Within The Column

Electronics Repair Manuals

Schematic Diagrams and Service Manuals

Get Instant Access

current vectors, temperature, and salinity

FIGURE 12.2 Schematic diagram illustrating how the atmosphere and ocean are divided into columns in a typical coupled general circulation model experiment. Ocean and atmospheric grid sizes are commonly different. Computations take place simultaneously for all grid boxes at all specified levels (McGuffie and Henderson-Sellers, 1997).

are too small to be represented by even a 2° x 2° grid spacing. In such cases, the process is represented in a simplified manner as a function of other variables, a procedure known as parameterization (parametric representation). This may be based on observed statistical relationships between, for example, temperature and humidity profiles and cloudiness, or on some other simplified model of the process in question. In fact, parameterization of all forms of cloud is one of the most difficult problems in atmospheric GCMs and is the focus of much research at present (Cess etal, 1995).

Atmospheric general circulation models may be coupled to the ocean realm in a variety of ways. At the simplest level, the surface temperature of the ocean is prescribed (predetermined) and the ocean region of the model interacts with the atmosphere only in terms of moisture exchange. This is often termed a "swamp ocean" (Fig. 12.3). At the next level a "slab ocean" is specified as a layer of fixed depth (50-100 m); heat and moisture exchange with the atmosphere occurs, enabling SSTs to vary as the model run progresses. However, in such models there is no


Moisture Energy








Prescribed SSTs



Calculated SSTs with prescribed advection. No vertical motion


Calculated SSTs with prescribed advection and detailed calculation of fluxes through the mixed layer to the deep ocean

Detailed Mixed Layer f f

Dynamic Ocean

Calculated temperatures, velocities and salinities at all levels

FIGURE I 2.3 Schematic diagram to show the different levels of complexity and coupling with the atmosphere in various types of ocean model (McGuffie and Henderson-Sellers, 1997).

mechanism for heat exchange with the deep ocean and only a crude representation of horizontal energy fluxes. The mixed layer ocean is a further improvement, also involving prescribed horizontal advection, but with computation of fluxes to and from the deep ocean. The most complex level is a fully coupled ocean-atmosphere GCM (OAGCM) in which the ocean has internal dynamics in three dimensions, and exchanges of energy, moisture, and momentum take place at the ocean-atmosphere interface. It is worth noting that a major problem in coupling atmospheric to oceanic processes is the vastly different response times characteristic of each domain (Fig. 12.4). The slower response time of the deep ocean must be taken into account when the two systems are linked. The problem is further compounded if one is trying to investigate climate system changes involving ice sheets, which have even longer response times. One approach is to couple models of each system "asynchronously," that is, to operate an atmospheric model for a time appropriate for that system, then use the resulting atmospheric conditions as input to a model of the system operating on a different timescale. For example, Schlesinger and Verbitsky (1996) used the climate at 115 ka B.P., generated by an atmospheric GCM (coupled to a mixed layer ocean), to drive an ice sheet/asthenosphere model, in order to investigate the areas most likely to develop ice sheets following the last interglacial

FIGURE 12.4 Estimated response or adjustment (equilibration) times for different components of the climate system. Note that they vary by 6-7 orders of magnitude (Saltzman, 1985).

period (see Section 12.3.1). The ice sheet model was run for 10,000 yr (an appropriate interval of time to examine ice sheet development) but the atmospheric conditions obviously could not be computed over such a long period and so were fixed as those obtained from the 115 ka B.P. simulation. Although this approach clearly has its limitations, it does allow two systems with strikingly different response times to be examined in a somewhat coupled fashion.

GCMs can also be used to trace the pathways of materials within the climate system (Koster et al., 1989). This has been put to good use in paleoclimatic applications (Jouzel, 1991; Jouzel et al., 1993a; Andersen et al., 1998). For example, the long-distance transport of desert dust particles in the atmosphere has been traced using an AGCM for both modern and last glacial maximum (LGM) conditions. Source regions of dust deposited from the atmosphere are identified by "tagging" the dust originating from different areas (Joussaume, 1987, 1990, 1993). Modern-day simulations show a strong seasonality in atmospheric dust production, with atmospheric dust loading in August more than twice that in February. The largest source of dust (by far) is the Sahara /Arabia /central Asia region. Australia is the principal source of dust reaching east and west Antarctica, whereas South America contributes the most dust to central Antarctica. Simulations for the LGM show greater atmospheric dust deposition especially over the tropical Atlantic Ocean and Europe, but the modeled increases significantly underestimate the observed changes (recorded in ice cores). This could be due to many factors, including model resolu tion (poor representation of source areas) and/or inadequate characterization of dust entrainment, transportation, and depositional processes (wet and dry fallout). In view of the potential climatic significance of dust during glacial periods (Over-peck et al., 1996) these first steps toward fully incorporating the dust cycle into GCMs are important contributions. Further studies with higher resolution GCMs are now needed.

