Projections of climate change suggest that the future climate will be warmer, that mean precipitation rates will increase moderately, and that extreme precipitation rates will increase substantially. If the latter comes to pass there will be severe consequences for natural ecological and hydrological systems and for built infrastructure such as storm sewers, bridges, roads and water control structures.
The Canadian Centre for Climate Modelling and Analysis (CCCma) develops comprehensive dynamical models of the coupled global atmosphere/ocean/sea-ice/land-surface system. These models, such as the one described in Flato et al. (2000), simulate variations in atmospheric and ocean circulation on a global grid on time scales ranging from less than an hour to centuries. If one were able to step within the simulated world, one would experience "weather" and climate (average weather) not all that different from that experienced in the real world. Models of this type are used as lab tools to probe the mechanisms of climate variability, and to project future climate change. The latter type of experiment is performed by prescribing to the model changes in the atmospheric composition (i.e., greenhouse gas and aerosol concentration) that have been observed since 1850 and that are projected to occur during the coming century. Such an experiment conducted with the CCCma model is described by Boer et al. (2000a,b) and summarized briefly on the CCCma web site (http://www.cccma.bc.ec.gc.ca/models/cgcm1.shtml).
The CCCma model simulates climate on a global grid with 4608 points, each point representing a region that spans 3.75 degrees of latitude by 3.75 degrees of latitude, or approximately 300x300 km at mid-latitudes. Model output is archived twice each simulated day for most meteorological variables, including basic quantities such as air temperature, winds, and precipitation amount.
CCCma scientists have performed a preliminary analysis of changes in the extremes of air temperature and precipitation simulated by the model in an ensemble of three independent climate change simulations extending from 1850 to 2100 (Kharin and Zwiers, 2000). The analysis was performed by comparing estimates of long period return values for a period representing present day climate (nominally 1975-95) with estimates obtained for two future periods (2040-2060 and 2080-2100). Because the CCCma experiment consists of an ensemble of three independent climate change simulations, the samples available for these three 21-year periods actually contain the equivalent of 63 years of data, which is felt to be adequate to estimate moderately long (50-100 year) return period values.
The analysis performed to date has not utilized spatial information, except in the case of precipitation, where this has been done in a very rudimentary way. That is, long period return value estimates are produced separately at each grid point, with little if any use made of information at adjacent grid points.
This case study will challenge those interested in extremes value analysis and the manipulation and analysis of large data sets. The specific statistical challenge is to improve upon the extreme daily precipitation return value estimates described in Kharin and Zwiers (2000) by incorporating a spatial aspect into the extreme value analysis so that behaviour at nearby grid points improves the return value estimates at the point of interest.
(a) a pdf copy of Kharin & Zwiers (2000) describing the analysis that has already been performed.
(b) three 21-year time series of daily precipitation amounts simulated by the CCCma climate model in each of three 21-year "windows" representing the climates of 1975-95, 2040-60 and 2080-2100. That is, a total of 3x(3x21)=189 years of simulated daily precipitation data will be available.
(c) advice by e-mail on data access, data manipulation, and interpretation from either Slava Kharin and/or Francis Zwiers (send queries to Slava.Kharin@ec.gc.ca and/or Francis.Zwiers@ec.gc.ca).
The datasets available for this case study are "clean". The model output has features similar to those seen in observations, such as an annual cycle and substantial daily variability, but the model output is free of problems such as observational error, censoring, and missing data.
The available data sets are also voluminous. They contain a total of 189 years of simulated precipitation data (3x21 years for each time window). Each of the 68985 (=189x365) daily precipitation records in the data sets is presented on a 96x48 array representing the 4608 points of the longitude-latitude grid covering Earth in the CCCma climate model.
Case study participants will have two options for accessing and analysing these data.
One option will be to download the nine global 21-year files (gzipped) from the CCCma ftp site (ftp://ftp.cccma.bc.ec.gc.ca/pub/fzwiers/global). The pdf copy of Kharin & Zwiers (2000) and a "readme" file containing basic documentation are included. This option will be appropriate for analysts desiring a challenge manipulating very large data sets, and who wish to perform the spatial extreme value analysis over all points on the global grid. Analysts selecting this option will need large amounts of disk space to manipulate the downloaded files and they will need to be proficient in manipulating and displaying spatial information.
A second option will to download nine regional 21-year files (gzipped) from the CCCma ftp site (ftp://ftp.cccma.bc.ec.gc.ca/pub/fzwiers/canada). The pdf copy of Kharin & Zwiers (2000) and a "readme" file containing basic documentation are included. Each record in the regional files will contain a sub-array of the global grid covering Canada and surrounding waters (312 points). This region should provide ample opportunity to explore spatial approaches to extreme value analysis, while substantially reducing the amount of data to be analysed, the associated need for disk space, and the burden of manipulating very large data sets.
Boer, G.J.; Flato, G.M.; Reader, M.C.; & Ramsden, D. (2000a). "A transient climate change simulation with historical and projected greenhouse gas and aerosol forcing: experimental design and comparison with the instrumental record for the 20th century", Climate Dynamics 16:405-425.
Boer, G.J.; Flato, G.M, & Ramsden, D. (2000b). "A transient climate change simulation with historical and projected greenhouse gas and aerosol forcing: projected climate for the 21st century", Climate Dynamics 16:427-450.
Flato, G.M.; Boer, G.J.; Lee, W.G.; McFarlane, N.A.; Ramsden, D.; Reader, M.C.; & Weaver, A.J. (2000). "The Canadian Centre for Climate Modelling and Analysis Global Coupled Model and its Climate", Climate Dynamics 16:451-467.
Kharin, V.V. & F.W. Zwiers (2000). "Changes in the extremes in an ensemble of transient climate simulation with a coupled atmosphere-ocean GCM", Journal of Climate 13:3760-3788.