isotopomers, whereas thermodynamical fractionation processes maintain a constant ratio between them.
The leads to the deﬁnition of a second-order isotopic signal, referred to as deuterium excess d (Equation 1.6).
D = δ D − 8 × δ18 O [◦/◦◦ ] (1.6) Due to the characteristics of isotope fractionation, stable water isotopes represent a major climate proxy.
The pioneering work by Dansgaard  reveals experimental and theoretical evidence about the stable water isotope ’effects’, as discribed above.
A detailed description of the isotope effects is given in Sturm et al. [2005b, cf.
For a comprehensive description of stable water isotopes in the climate system, the reader is conferred to Mook et al. [2001, cf.http://www.iaea.org/programmes/ripc/ ih/volumes/volumes.htm].
Thereafter, we focus on the stable water isotope signal in ice-core archives.
The longest ice-core records were recorded from the Antarctic ice-shield.
The Vostok ice-core [Petit et al., 1999, Jouzel et al., 1987] uncovered the climate history of the past 420 000 years, later extended by the Dome C ice-core [EPICA, 2004] to 740 000 years.
Several ice-cores were recovered from the Greenland ice-core [NorthGRIP, 2004, Johnsen et al., 1995, Johnsen et al., 2001], reaching down to 123 000 years B.P.
In the case of polar climate, the isotopic composition of precipitation is predominantly controlled by temperature.
Hence past temperature variations can be estimated from the isotopic signal in polar ice-cores.
Yet the precision and accuracy of temperature estimations from ice-core isotopic signal is still investigated [Jouzel et al., 1997, Krinner and Werner, 2003, Krinner et al., 1997, Werner and Heimann, 2002].
While the major stages of the 100 000 year tel-00010157, version 1 – 16 Sep 2005 glacial cycle are recorded in both Arctic and Antarctic ice-cores, the isotopic signal is not rigorously synchronous at both poles.
So, if past climate change witnessed phase differences between both hemispheres, the question remains: which climate conditions prevailed at mid- and low latitudes ? 1.3 Stable water isotopes in tropical South American ice-cores Numerous ice-cores were recovered from mid-latitude glaciers, eg as reported by Schotterer et al. .
Tropical icecores detain a highly valuable information, since they record the climate variability of the inter-tropical belt.
The latter is proved to be particularly sensitive to global climate variations.
Nevertheless, only high altitude sites can offer no-melting conditions in spite of intense zenith radiation at tropical latitudes.
The requirements are met at tropical latitudes in the South American Andes only, where summits commonly exceed 6000 m.
Table 1.2 summarises the locations of 6 tropical and 1 sub-tropical ice-core drilling sites in the Andes.
Several ice-cores records extend back to 20 000 years B.P., ie they comprise the transition from the last glacial maximum (LGM) to the present Holocene inter-glacial stage [Pierrehumbert, 1999, Thompson et al., 1995, 1998, 2000, Ginot, personal communication].
The no-melting condition is essential for preserving the isotopic signal in the ice-core archive.
The energy balance on high altitude / low latitude glaciers show that signiﬁcant sublimation of ice occurs.
However, Stichler et al.  reports that bulk sublimation is a non-fractionating process, hence it does not affect the isotopic signal.
Summit Quelccaya Huascarán Sajama Illimani Chimborazo Tapado Coropuna λ W 70.83◦ 77.62◦ 68.88◦ 67.77◦ 78.83◦ 69.83◦ 72.62◦ ◦ ϕ S 13.93◦ 9.117◦ 18.◦ 1 16.62◦ 1.5◦ 30.18◦ 15.52◦ ◦ z masl 5670 6048 6548 6438 6268 5550 6450 Reference Thompson et al., 1985 Thompson et al., 1995 Thompson et al., 1998 Ramirez et al., 2003 Ginot et al., 2002 Ginot et al., 2001 Ginot, pers.comm. Table 1.2: Location of the Andean ice-core drilling sites. λ stands for western longitude, ϕ for southern latitude [both in degrees ] and z for altitude [in masl – meters above sea-level.
Apart from the sub-tropical Tapado, all sites are located within the inter-tropical belt. ◦– 8 Chapter 1.
Introduction — 71 72 Chapter 4.
South America isotope climatology produces excessive precipitation over the ocean, hence stronger down-drafts and reduced pressure under the ITCZ.
Unlike over the ocean, MSLP in REMO/ECHAM T30 have a generalised negative bias over land.
Despite this underestimation, the relative patterns of MSLP, and thus the gradients guiding the geostrophic ﬂows, are well reproduced in REMO.
