SCIENCE

Andrew J. Pitman, BSc Liv., PhD Liv

THE RELATIVE IMPACT OF CHANGES OF CARBON DIOXIDE AND LAND COVER CHANGE ON CLIMATE

Several major research projects have been pursued by Professor Pitman through 2000 and while most of these continue, some interesting results have already been obtained. The work splits into two major areas of study: the uncertainty of climate model simulations resulting from how the land surface energy balance is parameterised, and a continuation of the work with Mei Zhao investigating the relative impact of observed change in land cover and carbon dioxide using the CCM3 climate model.

Climate models (AGCMs) require information on how the land surface affects the atmosphere. Land surface models (LSMs) provide this information by describing the exchanges of energy, moisture and momentum across the land-atmosphere interface. One of the fundamental components around which an LSM is constructed is the surface energy balance (SEB). Conservation of energy at the surface is achieved by balancing a change in the energy state of the surface with the flux of absorbed solar radiation and net land-atmosphere exchanges of longwave radiation and sensible and latent heat.

In collaboration with Professor Bryant McAvaney from the Bureau of Meteorology Research Centre, we used the CHAmeleon Surface Model (CHASM) to examine how four general extensions to the representation of the basic land surface energy balance affect simulated land-atmosphere interface variables: evaporation, precipitation, skin temperature and air temperature. The impacts of including separate surface energy balance calculations for: vegetated and non-vegetated portions of the land surface; an explicit parameterisation of canopy resistance; explicit bare ground evaporation; and explicit canopy interception were isolated and quantified (see Table 1).

TABLE 1

Table 1 Summary of the differences between the CHASM modes indicating which of them include explicit parameterisations for canopy interception, bare ground evaporation, canopy resistance and horizontal temperature differentiation. Also shown are the codes for the four differences obtained from the five experiments used in subsequent figures.

This work was conducted to explore the hypothesis that these aspects of surface energy balance parameterisation do not contain substantial information at the monthly timescale (and are therefore not important to consider in a land surface model) and we were able to show this hypothesis to be false.

Considerable sensitivity to each of the four general surface energy balance extensions was identified in average pointwise monthly changes for important land-atmosphere interface variables. Average pointwise changes in monthly precipitation and land evaporation are equal to about 40% and 31-37% of the global-average precipitation and land evaporation respectively (e.g. Table 2)

TABLE 2:

Table 2 Changes in average monthly land evaporation (mm/mon) resulting from the removal of each of the SEB extensions (see Table 1). All percentages are of global-average evaporation (60oS to 90oN) in the control experiment with SLAM. Statistically significant changes at 95% are shown in bold. The final two columns show the percentage of land points where statistically significant changes occurred at confidence levels of 80% and 95% using a two-tailed t-test.

Average pointwise changes for land surface skin temperature and lowest model layer air temperature are about 2K and 0.9K respectively. The average pointwise change and average pointwise biases are statistically significant at 95% in all cases. Substantial changes to zonally average variables are also identified as well as to the geographical distribution of evaporation (Figure 1a). These changes can be shown to be statistically significant (Figure 2b).

Figure 1(a) Global maps of July-average land evaporation differences resulting from the removal of each of the SEB extensions (defined in Table 1). Only differences greater that ± 5 mm month are shown

Figure 2(b) As Figure 1(a) but showing areas of statistical significant change calculated using a two-tailed student’s t-test calculated point-by-point. Significant areas of change are shaded at 80% and 95% levels.

Overall, this work suggests that a useful basis for assessing whether or not to include an explicit parameterisation of a particular process may be that: (a) it has a substantial effect on the simulation that is not simply due to parameter inconsistency; (b) it does not degrade the overall performance of the model and; (c) the basic effect of the physical mechanism it represents is clear at the spatial scale to which it is being applied. Our finding confirm that the first of these three criteria is satisfied for each of the four SEB extensions we explored and if one chooses to adopt the suggested criteria, the decision as to whether each of them should be included in an LSM should then be based on the second and third criteria. Our findings did not allow for an assessment of whether the SEB extensions satisfy these two additional criteria, but in addressing the first criteria they do go partways towards answering the question of whether or not the processes should be included in an LSM.

The work exploring land cover change has continued. An important facet of research within the CIC is the impacts on the Earth’s climate of increasing carbon dioxide. The role of carbon dioxide in increasing the Earth’s temperature has been the subject of considerable research including the major reports by the Intergovernmental Panel on Climate Change [Houghton et al., 1995]. The body of statistical evidence now points to a discernable human influence on the global climate [Santer et al., 1995], although most of the focus to date has been on the role of greenhouse gases and anthropogenic aerosols as primary causes for the changing climate. The IGBP and the IPCC have recently recognised that regional scale changes in climate can be caused by other Human-induced perturbations, specifically land cover change (LCC) which can affect the regional climate by changing the partitioning of available energy between sensible and latent heat, and by changing the partitioning of precipitation between evapotranspiration and runoff.