Isotopes (deuterium, 180) in the hydrological cycle have also been modeled with GCMs, for modern and glacial age conditions (Jouzel et al., 1987c, 1991, 1994; Joussaume and Jouzel, 1993). At each change of phase of water molecules in the model, appropriate fractionation factors are employed to calculate the mass of the isotopes in each reservoir (water vapor, precipitation, ice, groundwater). Iso-topic modeling is particularly important in paleoclimatic studies as it allows the direct comparison of model simulations with paleoisotope records (in ice, sediments, biological materials, speleothems, etc.), thus avoiding the need to calibrate the paleo-record in terms of, say, temperature for a comparison with modeled paleo-temperature output. Modern simulations reproduce the global pattern of 8lsO and 8D very well and the seasonal cycle is well captured at most sites in both high and low latitudes (Jouzel et al., 1987c). The LGM simulations show a similar overall 8180/temperature relationship to that derived from modern simulations (8lsO ~0.6°T, where T < -5 °C) and there were large decreases in 8180 at high latitudes (see Fig. 5.9) (Joussaume and Jouzel, 1993).

Sources of moisture can be traced using GCMs and this is useful in understanding how source regions might have been different in the past; this would be relevant, for example, in the interpretation of ice core geochemistry. Charles et al. (1994) used an AGCM to examine how source regions of precipitation reaching Greenland changed from the LGM to the present. The modern (control) simulation showed that 26% of Greenland precipitation was derived from the North Atlantic (30-50° N), 18% from the Norwegian-Greenland Sea, and 13% from the North Pacific. At the LGM, these values changed to 38%, 11%, and 15%, respectively. However, northern Greenland received distinctly more moisture at the LGM from the north Pacific source region, due to displacement of storm tracks around the Laurentide Ice Sheet. Southern Greenland received most of its snowfall from North Atlantic moisture sources. Because of the much longer (and colder) trajectory of the Pacific air masses, snow deposited on Greenland from such sources was much more depleted in 8lsO than snow from North Atlantic sources (~15%o lower). Charles et al. (1994) point out that if there was no change in temperature in Greenland, but only a shift in source region from purely North Atlantic moisture to a 50:50 mix of North Atlantic and North Pacific moisture, changes in 8lsO of snowfall could change by ~7%o, equivalent to the large amplitude oscillations seen during late glacial time in the GISP2/GRIP ice cores. This raises the interesting possibility that abrupt changes in 8180 seen in the ice cores from Greenland may be partly related to changes in storm tracks rather than large-scale (hemispheric) shifts in temperature. However, it is worth noting that this model did a very poor job of simulating modern precipitation in Greenland (with simulated precipitation exceeding observations by as much as 100%, or ~1 mm day"1, especially in summer); thus this experiment, though interesting, begs the question whether similar results would be found with a more accurate, higher resolution model.

Although the discussion so far has focused on general circulation models of atmospheric and oceanic systems, further development of GCMs will be towards total climate system models (CSMs), which will incorporate in a fully interactive way land surface and cryospheric processes (both terrestrial snow and ice, and sea ice) and biomes. Models of global biomes are now available (Prentice et al., 1992; Hax-eltine and Prentice, 1996) and have been used with output from GCM experiments to predict vegetation at times in the past (Harrison et al., 1995; TEMPO, 1996). Models with full biogeochemical cycling of materials in the atmospheric and oceanic realms are also under development (Brasseur and Madronich, 1992; Sarmiento, 1992). Nested models, in which a very detailed grid network for a specific region is used to model detailed geographical variations of the climate, given initial input from a larger scale GCM, will also become more widely available (Hostetler et al., 1994). Such models are especially important in mountainous areas where large-scale GCMs cannot provide the necessary topographic detail to produce meaningful regional simulations.

Was this article helpful?

0 0

Post a comment