GENIE:QuaternaryBoundaryConditions

Bob Marsh, National Oceanography Centre, Southampton, ([mailto:rma@noc.soton.ac.uk rma@noc.soton.ac.uk])

Specifying boundary conditions for Quaternary experiments
---This page is still under construction---

Through two studies with the 36x36s grid, we have developed boundary conditions and experimental designs for carrying out long simulations through all or part of the last glacial cycle. This page provides information on:


 * which boundary conditions
 * interpolation in time
 * interpolation in space
 * filenames (specified in genie-main/namelists.sh) and flags
 * new parameters

In the full glacial cycle experiments (Marsh et al., submitted), the ocean and atmosphere are initially at rest and prognostic variables are set to constant values: the ocean with temperature 20°C and salinity 34.9 psu; the atmosphere with temperature 0°C and relative humidity 0%. Each experiment has a 124,000-year duration. For the first 4000 years, boundary conditions are fixed at values for 120 ky BP in order for the thermohaline circulation and global climate to reach an equilibrium state. Lunt et al. (2006) find that this is sufficient time for a simulation to come to equilibrium, and that an equilibrated climate is a good approximation to the equivalent snapshot of a transient climate in the absence of short-term variability such as Heinrich events. The implication is that, if we had started from, say, 150 ky BP and carried out a 30-ky spin-up under transient boundary conditions for 150-120 ky BP, we would obtain a very similar initial state.

Orbital forcing
Solar insolation changes with variations in the Earth's orbit, which are caused by gravitational interaction between the sun, the planets and their satellites (Pälike, 2005). Consequently the amount of energy received at a given latitude and in a given month is dependent upon the Earth's orbital and rotational parameters defined as obliquity, eccentricity, and precession, which vary on several dominant time scales, including ~20Kyr (precession), ~40Kyr (obliquity) and ~100 and 400Kyr (eccentricity). Obliquity variations have a substantial effect on local annual-mean insolation but do not influence the global mean annual incoming solar radiation. Eccentricity variations do affect annual insolation, although the effect is relatively small. Variations in precession control the timing of the approach of the perihelion (when the Earth is closest to the Sun) and consequently the seasonality of insolation. The model uses orbital parameters to generate the insolation forcing at all latitudes. These parameters are constructed using data from Berger (1978).

The orbital forcing filename is ...., and is read as follows ...

Radiative trace gases
1. CO2

Ice core records show that CO2 is an important amplifier of orbital forcing (Genthon et al., 1987). We use the record of CO2 directly measured in air bubbles trapped in the Vostok ice core (Petit et al., 1999), interpolated to every 1000 years, shown in Figure 2a. The CO2 concentrations are set to the GT4 timescale (Petit et al., 1999), an extension of the original timescale (Lorius et al. 1985) including subsequent improvements (Jouzel et al. 1993). The chronology is derived from ice-flow and ice accumulation models and therefore is purely based on physics rather than climatic correlations, apart from a number of control ages. The timescale accuracy of the CO2 reconstruction over the last 120 ky is ±5 ky for the last 110 ky, which decreases to ±10 ky for the remaining 10 ky. The record indicates an approximate 100 ppmv glacial-interglacial range, with minimum values around the LGM.

The CO2 concentration filename is ...., and is read as follows ...

Ice Sheets
The "ICE-4G" reconstruction provides terrestrial ice sheet fraction and orography for the period from 21 ky BP to the present, at a time resolution of 1 ky (Peltier, 1994). The ICE-4G method is based on an understanding of how relative sea-level records the viscous deformation of the earth, due to the presence of terrestrial ice sheets, using exponential relaxation curves for ice-covered regions. These curves contain information about the internal rheology of the planet and the space- and time-dependent thickness of the ice load when this is combined with relative sea-level data, and are deconvolved to produce the paleotopography of the terrestrial ice sheets.

In a related GENIE-1 study, Lunt et al. (2006) used the ICE-4G reconstructions every 1000 years over 0-21 ky BP, along with the Arnold et al. (2002) reconstruction of northern ice sheets at 30 ky BP, and combined the ICE-4G data over 0-21 ky BP and the Vostok delta O-18 record (Petit et al., 1999) over 0-29 ky BP to obtain ice sheet extent and orography over 22-29 ky BP. As the ICE-4G record spans the period of maximum ice cover (LGM) to the present, which can be assumed as an approximation of the minimum amount of ice cover over the glacial-interglacial cycle (neglecting more extensive melting of the Greenland ice sheet during the Eemian), this period can be likewise used as an analogue to approximate the ice volume change during the period since 120 ky BP. Marsh et al. (submitted) therefore reconstructed a millennial time series of ice sheet extent and orography over 22-120 ky BP using the low-resolution sea level reconstruction of Siddall et al. (2003), as an indirect measure of change in global ice volume. This reconstruction spans +8.1 m to -112.4 m, relative to present day sea level.

