exposure, or instrumentation

and observing practices.

Such inhomogeneities would

hinder the identification and assessment of change

in climate, and thus Environment Canada has also

developed and made available a free R-based data

homogenization package called RHtest.

A detailed user

manual for RHtest is provided in both the English and

French languages and is continually updated to reflect

results from new research. For example, the most recent

version of RHtest is able to automatically detect single

or multiple change-points in climate time series consist-

ently throughout the length of the series. Both software

tools will allow all interested parties to benefit from

improved monitoring of climate change with broader

spatial coverage than was previously available, thus

directly contributing to the goals of the Nairobi Work

Programme.

The ETCCDI workshops

To promote the analysis of extreme events around the

world and help build capacity in the less developed

world, ETCCDI has also organized several regional

workshops. These workshops, which followed the

model pioneered in December 1998 at an Asia-Pacific

Network for Global Change Research (APN) meeting in

Melbourne, have proven to be very successful. The core

component of each workshop is the hands-on analysis of

national observational data with daily resolution, which

have often never been analysed prior to the workshop.

RClimDex and RHtest are used to perform the analysis.

An ETCCDI workshop usually involves participants

from neighbouring countries and several well-qualified

experts from around the world to provide guidance on

the analysis of climate data. In some cases, computers

for the participants have been provided by sponsoring

agencies for the workshops.

A workshop typically starts with each participant

presenting information on the climate of their country,

as well as their daily precipitation and temperature

data. The participants then learn data quality control

and homogenization procedures, and conduct the

computation and trend analysis of climate indices. At

This has proven to be a difficult task. Detection becomes increasingly

difficult at smaller regional and local spatial scales and for extreme

events, because the signal to noise ratio is reduced in both of these

circumstances.

This feature is not attributed to a weaker signal at

the local scale compared to the global scale, rather it is due to the fact

that influences of natural variability on all aspects of the climate are

larger on local scales. In addition, it remains a challenge to document

past climate change for some places of the world due to the limited

availability of assessable data or resources. Understanding the past

and current climate at small spatial scales will not only provide a

baseline for the future, but also a means to validate and constrain

future projections. It is, therefore, a necessity for the development of

sound adaptation strategies.

The ETCCDI indices

Long-term, high quality and reliable climate records with a daily (or

higher) time resolution are required for assessing changes in extremes.

However, the compilation, provision, and update of a globally

complete and readily available full resolution daily dataset are very

difficult tasks. This comes about, in part, because of the traditional

focus of climatologists on monthly data and the inability of some

countries to exchange long-term daily climate records. Nevertheless,

adequate adaptation requires the development of quality, daily

climate records and the ability to use those records to track changes

in the climate and its extremes. The joint Expert Team on Climate

Change Detection and Indices (ETCCDI) and its predecessors have

coordinated the development and application of a suite of indices for

monitoring climate change from daily data.

The use of agreed indices

allows comparison of analyses conducted in any part of the world.

The indices focus primarily, but not solely, on extremes. The

extreme indices describe different aspects of temperature and precipi-

tation including frequency, intensity and duration. There are three

different types of indices. One type involves counting the number

of days in a season or a year that exceed specific thresholds at which

impacts of weather may occur – such as daily temperature below 0°C

or daily precipitation amounting to greater than 20 millimetres. The

number of such events may not be evenly distributed across a large

region. The specific thresholds may, or may not, represent extreme

events in a given region under current climate conditions, and it is

possible such events may not occur at all in some regions. Impacts

also vary across regions. To overcome such shortcomings, a second

type of ETCCDI index uses thresholds based on percentiles to assess

moderate extremes that typically occur a few times every year – such

as daily temperature greater than its 90th percentile. The third type

of index is of more relevance to the derivation of design values in

applications that involve values of absolute extremes. These indices

include, as examples, the annual maximum daily temperature and

highest five day precipitation amount in a year. The indices are widely

used for monitoring changes in extremes,

climate model evaluation

and the assessment of future climate.

The software

To facilitate the calculation of the indices, Environment Canada,

under the auspices of WMO and ETCCDI in particular, has devel-

oped a standard software package RClimDex. This software uses the

open source statistical programming language R

(www.r-project.org

which runs on a variety of computer platforms. It is freely available

from

http://cccma.seos.uvic.ca/etccdi

and comes with a tutorial (in

both English and Spanish). A graphical user interface is provided,