Work by Pitman and Zhao has used a climate model to perform simulations which provide the relative impact of the increase in carbon dioxide compared to the impact of LCC. Our aim is to determine whether LCC causes systematic changes in the Earth’s climate of sufficient size (compared to carbon dioxide) to be worthy of inclusion in future simulations of global change. We performed a series of 17-year integrations using the NCAR CCM3 (at about 2.8o x 2.8o resolution) to determine the magnitude of historical land cover change impact on the global scale climate, and on the regional scale climate over the regions where most occurred: Europe, India and China. The change from natural to current land cover was estimated using BIOME3 to predict the natural vegetation type, and then using remotely sensed data to estimate the locations where land cover had been changed through human activity. Results show statistically significant changes in the 15-year averaged 1000 hPa wind field, mean near surface air temperature, maximum near surface air temperature and the latent heat flux over the regions where land cover change was imposed. These changes disappeared if the land cover over a particular region was omitted indicating that our results cannot be explained by model variability. We also found impacts on the global scale climate (see Table 3) which remained even if we increased the carbon dioxide levels in the atmosphere to 355 or 430 ppmv:

Table 3 Percentage of global area showing statistically significant change as a result of land cover change at 80%, 90%, and 95% significance levels using a two-tailed Z-test.

An analysis of changes on an averaged monthly timescale showed large changes in the maximum daily temperature in (northern hemisphere) summer and little change in the minimum daily temperature, resulting in changes in the diurnal temperature range. The change in the diurnal temperature range could be positive or negative depending on region, time of year and the precise nature of the land cover changes.

At a regional scale, we analysed the results over China in more detail. Figure 3 shows the impact of land cover change at various CO2 levels on the surface wind field averaged over December, January, February (DJF). The land cover change clearly affects the direction and strength of the surface winds.

Figure 3 The changes in the surface wind field for DJF resulting from (a) LCC at 280 ppmv; (b) LCC at 355 ppmv; (c) LCC at 430 ppmv, (d) LCC at 505 ppmv; (e) a change in CO2 from 355 ppmv to 280 ppmv; (f) a change in CO2 from 430 ppmv to 355 ppmv; and (g) a change in CO2 from 505 ppmv to 430 ppmv. The CO2 experiments were conducted using natural land cover.

Overall, our results indicate that the inclusion of land cover change scenarios in simulations of the 20th Century may lead to improved results. The impact of land cover changes on regional climates also provides support for the inclusion of land surface models which can represent future land cover changes resulting from an ecological response to natural climate variability or increasing carbon dioxide. This provides an indication of the ways forward we are planning.

Neil J Holbrook, BSc Syd., PhD Syd

INVESTIGATING LARGE SCALE OCEAN PROCESSES USING MAPPED OBSERVATIONS AND SIMPLE MODELS

Investigating the role of Ekman pumping and Rossby wave propagation in the Pacific Ocean using a linear vorticity model

Perkins and Holbrook (2001) have attempted to reproduce the salient features of the variability in the depth of the thermocline in the marginally eddy-resolving Parallel Ocean Climate Model (POCM) of Semtner and Chervin, using a simple linear vorticity model which only permits local Ekman pumping and the propagation of long Rossby waves. The dynamic upper ocean variability in the POCM is examined in response to changes in daily European Centre for Medium-range Weather Forecasting wind stresses across the tropical and subtropical Pacific Ocean (31oS-31oN) between 1983 and 1989. The POCM provides a complete and physically consistent representation of the state of the Pacific Ocean, with the phase of the thermocline depth anomalies being consistent with the observed El Niño/La Niña variations in the near-equatorial zone and southwest Pacific during the decade.

A series of vorticity model sensitivity experiments, incorporating scaled Rossby wave speeds (following a recent study by Holbrook and Bindoff (1999)) based on recent observations from the TOPEX/Poseidon satellite altimeter, is used to examine and compare the phase and amplitude variations in the depth of the internal surface against changes in the depth of the 14oC isotherm (D14, used as a proxy for the depth of the thermocline, or pycnocline) as simulated in the POCM. This study demonstrates that the simple linear vorticity model can reproduce the Pacific Ocean thermocline depth anomalies in the interior of the subtropical gyres as simulated by the POCM. These variations are both qualitatively and quantitatively consistent with an ocean forced by only Ekman pumping and long Rossby waves which traverse the basin, with isolated topographic and background influences. A number of experiments in the study demonstrate that the phase similarities, from correlation analyses, between results from the POCM and those from the simple dynamical model are statistically significant (at the 95% level) across the majority of zonal transects located at 11oS, 11oN and 21oN in the western, central or eastern Pacific basin. At 11o and 21o latitude, the amplitude of the variability is similarly comparable across much of the basin. The model is generally less successful at 31o latitude where higher baroclinic modes of the mean flow become important.