In particular, the build-up of high temperatures east of the Andes during austral spring result in a distinct thermal low during summer.
In agreement with the ERA40 data set, the structure (referred to as Chaco low) develops in austral summer (DJF) at the eastern edge of the Andes.
REMO locates it at [60◦ W; 25◦ S], slightly higher North than the re-analyses.
Several modelling studies show discrepancies in MSLP as observed in REMO [Lenters and Cook, 1995, Rojas and Seth, 2003].
The GFDL GCM overestimates by up to 30m the geopotential at 1000 hPa over the ocean, which corresponds to an underestimation of MSLP by up to 3 hPa.
This is in the same range as REMO’s error over the ocean.
Rojas and Seth  analyses the MSLP bias in several RCM runs, which show a similar land/sea contrast.
The bias is larger in Amazon and Southeast Brazil than in the Nordeste.
The bias reaches 4 hPa, with MSLP values lower than re-analyses in austral summer (DJF), and higher in austral autumn (MAM).
Due to the different methods in computing MSLP, no direct comparison is possible between these results and observations with REMO.
Nevertheless, the seasonal variations of the bias is consistent between RegCM and REMO runs.
In general, Christensen et al.  reports MSLP errors by different RCM that commonly reach 3 hPa, with generally negative bias over land.
Under such perspectives, MSLP simulated by REMO is in reasonably good agreement with re-analyses. tel-00010157, version 1 – 16 Sep 2005 The Bolivian High (BH), and its counterpart the Nordeste Low (NL), are characterised by their geopotential heights.
Figure 4.4 shows the vertical cross section at ∼ 12◦ S of geopotential and temperature deviations from their zonal mean.
The Bolivian High builds up during austral spring (SON), to reach its maximum in austral summer (DJF).
As described in Lenters and Cook , the BH is then marked by a 50 m increase in geopotential, centred at 200 hPa and ∼ 65 ◦W .
This is the result of a warm core at 400 hPa, topped by a cold lens at 100 hPa.
The Nordeste low, located at ∼ 20 ◦W has an inverted structure: a cold core at 400 hPa produces a decrease by 40 m in geopotential height at 200 hPa, topped by a warm core at 100 hPa.
Similar to the GFDL GCM used by Lenters and Cook , the warm core simulated by REMO is lower than in the observations (300 hPa).
The latter points out an underestimation of deep convection with REMO’s standard parameterisation, in agreement with the tropical sensitivity run.
In absence of intense convection over the Amazon, the BH-NL vanishes during austral autumn (MAM) and winter (JJA).
The remaining radiative heating over the Altiplano, and associated geopotential anomaly, leaves the 200 hPa wind ﬁeld unaltered.
The mean seasonal geopotential height, wind ﬁeld and divergence at 200 hPa are shown on ﬁgure 4.5.
It is possible to identify during austral summer (DJF) a synoptic maximum of geopotential, which lies 5 ◦ North of the minimum of MSLP at [63◦ W;19◦ S], and marks the Bolivian High.
The BH is further stressed by the anti-cyclonic (counter clockwise) vortex developing around it.
On the opposite, the Nordeste Low, located at [22 ◦ W;20◦ S], is characterised by a 100m decrease in geopotential height, which coincides with a cyclonic (clockwise) wind circulation.
The major domain of divergence at 200 hPa (corresponding to lower troposphere convergence) is located to the North and the East of the BH.
The areas of maximum 200 hPa convergence correspond to high precipitation (as compared to Figures 4.2, 4.7, and 4.8).
Strong convergence extends to the North-East over the Atlantic, joining the Inter-Tropical Convergence Zone (ITCZ).
A second zone of convergence forms at [30-40◦ W;30◦ S], known as the South-Atlantic Convergence Zone (SACZ).
In austral winter (JJA), the geopotential height in the Southern Hemisphere adopts a zonal distribution, where both the BH and the NL vanish.
The wind ﬁeld, in geostrophic equilibrium, is perpendicular to the gradient of geopotential, ie it ﬂows as westerlies.
The major zone of convergence (ITCZ) lies at 10◦ S over the Paciﬁc and penetrates across the Andes into Columbia and Ecuador.
This high convergence is responsible for the heavy precipitation over the Eastern ﬂank of the Andes.
Intermediate seasons (MAM and SON) show the abrupt transition from the winter to the summer mode.
These results are in good agreement with Hastenrath , reporting the location of the Bolivian High at [65◦ W;15◦ S] and similar divergence patterns.
