SOEE2810: Coursework documentation

The Last Glacial Maximum

1 Background

1.2 Ice ages

The last ice age was punctuated by an alternation of cold and warm periods, called glacial-interglacial cycles. The Milankovitch theory tells us that these were driven by cycles in the three main parameters that  influence the  Earth’s  orbit:  the  eccentricity,  obliquity  and  precession  (Figure  1).  Eccentricity describes the shape of the Earth’s orbit around the sun, varying from nearly a circle to an ellipse with a period of about 96,000 years. Obliquity is the tilt of the Earth’s axis of rotation with respect to the plane of its orbit, which changes with a period of about 41,000 years. Precession refers to the fact that both Earth’s rotational axis and orbital path precess (rotate) overtime – the combined effects of these two components and the eccentricity produce approximately a 21,000-year cycle. The three cycles have  different  effects  at  different  places  on  our  globe.  Obliquity  has  a  strong  influence  at  high latitudes,whereas precession has a notable impact on tropical seasons.

The Milankovitch cycles have a control on the variability in shortwave radiation that reaches Earth. The sum of the three Milankovitch cycles have been shown to be in phase with variability in aspects of climate,  such  as global  mean  temperature  and  global  ice  sheet  volume  (Hays  et  al.  1976).  In particular, the amount of shortwave radiation that reached the surface at the latitude of the Northern Hemisphere ice sheets (60N) controlled their growth and retreat and can explain the variations in ice sheet  volume  and  sea  level  during  glacial-interglacial  cycles  (Figure  2).  Glacial  to  interglacial oscillations occur on approximately a 100,000 year timescale (Figure 3). The last glacial period was at a maximum approximately ~21,000 years ago.

Figure 1 The main cycles in Earth’s orbit (Source Maslin, 2016)

Figure 2 The combined effect of precession, inclination and eccentricity on incoming shortwave radiation and temperature (sourcehttps://www.periodni.com/gallery/milankovitch_cycle.png)

Figure 3 Ice ages during the past 2.4 billion years (upper) and glacial-interglacial cycles over the past 450,000 years (Eldredge and Biek 2010).

1.3 Last Glacial Maximum

The  Last  Glacial  Maximum  (LGM),  is  defined  climatologically  as the  period  of stable  cold  climate around the maximum of global ice sheet volume approximately 21,000 years ago. During the LGM, land-based ice sheets more than 2km in depth, covered Canada, north USA, northern Europe and parts of Eurasia (Figure 4). Global sea level was 130 m lower than the present day, joining Britain to Europe, Australia to Asia and Alaska to Siberia. South of the ice sheets, conditions were cold, windy and dusty. There was 20-25 times more dust in the atmosphere due to reduced vegetation, stronger winds and less  precipitation to wash out  dust from the atmosphere. The  modern forests of  North America, Europe  and  Asia  were  tundra  and  grasslands.  There  were  lower  atmospheric  concentrations  of greenhouse gases such as carbon dioxide and methane, causing cooling and drying across the tropics. The main influencing drivers on LGM climate and how well we understand them are summarised in Figure 5.

Changes in orbital parameters over the 10s of thousands of year before the LGM lead to the build-up of these large ice sheets. However, by the time of the LGM, incoming shortwave radiation from the Sun had returned to levels that are similar to the present day (which subsequently initiated the move into today’s interglacial period). There is a time lag in the climate system between the minimum in incoming shortwave radiation and the maximum in ice sheet cover because (1) it takes time for the build-up of ice and (2) negative climate feedbacks reinforce the cooling (see section 1.4).

Deglaciation in the Northern Hemisphere started ~20,000 years ago. Deglaciation in Antarctica started ~14,500 years ago – at the sametime as abrupt sea level rise.

Figure 4 Land-based ice sheet coverage during the Last Glacial Maximum.

Figure 5 Influences on LGM climate cooling (generally feedbacks in glacial-interglacial cycles) and he  level of scientific understanding associated with each one (IPCC, 2007, Figure 6.5 in Chapter 6.4.1.3).

Why study the last glacial maximum?

•    It provides a means for evaluating the understanding and modelling of the response of the climate system to large radiative perturbations.

•    The LGM and the subsequent deglaciation have been widely studied because the radiative forcings, boundary conditions and climate response are well known compared to previous paleo-eras.

•    The response of the climate system at the LGM included feedbacks in the atmosphere and on land, amplifying the orbital forcing.

•    Concentrations of well-mixed greenhouse gases at the LGM were reduced relative to pre- industrial values, amounting to a global radiative perturbation of –2.8 W m–2.

