Simulation of climate change over Europe using a global variable resolution general circulation model

Simulation of climate change over Europe using a global variable resolution general circulation model - pdf for free download
Uploaded by: Pascal Marquet
Category: Oceanography, Climate Change, Atmospheric Modeling, Sea surface temperature, Mediterranean Sea, Climate Dynamics, Seasonality, Atmospheric sciences, Soil moisture, Boundary Condition, Climate Dynamics, Seasonality, Atmospheric sciences, Soil moisture, Boundary Condition

86062 downloads 86125 Views 1MB Size

Recommend Documents

No documents

Story Transcript


Climate Dynamics (1998) 14 : 173—189

( Springer-Verlag 1998

M. De´ que´ · P. Marquet · R. G. Jones

Simulation of climate change over Europe using a global variable resolution general circulation model

Received: 26 February 1997 / Accepted: 21 October 1997

Abstract This study presents results from a downscaling simulation of the impact of a doubling of CO 2 concentration. A multidecadal coupled simulation of a 1% per year increase of CO concentration with the 2 Hadley Centre ocean-atmosphere model provides its sea-surface temperatures and deep soil climatological temperatures as a boundary condition to two 10-year integrations with a version of the ARPEGE-IFS atmosphere model. This global spectral model has a horizontal resolution varying between 60 km in the Mediterranean Sea and 700 km in the southern Pacific. The global impact as well as the regional impact over Europe in this time slice are examined and compared with results from other studies. Over Europe, our main focus, the model impact consists of a warming of about 2 °C, relatively uniform and with little seasonal dependence. There are precipitation increases of about 10% over the northern part in winter and spring, and 30% over the southern part in winter only. Precipitation decreases by 20% in the southern part in autumn. The day-to-day variability of the precipitation increases, except over the southern area in summer. No strong impact is found on the soil moisture. Budgets of physical fluxes are examined at the top of the atmosphere and at the land-atmosphere interface.

1 Introduction The greenhouse gas phenomenon plays a fundamental role in the equilibrium climate of the globe. The assoM. De´que´ ( ) · P. Marquet Me´te´o-France Centre National de Recherches Me´te´orologiques, 42 Av. Coriolis, Toulouse, France R. G. Jones Hadley Centre for Climate Prediction and Research, Meteorological Office, Bracknell, UK

ciated gases have been increasing since the beginning of the last century due to the anthropogenic emission of CO and other gases (IPCC 1995). However a steady 2 temporal and uniform spatial warming has not been observed, since: 1. A part of any additional heating is absorbed by the oceans. 2. The remaining part will modify the atmospheric circulation which in turn will distribute the heating in a non-homogeneous way. 3. Any climatic signal is sufficiently weak that it is masked by natural variability. 4. Associated aerosol emissions tend to cool the atmosphere. The aim of this study is to contribute to point (2) for the European region. Coupled atmosphere-ocean models are the major tool to investigate the geographical distribution of impacts of an increased greenhouse gas warming. The high cost of their numerical integration over long periods using present generation computers allows us to study the impact only at the continental scale: the horizontal resolution involved is about 300 km (see for example Murphy and Mitchell 1995; Timbal et al. 1995). To investigate the regional impact, i.e. to account for the the distribution of land surface properties, e.g. the Alps, Pyrenees and Apennines which contribute to temperature and precipitation anomalies on horizontal scales smaller than 1000 km, these general circulation models (GCM) are not accurate enough. A solution to this limitation consists of assuming that the ocean surface temperature response is weakly affected by the regional distribution of the anomalies. Thus we can run ‘‘time slice’’ or ‘‘snapshot’’ experiments consisting of an atmospheric GCM forced by the SSTs (and if needed also the deep soil temperatures over land) calculated by a coupled ocean-atmosphere model (Stephenson and Held 1993). The coupled model is generally a coarser and thus less expensive one (e.g. Mahfouf et al. 1994). Then, AGCM integrations of only