Time variation of reconstructed ice volume should be consistent with the time-varying glacial freshwater fluxes that include both global net evaporation (for ice sheet growth) and meltwater pulses (MWPs). The modelled ice sheets are therefore directly based on the smoothed and HE-adjusted global mean sea-level curve.

The ice sheet extent and orography filenames are ...., and are read as follows ...

Land Ice-Ocean Freshwater Exchange
We use the low-resolution sea-level reconstruction of Siddall et al. (2003) to develop a time series of net global surface freshwater flux. The procedure comprises four stages:

(1) The raw sea-level data (Fig. 2b) are too noisy for the derivation of rates of change. We therefore fit the glacial-interglacial cycle with two trends (-128 m over 120-20 ky BP and +128 m over 20-10 ky BP). (2) We then smooth the anomalies about these trends using an exponentially weighted moving average. (3) We then add the smoothed anomalies at 500-year intervals to the trends, providing a smoothed record that also incorporates the abrupt onset of deglaciation at 20 ky BP. Approximating that the total area of the global ocean remains constant over time, rates of change in this global-mean sea level record (every 500 years) are simply converted to global net freshwater fluxes. (4) Finally, we adjust fluxes at the end of deglaciation (12-8 ky BP) to ensure a gradual decline in the rate of sea-level rise that is consistent with higher-resolution sea-level records for that interval of time.

To this smoothed time series of net glacial freshwater fluxes, we then add a series of nine MWPs (positive freshwater fluxes), compensated by weaker net evaporation during intervening periods to ensure global conservation of ice/seawater over the entire glacial-interglacial cycle. Our choices of chronology, duration and magnitude of MWPs are based on published evidence.

When the global-mean glacial freshwater flux is negative, we apply the flux equally at the surface, at every ocean gridpoint in the World Ocean. When the global-mean glacial freshwater flux is positive, we apply the flux at the surface, equally at every ocean gridpoint in the North Atlantic, in the zone 50-70°N. This choice is based on a general assumption that glacial sea level variability was associated with discharges of meltwater from the Laurentide and Fennoscandian ice sheets into the Atlantic within this latitude range, similar to common practice in previous model experiments (e.g., Manabe and Stouffer, 1997). In the context of MWPs, this choice is further supported by field evidence for Heinrich events in these mid-latitudes (Hemming, 2004). In the experiments, the effect of net glacial freshwater fluxes over 120-20 (20-0) ky BP is to increase (decrease) global-mean salinity in the model by 1.245.

The freshwater flux filename is ...., and is read as follows ...

Time-varying Atlantic-Pacific moisture flux
In a further study (Marsh and Ridgway, in prep.), we allow the prescribed Atlantic-Pacific moisture transport, F, to vary according to three aspects of glacial climate change: (1) F decreases with global cooling, which dries the global atmosphere and weakens moisture transports everywhere (2) F increases with LIS height, which acts to assist Atlantic moisture export, based on evidence that mountain ranges favour a vigorous Atlantic thermohaline circulation (Sinha, pers. comm.) (3) F decreases with Pacific cooling, which favours weaker and/or less frequent El Niño conditions, hence weaker Atlantic moisture export (Schmittner and Clement, 2002)

We investigate the sensitivity of ocean circulation to each influence by specifying two values for each controlling parameter, corresponding to high and low sensitivity respectively: global-mean air temperature range = 2.5°C or 5°C; height range = 500 m or 1000 m; ropical Pacific sea surface temperature range = 1°C or 2°C. The choice of these values was guided by prior knowledge (from Marsh et al., submitted) of the glacial-interglacial ranges in global-mean air temperature, the Laurentide Ice Sheet height, and tropical Pacific sea surface temperature, combined with a hypothesis that the Atlantic-Pacific moisture flux may change on the order of ±100% under glacial climate.

The three new parameters are ..., and these are listed in ...