1.4 Climate feedbacks

Small changes in incoming shortwave radiation at the top of Earth’s atmosphere, driven by changes in  the Earth’s orbit, can push the planet into or out of an ice age alongside Earth’s “climate feedback” mechanisms:

•    An  example  of  a  positive  climate  feedback  in the  LGM  is  the  ice-albedo  feedback;  when incoming solar radiation is reduced, global temperatures decrease, encouraging the formation of ice. Ice is white and reflects more sunlight than the ocean or land surface below it (high albedo), which further cools the area, reinforcing the cooling and allowing even more ice to form.

•    A  second example of a positive climate feedback is that the ocean is able to absorb  more carbon dioxide when it is cooler, so during glacial periods, when global temperatures were cool, the ocean could absorb more carbon dioxide. Lower concentrations of this greenhouse gas in the atmosphere reinforced the cooler temperatures.

•    An example of a negative climate feedback in the LGM is that the climate became much colder and drier with reduced vegetation, which increased the amount of dust in the atmosphere. This dust was deposited on the ice surface, making it darker and thus allowing it to absorb more  solar  radiation.  This  would  act  to  increase  the  surface  temperature  and  melt  ice, allowing vegetation to become re-established and reducing dust emissions.

It is the fine balance between external forcings such as changes to incoming solar radiation and climate feedbacks, that determines what the climate is like and how quickly it is changing at anyone time. In the LGM, variations in the planetary orbits decreased the incoming solar radiation, which started the glacial period but this was 1000’s of years before the glacial maximum. Climate feedbacks amplified the  reduced  solar  radiation;  by  the  time  of  the  LGM,  the  incoming  solar  radiation  had  almost recovered to present-day values but climate feedbacks and the long lag time between decreased solar radiation and global cooling kept the climate cool and ice sheets present for many more years.

There are many limitations and challenges to our current understanding of the climate during the LGM and how it developed. Lack of a substantial observational records limits our ability to understand and study the period and many open questions remain.

1.5 Surface energy budget

Earth’s surface energy budget describes the balance between radiation, conduction and convective heat flow into and out of the surface (Figure 6). By convention, flow towards the surface is positive and flow away from the surface is negative. Earth receives its energy from the Sun, through incoming shortwave radiation at the top of the atmosphere. Some of this energy is reflected back off the surface and the remainder is absorbed by the surface. The surface albedo defines the proportions. Heat is transferred  back to the  atmosphere via the  longwave  radiation  flux,  and the sensible  and  latent (convective) heat fluxes. Heat can also be transferred into the ground through the conductive ground heat flux.

Clouds complicate this system by reflecting incoming shortwave radiation back towards space and reflecting upwelling shortwave radiation back towards the surface. Clouds can also absorb and emit longwave radiation (Figure 7).

Figure 6 The components of the surface energy budget and how to compute the surface albedo.

Figure 7 A more complete picture of the Earth’s entire energy budget, including the impact of clouds.

2 Climate model documentation

2.1 What is a climate model and how does it work?

A climate model is the only way to project how climate may change in the future (Figure 8). They can also be used to reproduce present and ancient climates. Climate models solve complex mathematical equations that are based on well-established physical laws that define the behaviour of the weather and climate. However, it is not possible to represent all the detail in the real world in a computer model, so approximations have to be made.

Understanding climate involves running long (~500-1000 year) simulations to represent the climate in a chosen period. Different climate forcing scenarios can be implemented into the simulation such as  incoming shortwave radiation, natural and anthropogenic sulphur emissions, atmospheric aerosols such as sea salt, dust, volcanic emissions, ozone, greenhouse gases (e.g. carbon dioxide, methane, CFCs). Comparing a ‘control’ simulation, such as that for pre-industrial times with an ‘experiment’ simulation, such as one for the LGM isa means of exploring how different climate forcings may affect global climate. This provides a set of internally-consistent pictures of past climates, each dependent on a set of prior assumptions.

Figure 8 Schematic of a climate model. Data is stored at each ‘gridbox’ . Many processes are taken into account (see insert).

After the model is initialised there is a period of rapid adjustment (1st  year of simulation) and then it takes 10’s or 100’s of years to reach climate equilibrium. In the example in Figure 9 it takes 15 years of model simulation to reach equilibrium, so only the last 5 years of model data are used for analysis.

Figure 9 Model spin-up towards equilibrium.

There are many uncertainties in climate models; they are not truth, rather our best representation of climate in past, present and future periods. Uncertainties are due to three main reasons:

•    Uncertainties in climate model input data i.e. we are unable to accurately quantify aspects of the past climate such as the greenhouse gas concentrations and ice sheet extent.

•    The model simulates or approximates many climate processes such as ocean currents, clouds, ice melt and atmospheric circulation. We know that these processes are not represented with 100% accuracy in present-day climate simulations and are likely to be even less accurate in simulations of past climate.