174

De´que´ et al.: Simulation of climate change over Europe using a global variable resolution general circulation model

several years are necessary to simulate the control and the perturbed climate. In a time-slice experiment, it is not necessary for the two atmosphere models to be identical. As with an AMIP simulation (Gates 1992), the second model ‘‘sees’’ the SST as a constraint and is not allowed to interact with it. The difference with an AMIP exercise is that the SST is not perfect and includes systematic errors from the coupled system, so we have to choose a coupled experiment that is as accurate as possible. The lack of feedbacks of the atmospheric fluxes on the SST is the price paid for saving computer time: if one wants an atmosphere in equilibrium with its SST, a multidecadal run is necessary because of the time constants of the OGCM; in the case of a transient climate change, this implies century time scale experiments. The lack of feedback is particularly detrimental if the second model calculates much more realistic large-scale surface fluxes (this is obviously not the case in an AMIP run where the coupled model is perfect). Three kinds of AGCM can be used to improve the regional-scale boundary conditions: 1. Global high resolution models (e.g. Wild et al. 1995). This is generally the best and the most expensive way to proceed. 2. Varying resolution global models (e.g. De´que´ and Piedelievre 1995 referred to as DP95 in the following). This intermediate approach focuses on a given region (in our case Europe). With the same amount of computation as in (1), a higher resolution in the domain of interest can be achieved, at the expense of a coarser resolution over the rest of the globe. In DP95, this method is shown to provide a better climate over Europe than a more expensive global high resolution version of the same model. 3. Limited area models (LAM) forced at their lateral boundaries by a global coupled GCM (e.g. Marinucci and Giorgi 1992; Jones et al. 1995). This approach is the most flexible and allows a higher resolution than in (2) for the same amount of computation. The difficulty comes from the possible conflicts at the lateral boundaries, since the driving GCM ‘‘ignores’’ the atmospheric circulation of the LAM. In the present study, we use the second approach. The ARPEGE—IFS model is a spectral atmospheric model developed by Me´te´o—France and ECMWF for operational weather forecasting in the short— and medium—range. A climate version has been derived from this code (De´que´ et al. 1994) to study anthropogenic climate changes and seasonal predictability. This model offers the possibility of varying the horizontal resolution with the geographical position. This possibility is used daily Me´te´o—France for short—range prediction. In DP95 and Machenhauer et al. (1996) we have shown that a variable resolution version reproduces, with some realism, the seasonal and geographical variations of the main climatological parameters over Europe. This model is described in Sect. 2. A long

coupled integration has been carried out at the Hadley Centre with the UKMO unified model and with an increasing CO rate. Two 10—year ‘‘time slice’’ integra2 tions have been performed with ARPEGE—IFS, using the results of the coupled integration as a boundary condition. This experiment is presented in Sect. 3. The impact on the global scale of a CO doubling is sum2 marized in Sect. 4. Section 5 details the response over Europe, and the significance of the results is discussed in Sect. 6. Finally, the role of the SST anomalies is examined in Sect. 7.

2 The variable resolution model The principle of the variable resolution model is to discretize the atmospheric variables and equations with a system of horizontal coordinates different from classical latitude-longitude. The new system corresponds to the latitudes and longitudes on a new sphere, obtained by a suitable geometrical transformation of the globe (assumed to be also a sphere). A constraint imposed on the choice of the geometrical transformation is that the length of the image of a tangent vector depends on its position, its length, but not its direction (isotropy). This property, used in the marine chart mapping (the target is a plan instead of a sphere) since it conserves the angles, has two advantages in our case: 1. The equations after discretization are simpler since the derivative with respect to x and y are multiplied by the same factor; thus the gradient, divergence or curl operators remain simple. 2. The discretization does not introduce artificial effects by favouring a particular direction, for example by filtering certain mountain ranges and not others; the physical laws are isotropic. Courtier and Geleyn (1988) demonstrated that the family of the isotropic transformations between two spheres involves only two parameters: a point on the sphere (the pole of stretching), and a coefficient (the homothetic factor in the polar stereographic plan). In DP95 the pole of stretching was chosen in the Tyrrhenian sea (40 °N, 12 °E), and the stretching factor was 3.5. With a spectral truncation T63, such a model has thus a horizontal grid mesh of 60 km in the Mediterranean sea, increasing to 120 km in the mid—Atlantic, and eventually to 700 km in the southern Pacific. More details on the geometric transformation and on the stretched grid can be found in DP95. As far as the vertical resolution is concerned, the model has 31 vertical levels from the Earth’s surface to about 80 km. The 20 upper levels are constant in pressure and located above 200 hPa. This allows a rather good representation of the stratosphere. The physical parametrizations active in the stratosphere are the radiation (from Geleyn and Hollingsworth 1979), the orographic gravity wave momentum transfer, and

De´que´ et al.: Simulation of climate change over Europe using a global variable resolution general circulation model

a linearized scheme used to calculate the ozone photochemical sources and sinks (Cariolle and De´que´ 1986). The ozone concentration is a three-dimensional prognostic variable coupled to the rest of the model by the radiation...

Life Enjoy

When life gives you a hundred reasons to cry, show life that you have a thousand reasons to smile

Get in touch

Social

© Copyright 2013 - 2019 DOCKUN.COM - All rights reserved.