AMIP II Diagnostic Subproject 18


Intercomparison of Surface Climate Extremes
Project coordinator:
Francis W. Zwiers and Viatcheslav V. Kharin

 Canadian Centre for Climate Modelling and Analysis, Victoria, B.C., Canada
 (Also affiliated with the Department of Mathematics and Statistics,
 University of Victoria, Victoria, B.C., Canada V8W 2Y2)

    Climatic extremes have acute effects on various aspects of human life. For example, the viability of many crops is constrained by  the number of frost free days per year, the frequency and duration of  high temperature events that expose crops to damage from heat stress,  and the availability of moisture.  Insurance schemes spread risk  across large pools of users and a number of years by assessing  premiums actuarially from historical claims data, historical extreme  weather event data and other related data.  Changes in the frequency  and/or intensity of extreme events will affect these economic risk  sharing arrangements. Human health is affected by weather extremes  directly through the physiological effects of heat and cold and  indirectly by floods, pollution episodes, and the like.

    It is therefore of great interest to document the extremes of surface  temperature and precipitation that are simulated by modern general  circulation models and compare them to the observed extremes.

    Objectives
     

    1. Evaluate and document the ability of current generation GCM's to simulate the extremes of surface temperature and precipitation.
    2. Relate the ability of GCMs to simulate the extremes to model's characteristics.

      3.  
    Methodology
     
      Two approaches will be used to characterize the extremes of the  simulated climate:

      1) 10-, 20- and 50-year return values will be estimated at every grid point by inverting the estimated Generalized Extreme Value (GEV) distribution fitted to the sample of annual extremes (annual extremes of 24-hour accumulated precipitation, daily maximum surface temperature and daily minimum surface temperature). (Zwiers and Kharin, 1998a;  Kharin and Zwiers, 1998b).

      2) threshold crossing frequency and duration analysis.

       The properties of the simulated extremes, and the extent to which their behaviour can be represented by standard  statistical models, will also be examined.  For example,  Kharin and Zwiers (1998b) find that the GEV distribution  does not properly fit the annual extreme minimum and  maximum temperatures simluated by the CCC coupled model  at all locations.  This happens when there is year to year  variation in the physical processes that govern the model's surface temperature. For example, Kharin and Zwiers (1998b)  display a point in central Australia for which the median annual  extreme minimum temperature is about -10C.  However, there  are several years in the sample for which the annual extreme  minumum temperature is very nearly 0C.  This clustering is  an artifact of the model's simple bucket type land surface  and the interpolating scheme that is used to diagnose screen  temperature.  The temperature of the land surface remains at  0C until all liquid soil moisture is frozen.  Thus there are  instances when the atmosphere would like to cool the surface  to a temperature substantially below zero, but the one-layer  land surface prevents this because it has a large thermal mass  of liquid water that is not properly isolated from the atmosphere.  We will use goodness of fit tests and other diagnostic techniques to determine whether other models exhibit similar types of  pathological behaviour. Our expectation is that models with  modern, multilayer land surface schemes (such as CCC's  GCM3 which will be used in AMIP2) will exhibit fewer such  problems.
       

    Data Requirements
       
      1. 6- hourly data of total precipitation rate (pr) will be required to calculate 24-hour accumulated precipitation.
      2. Daily maximum and minimum surface (2m) air temperature (tasmax and tasmin).
      3. Corresponding station data and reanalysis data will be required to validate the simulated extremes. NCEP reanalysis data are already housed at CCCma. Yet to be identified collaborators are required to provide access to station data.
    References
     
      Zwiers, F. W., and V. V. Kharin, 1998a: Changes in the Extremes of the  Climate Simulated by CCC GCM2 under CO2 Doubling. J Climate, in press.

      Kharin, V. V., and F. W. Zwiers, 1998b: Changes in the Extremes in an  Ensemble of Transient Climate Simulation with a Coupled  Atmosphere-Ocean GCM. (http://ams.allenpress.com/amsonline/?request=get-abstract&issn=1520-0442&volume=011&issue=09&page=2200)

For further information, contact the AMIP Project Office (amip@pcmdi.llnl.gov).
Last update: 18 June 1999.  This page is maintained by mccravy@pcmdi.llnl.gov

LLNL Disclaimers
UCRL-MI-127350