•    Lack  of  available  observations  of   past  climates.   Observations  of   key  variables  such  as temperature,  greenhouse  gas  concentrations  and  ice  sheet  extent  are  very   limited,  so evaluating and verifying model simulations is very challenging.

2.2 HadCM3 model

HadCM3 (abbreviation for Hadley Centre Coupled Model, version 3) is a coupled atmosphere-ocean general circulation model (AOGCM) developed at the Hadley Centre in the United Kingdom. It was one of the major models used in the IPCC Third Assessment Report in 2001. Although now quite an old model, HadCM3 is still used in paleoclimate research because it’s relatively coarse resolution and simple representation of climate processes means it is quick to run on supercomputers, so can be run for thousands of  years, showing little drift in its surface climate.

HadCM3 is composed of two components: the atmospheric model HadAM3 and the ocean model HadOM3 (which includes a sea ice model). Simulations use a 360-day calendar, where each month is 30 days.

2.2.1 Atmosphere model (HadAM3)

HadAM3 is a grid point model that has a horizontal resolution of 3.75 × 2.5 degrees in longitude × latitude. This corresponds to a spacing between points of approximately 300 km. There are 96 × 73 grid points on the scalar (pressure, temperature and moisture) grid and the vector (wind velocity) grid is offset by 1/2 agrid box (see Arakawa B-grid). There are 19 levels in the vertical using a hybrid (sigma and pressure) coordinate system.

2.2.2 Ocean model (HadOM3)

The ocean model has a horizontal resolution of 1.25 × 1.25 degrees, and thus there are six ocean grid points for every atmospheric one. There are 20 vertical levels. For ease of coupling the two models the grids are aligned and the ocean coastline is forced to be aligned to the atmospheric grid.

2.2.3 Coupling

The atmospheric model is run for a day, and the fluxes (of heat, moisture and momentum) at the atmosphere-ocean interface are accumulated. Then the ocean model is run for a day, with the reverse fluxes accumulated. This then repeats through the length of the run.

2.3 HadCM3 simulations

Here we compare the climate model’s representation of the MODERN period (which relative to 21,000 years ago is close to present-day climate) with the climate model’s representation of the LGM.

HadCM3 is run twice; once for the MODERN period and once for the LGM period. The two model simulations are initialised with the appropriate ancillary data and are run for 500 years. The first year of each simulation is always discarded to allow for the initial model rapid adjustment to its ancillary conditions. The model climate settles down to an equilibrium over the course of the first 450 years of the simulations, so only the last 50 years of the simulations should be used to analyse the differences in climate between the LGM and MODERN periods.

2.3.1 Simulation details

2.3.2 Directory structure

•    /ancillary_files contains data used to initialise the climate model

•    [experiment_ID]/climatology  contains  50-year  climate  means  for  each  gridbox  (calculated from year 451-500 of the run). This data should be used to analyse the mean climate of the LGM and MODERN periods.

•     [experiment_ID]/time_series contains monthly means for 499 years of the simulation i.e. it does not include data from the first year of the simulation. This data should be used to analyse how long the climate in the simulations takes to reach equilibrium.

2.3.3 File naming convention of climatology files expID[a,o].XXYYZZZ.nc, where:

-             expID is the 5 letter simulation code (teada or teadv)

-             a/ois for atmosphere [a] or ocean [o] model output

-             XX is for the filetype (pc, pd, pf, pg, pt, as above)

-             YY is ‘cl’ for climate mean and ‘sd’ for standard deviation

-             ZZZ is ‘ann’ for annual mean, ‘jan’ for January mean, ‘feb’ for February mean. 2.3.4 File contents

Ancillary files

Ancillary files contain important input data required to initialise the climate model. These fields are static over the course of the climate  model  simulation,  but they  are  different  in the  two  model experiments. The ancillary files available to you are:

Climatology files

Within [experiment_ID]/climatology there are 4 subfolders:

•    /pcpd contains atmospheric output. Generally pd files contain single layer (2D) fields (e.g.    1.5m temperature and pc files contain multi-level (3D) fields (e.g. temperature on pressure levels). Note sea ice fraction is available on both the atmospheric and ocean grids.

•    /pf contains monthly and annual mean ocean model output (mostly from the top ocean layer).

•    /pg contains annual mean ocean output (mostly 3D fields).

•    /pt contains vegetation model output.

•    Standard deviations are also included.

‘Diagnostic’ is the term used for data that is saved out from the climate model simulations e.g. near- surface temperature is a diagnostic. Summary of key diagnostics in the climatology files (many more are contained in the files, open the netCDF files to view the full list of diagnostics).