Investigating the stability of the thermohaline circulation using a coupled ocean-atmosphere-sea ice box model

Grigg and Holbrook (2001) have used a box model of the North Atlantic to test the stability of the thermohaline circulation under two different climatic regimes, (i) a warmer regime where deepwater formation occurs through open ocean convection, and (ii) a cooler regime where deepwater formation occurs with the aid of brine rejection from sea ice formation. We found that the brine rejection mechanism produces a more stable modelled thermohaline circulation than the open ocean convection mechanism, i.e., a circulation that is able to withstand a larger high latitude freshwater perturbation. Our results demonstrate that the effects of leads and polynyas on brine rejection rates are important and suggest that the presence of open water within modelled sea ice contributes significantly to the sensitivity of the climate response and cannot easily be ignored.

Investigating the volume transport variability in the southwest Pacific Ocean using gridded observations from DASPOT

Maharaj and Holbrook (2000) have investigated and quantified changes in the southwest Pacific Ocean circulation on annual, seasonal and interannual time scales (between 1973 and 1988), based largely on geostrophic volume transport estimates derived from gridded temperatures in the Digital Atlas of Southwest Pacific upper Ocean Temperatures (DASPOT) (Holbrook and Bindoff 2000a). The geostrophic transports were derived from the application of a series of pre-determined linear regression equations of the depth-integrated steric heights relative to 2000 m depth, P2000, on the temperatures at 450 m depth, T450, as a function of space and time. Annual and seasonal total volume transport (geostrophic + Ekman) budgets were calculated in this study for a boxed region bounded by the east Australian coast. An inverse technique was applied to determine the non-zero horizontal velocities at the 2000 m reference level, causing the initial geostrophic transport estimates to be adjusted by no more than 2-3 Sv across each transect. Climatological monthly and interannual geostrophic volume transports have also been determined and presented at 2o grid scales across zonal and meridional transects. Transport errors derived from the temperature mapping errors in DASPOT (Holbrook and Bindoff 2000b) do not exceed +-1 Sv in any of the sub-transects.

We found that the subtropical gyre circulation in the southwest Pacific appears to be most vigorous in winter, with western boundary transports of 30.2+-0.5 Sv north across 15oS and out of the Coral Sea, 24.9+-0.1 Sv south across 29oS in the Tasman Sea, and 10.2+-0.2 Sv south across 43oS into the Southern Ocean, which recirculates into the Tasman Sea. This anticyclonic recirculation across 43oS occurs only in winter and spring. The East Australian current is also shown to be variable and has a semi-annual cycle. Interannual geostrophic transport estimates indicate year-to-year changes of about 4-12 Sv (per 2o grid) including El Niño – Southern Oscillation and quasi-biennial time scale variations observed in the subtropics and mid-latitutudes. Southward propagation of interannual transport anomalies are observed quite clearly along 155oE, between 20oS and 30oS, the time scales of which are consistent with gyre strengthening/weakening and the westward propagation of long Rossby waves.

Ann-Maree Graham, BSc(Hons) (PhD)

TRENDS IN SCIENCE: THE SUNSPOTS AND ENSO CASE STUDIES

This work is of an interdisciplinary nature with an historical emphasis. The research explores aspects of the physical and social components of both the sunspots and ENSO phenomena. The social component collates and uses bibliometric data (entered into a coded database) to measure the (1) interest over time of the case study and (2) foci areas of interest within the case study area.

For each case study, wavelet analysis is employed, using the physical data and the bibliometric data. Wavelet methods are used to identify the signals/cycles in Sunspots and ENSO data. The results from the physical data analysis concur with much research conducted in this manner. An historical and bibliometric study is conducted in an attempt to measure the growth, trends and areas of foci in each case study. A database for each of ENSO and Sunspots was created, cataloguing items published in serial literature on each subject. In addition data was gathered from three specific journals to enable analyses of a `stable' data series. Results of the wavelet analysis indicate some periodicity in the publications data as well as both quiet and productive times. The changes reflected in the data are a result of wide-ranging factors, both internal and external to the study area.

One such factor is the influence of the physical cycle on the patterns of publications. This study also takes a preliminary look at the relationship of the findings above to the actual physical manifest of the phenomena. The preliminary results show evidence of a relationship between the physical cycle and the publication output. For example, sunspots publications over the longer term tend to follow the sunspot cycle. Interestingly, the peak in publications on sunspots tends to be around sunspot maximum or just prior. The results are yet to be published.

The focus of this interdisciplinary study is unusual, as very little wavelet analysis has been used in bibliometrics (or scientometrics). Such a study can serve to provide: an objective measure of a particular subject; can also measure and monitor changes in the direction of research in subject areas; provides a useful database for other researchers; historically it can reflect the contemporary interest in the subject and it can indicate points in time where distinct changes in foci occur.