Lenters and Cook  investigate the origin of the BH using the GFDL GCM.
As for REMO, the GFDL GCM locates the BH 5◦ to the South-East as compared to NASA-DAO analyses.
Latent heat release related to high convection over the Amazon is the primary cause of the BH.
In second order, Lenters and Cook  4.3.
Evaluation against observations 73 tel-00010157, version 1 – 16 Sep 2005 Figure 4.4: Vertical cross section of virtual temperature anomaly (shaded, 1K interval) and geopotential anomaly (contour interval 10m, negative contours dashed) from zonal means, averaged from 10 ◦ to 25◦ S.
X-axis represents longitude (degrees), Y-axis the pressure coordinate (100 hPa). 74 Chapter 4.
South America isotope climatology tel-00010157, version 1 – 16 Sep 2005 — 2 (4.1) ∂z(x,y) 2 ∂z(x,y) 2 + ∂x ∂y We select the points above 1500 m, with a mean topographic slope ||∇ · z(x, y)|| > 500m/1000km.
These requirements are met by 346 (84) points from the REMO (ECHAM T106) gridded topography.
The corresponding scatter plot is shown in Figure 4.10.
Annual and seasonal means are ﬁtted with a second order polynomial, to account for the increase of ∂δ18 O /∂z with height.
Positive ∂2 δ18 O /∂z2 is noticeable in ECHAM T106. ∂δ18 O /∂z lies around 0.2 ◦/◦◦ /100 m below 3000 m, and increases towards 0.4 ◦/◦◦ /100 m above 3500 m.
Large scatter exist around the quadratic ﬁt, which accounts for 30% of the variance.
At T106 resolution, no signiﬁcant seasonal dependence of ∂2 δ18 O /∂z2 can be observed.
REMO records a stronger ∂2 δ18 O /∂z2 than ECHAM T106. ∂δ18 O /∂z increases from 0.2 ◦/◦◦ /100 m at 1500 m to 0.6 ◦ /◦◦ /100 m at 4000 m. 46% of the mean annual ∂δ18 O /∂z is captured by the quadratic ﬁt.
Despite of the remaining scatter, signiﬁcant differences appear between the rainy season (DJF) and the dry season (JJA).
During DJF, ∂2 δ18 O /∂z2 is less steep than for the rest of the year.
Austral spring (autumn) displays an intermediate behaviours, recording the dry-wet (wet-dry) transition.
The shape and range of ∂δ18 O /∂z variations agree well with pure Rayleigh distillation, as modelled in Gonﬁantini et al. .
The quadratic ﬁt for DJF resembles the pure Rayleigh distillation, assuming a tropospheric lapse rate of −6.5 ◦C · km−1 and initial relative humidity of 80%.
The JJA ﬁt agrees best with Rayleigh distillation from an initial 60% relative 4.4.
Climatic interpretation of the water isotope signal 83 tel-00010157, version 1 – 16 Sep 2005 Figure 4.10: Variation of the verticalδ18 O gradient (∂δ18 O /∂z, in [◦/◦◦ /100 m]) with altitude (in meters) for REMO (left panel) and ECHAM T106 (right panel).
Altitude gradient is given in absolute values.
Mean annual values are represented as circles.
Mean seasonal values are pictured as follows: DJF – stars, MAM – downwards triangle, JJA – crosses, SON upwards triangle.
Thick lines represent the quadratic ﬁt for annual means and seasonal means.
R 2 indicated in the legend indicates the squared correlation coefﬁcient for the quadratic ﬁt. 84 Model δ18 O ◦ /◦◦ -4.50 -4.09 -3.15 -3.94 -4.05 -3.88 -3.16 -3.97 -6.15 -4.80 -3.93 -3.64 -2.72 -1.33 -1.42 -0.62 -2.04 -3.49 -2.32 -4.71 δ18 O ∆δ18 O /∆λ (◦/◦◦ /10 ◦ ) -1.1 0.13 0.26 -0.024 -1.2 0.32 -0.026 0.0028 -1.5 0.033 -0.014 -0.025 -0.39 -0.082 1.3 0.36 -0.48 -0.94 -0.24 -0.088 2 rδ Chapter 4.
South America isotope climatology Precipitation ∆P/∆λ mm/10 ◦ 46 39 17 19 29 -6.7 31 32 5.7 42 26 13 92 71 57 tel-00010157, version 1 – 16 Sep 2005
Read more about Transition : they comprise the transition from the last glacial maximum LGM….: