Temperature reduction and local last glaciation maximum (LLGM) The example of the east-Andean Cordillera around Cochabamba, Bolivia (17°S)

geomorphologic and pedogenic evidence


Introduction
The Central Andes of South America (16°S-30°S) experienced major climatic changes during the late Quaternary, making this high altitude mountain region a key area for the study of processes and dynamics of tropical and extratropical climate variability.In the International Geosphere-Biosphere Program (IGBP), the Pole-Equator-Pole transect of the Americas (PEP I) has been analysed in detail by PAGES (Past Global Changes).The synthesis points to the climatic sensitivity of this region (Markgraf et al. 2000).The challenge is to disentangle local and regional influences from hemispheric or global influences and processes of past climate variability (see Kull & Grosjean 1998).
Ice cores (Thompson et al. 1998; Thompson et al.   2003), lake Sediments (Abbott et al.  provide quantifiable data for these environmental changes which occurred during the late Pleistocene and Holocene.The derived spatial and chronological pattern of climate variability in the western cordillera of northern Chile shows dynamical links to the extra¬ tropical westerly circulation south of 27°S and to the tropical influence in the northern part (Kull et al.

2002).
The eastern cordillera north of 27°S, situated close to the Amazon and Chaco basins, is mainly influenced by tropical circulation.There is still an ongoing debate about the chronology, the geographical extension and the related magnitude of late Pleistocene -Holocene environmental changes (Baker et al. 2001; Mourguiard & Ledru 2003; Sylvestre et al. 1999).
Due to the aridity of this high elevation mountain ränge, morphological structures of former lake shore- lines and glacial moraines, for example, remain well preserved.An exact reconstruction of the past envi¬ ronment is possible.Actual annual preeipitation in the center of the Arid Diagonal (Fig. 1) is about 100 mm, with a trend to higher values towards the north, east and south.This indicates that changes in the hydrologi¬ cal cycle are as important as changes in the temperature regime.Recent and past glaciological conditions con- firm this fact, with missing glaciations in the extremely dry Arid Diagonal and above the continuous perma¬ frost belt (above 5500 m asl) being explained by lack of preeipitation.Glacier dynamics, therefore, is mainly controlled by changes in the hydrological cycle (Kull et al. 2002).In the wetter areas north, south and to the east of the most arid zone, where annual preeipitation reaches values of about 1000 mm as, for example on the eastern slope around Cochabamba, temperature (besides preeipitation) plays an important role triggering glaciations.In this area, actual perennial snow cover and glaciers are limited to peaks above 5000 m asl.The different spatial sensitivity to moisture and/or temperature changes of glacier dynamics also influ¬ ences the chronology of local maximum glacier extent in the region, making it obvious that glaciers in the Central Andean area did not reach their maximum extent at the same time in all the areas (Kull et al.   2002).Large moraine Systems in the region are the residual morphological features of former glaciations.
The maximum extent was reached with a lowering of the equilibrium line altidudes (ELA) by as much as 1400 m (Mark et al. 2005).
During the past years, several studies focused on the detailed reconstruction of past glaciations in the central Andean region (Ammann et al. 2001; Kull &  Grosjean 2000; Kull et al. 2003).Exposure age dating has been applied and is still in progress in order to establish a chronology of glacier advances in many parts of the Central Andes (Zech et al. 2006b).Furthermore, detailed studies using the mapped glaciers and subsequently applied mass balance modeis were successfully carried out to quantify the related past climatic conditions (Kull et al. 2002).These activi¬ ties concentrated on the western Chilean cordillera and the region of north-western Argentina, thereby encompassing the southern Altiplano (Fig. 1).Evi¬ dence for massive changes in the effective moisture regime during late glacial times due to an enhancement of the tropical circulation was found especially in the arid western cordillera between 18°S and 25°S (BETANCouRTet al.2000; Clayton & Clapperton 1997; Grosjean et al. 1995; Kull & Grosjean 2000; Placzek   et al. 2006).In contrast, the westerly influenced area south of 26°S experienced a substantial temperature depression combined with a moderate increase in pre¬ eipitation producing maximum glaciation during füll glacial times, before and after the LGM (last glacia¬ tion maximum, 25-15     Lower panel: the area has been divided into three parts to allow for the influence of mean moisture flux from the east lo be taken into consideration.
The black Star in the detailed map shows the paleoglacier used for 3D-modeling.
Model results of research on north-western Argentina also indicate that maximum glaciation was triggered by massive temperature reduction and moderate humid¬ ity increase (Kull et al. 2003)   In this study, we employ a glacier climate model to test areas around Cochabamba in order to get a detailed picture of past glacier-climate dynamics on the central eastern Andean slope.
2 Study area and present climate 2.1 Study area The research area (Fig. 1) is located in the Cordillera de Cochabamba on the eastern slope of the Cordillera Oriental at 17°15'S, 66°15'W.Today, the study area is free of ice, but both satellite imagery and field work evidence large moraines triggered by at least one mas¬ sive glacier advance. 17paleoglaciers, covering an alti¬ tudinal ränge between 3700-5000 m asl, were mapped in detail to obtain glaciological key parameters such as paleo-equilibrium line altitudes (paleo-ELA).accu¬ mulation area ratios (AAR), hypsographic curves and bed topography.Not only valley glaciers.but also ice caps characterized the landscape during the LLGM as the morphology of this mountain area confirms.
Exposure age dating was applied to some moraines in our study area and is described in Zech et al. (2006b).Some of the moraines were dated to 12 and 10 kyr B.P. These dates suggest a late deglaciation on the eastern flank of the Cordillera Oriental.reflecting results from earlier studies (Abbott et al. 2003).
Paleo-ELA's of the last local glaciation maximum (LLGM), evidenced by the top onset of the lateral moraines, were situated between 4350-4550 m asl in the central and western part of the research area and at about 4250 m asl on the eastern slope (Fig. lj.This significant difference between the eastern and western sections of the study area points to the important role of moisture advection from the eastern lowlands and the spatial pattern of preeipitation.The AARs of the former glaciers vary between 0.59-0.94.Whilst high AARs (small ablation area, i.e. short tongues) point to a wet climate with a high balance gradient in the ablation area.low AARs (around 0.6) are character¬ istic for cold climates and glaciers with long tongues.
The observed difference in geometry and AAR may be an indication of different phases during past glacial advances.Research on this aspect is still ongoing.

Present climate
The present climate is dominated by dry winters and wet summers with convective preeipitation originat¬ ing from the tropical lowlands toward the east (Fig. 1).
Local climatic data was obtained from a temperature Station and eight preeipitation gauging stations that are scattered around the study area at different altitudes (Fig. 1).Daily climatic data, although incomplete.was available from the climate Station Jana Mayu (2002, 3770 m asl, Fig. 1, Table 1).
Extrapolation of the climatic data for the formerly glaciated areas suggests annual preeipitation between 780-910 mm and 7.0°C mean annual temperature at 3500 m asl.A summary of present climatic key parameters is given in Table 1.
Daily mean temperatures were calculated on the basis of daily minimum and maximum tempera¬ tures.These values differ by +0.65°C from the «real» values, calculated from continuous measurements (Jana Mayu).This discrepancy has been corrected in the mean annual temperatures mentioned above.
Free atmosphere lapse rates are derived from radiosonde measurements at La Paz between 4050-5800 m asl (National Climatic Data Center, NCDC 1980).
Monthly mean values vary from -0.64°C/100 m in December to -0.42°C/100 m in June and show a sinusoidal trend.Values are within the ränge mentioned in Kalnay et al. (1996).Regional ground-based lapse rates were calculated from temperature sta¬ tions between 2500-3700 m asl and show significantly higher values (-0.87 to -1.07°C/100 m).This is at least partially caused by local climates, lack of information about the exaet location of stations (Kageyama et al.  2005) and the different altitudinal ranges considered.
For this reason, two different scenarios were used for the modeling process.
Preeipitation gradients were calculated from the eight preeipitation stations in the study area.It should be noted that no significant positive preeipitation lapse rate could be observed (multiple regression analysis).
On the contrary, calculated values varied between -3 mm/100 m in the central and western parts of the study area and -1 mm/100 m in the eastern part.As a result, a value of 0 mm/100 m was selected for the pre¬ eipitation lapse rate in the model (    Kull (1999).The model is based on the actualistic principle.a) present climatic parameters such as diurnal and annual cycles, amplitudes and lapse rates (Table 2), b) 3D-mapping of the former glacier geometry based on field measurements and remote sensing data (Fig. 1), c) empirical-statistical Sublimation, melt and accumu¬ lation modeis (Table 3).
Furthermore. the model assumes that former glaciers fulfilled the following two conditions: I.The glaciers reached equilibrium (mass balance over the whole glacier 0!).A 2D-modeling ap¬ proach that fulfils this condition results in several possible temperature-preeipitation scenarios.II.Annual mass flux through a cross section in the ablation area is equal to annual mass balance below this cross section.With respective dynamical mass flux calculations derived for different cross sections, only one temperature-preeipitation scenario fulfils both conditions.
The 2D-mapping is based on field work and a 30 m and 90 m digital terrain model (DTM), whereas 3D-geo- metrical reconstruction of the glacier bed is based on field measurements only.With the three sub-models (accumulation, Sublimation, melt), the mass balance is calculated for each altitudinal belt (100 m equi- distance).By changing the climatic key parameters cloudiness (C) and temperature (T) and the depend¬ ent parameters preeipitation (P).relative humidity (RH) and global radiation (G), the calculation of a climate scenario is done iteratively (Fig. 2).For all modeled scenarios.the amount of winter rain appears to be the same as today and varies between 100 mm and 150 mm.depending on annual preeipitation.Con¬ sequently.for the scenario we assume that the main increase in annual preeipitation is caused by stronger summer preeipitation, a point that is also supported by field evidence (paleo-ELA distribution).
In a 2D-modeling approach (area-elevation distri¬ bution).only condition I (see above) has to be ful¬ filled.This leads to several temperature-preeipitation scenarios.Results are especially sensitive to areaelevation distribution and vertical balance gradient (kg/(m2*m)).Reconstruction of the former glacier sur¬ face in the ablation zone is done manually as the ice surface is given by the height of the lateral moraines.
In the accumulation zone.a model (Sailer et al. 1999)  is used to approximately calculate the paleo-ice sur¬ face.Especially in complex Valley structures where the actual ice thickness is often under or overestimated.modeled surfaces are modified manually.For the global radiation input, a model approach is used (Mölg 2002: Mölg et al. 2003).Incoming shorlwave radiation is modeled every 30 minutes for each pixel (30 m x 30 m) of the reconstrueted glacier surface as a function of latitude.exposition, shading.altitude and cloudiness.In the 2D-modeling scenario.all 17 glaciers Modern climate: Parameters (C.P, T, RH, W, G and DDF) => amplitudes (day, year).average values.lapse rates => modeis for melt, Sublimation and accumulation under modern conditions In order to achieve equilibrium => iterative process Fig. 2: Flow chart of the model: C (cloudiness).T (temperature), P (preeipitation), RH (relative humidity), G (global radiation) and W (wind velocity) The degrec-day factor (DDF) varies with changing paleoclimatic conditions (Kull 1999).C and T are the key parameters ofthe model and therefore control the other climatic factors.The Ihree submodels are based on climate transfer funetions calibrated under present conditions (Kull 1999;Table 3).The mean annual values ofC andT are changed ileratively in order lo reachzero mass balance and ELAs close to observed ELAs in the field.Schemel eles Modellierungsablaufs: C (Bewölkung), T (Temperatur), P (Niederschlag), RH (Relative Feuchte), G (Globalslrahlitng) und W (Windgeschwindigkeit) Schema de la modelisation: C (nebulosiie), T (temperature), P (preeipitation), RH (humidiie relative), G (radiation globale) el W (vitesse du vent) CLIMATE: Temperature ""  'Budyko (1974); Hastenrath (1984); Mölg (2002); Mölg et al. (2003)   Tab.2: Climatic parameters used in the model for daily and annual cycles and amplitudes Modellparametrisierungen für die Tages-und Jahresgänge sowie Amplituden  l).The first approach makes use of radiosonde measurements from La Paz, Bolivia, the second approach uses calculations based on local climatic data adjusted to take higher lapse rates during the dry season into consideration.This adjustment follows results of temperature reconstruetions during LGM (or generally colder periods) in South America, which indicate that a higher tem¬ perature depression was to be found in the highlands than in the lowlands and, therefore steeper lapse rates.
This effect is typical for a dryer atmosphere during colder periods (Farrera et al. 1999; Kageyama et al.  2005; Pinot et al. 1999).For this reason, in the second approach, the assumption is made that the lapse rates during the dry season are higher.
In order to use the 3D-modeling approach (flow behavior included).at least two detailed cross sections in the ablation zone of the former maximum glaciation are necessary.Polished rock surfaces at the base of the cross sections are ideal features, as they mark the former ice basis.Unfortunately, only one glacier (Wara MASS BALANCE: Sublimation a subldh -1.33 + 0.12(Wdll) + 0.24 (Dddh) + 0.27(Gdh) with: hourly mean temperature [mm ]   [°C] Specific annual mass balance bh 2, ~{subld h + meltd h) + accd h d [mm a '1]   jVuille (1996) b Kull (1999) Tab.3: Empirical-statistical modeis developed for mass balance modeling Empirisch-statistische Modelle zur MassenbilanzmodelUerung Modeles empirico-stalistiques developpes dans la zone de recherche pour la modelisation du bilan cTequilibre Quelle: Kull et al. 2002   Wara.black star in Fig. 1 and Fig. 3) fulfilled all above- mentioned requirements.allowing for the application of dynamical ice flow calculations.Two cross sections below the paleo-ELA were measured in detail during a field trip in May 2005.A summary of the results and the uncertainties in this connection is listed in Table 4. Uncertainty in the depth of a cross section can only be positive because polished rock surfaces were found at the deepesl point and therefore point to a former ice basis.Thus uncertainty arises only from underes- timation of the former ice thickness (erosion of lat¬ eral moraines).The other uncertainties are related to errors in field measurement such as the reconstruction of the former glacier surfaces.slope (sin a) and estima¬ tion of the flow parameters.Glacier flow is calculated after Oerlemans (1997) and Budd (1969)  p: ice density (900 kg/m') g: gravity (9.807 m/s:) The calculation of the form factor «F» is done after Budd (1969) The large lateral moraines define the former ablation area in detail.
Le paleoglacier Wara Wara utilise pour la modelisation 3D.A gauche: distribution etendue-aliitude (ligne con¬ tinue) et etendue du glacier cumulee avec /'altitude.A droite:photo de la vallee anciennement glacee.Source: NASA landsat7_233_72_tif_panchromatic The average ice flow velocity through a given cross section allows for the calculation of the mass flux by multiplying the velocity with the cross section.In the end, the calculated climate scenario has to estimate the same negative mass balance value below the cross sec¬ tion as mass flux through the respective cross section. 4Results

Recent glacio-climatological conditions
Today, the study area is not glaciated and a recent ELA has not been observed in the field.However.both sat¬ ellite imagery and field work (for example, on huge moraine Systems) provide evidence of large glaciations in the past.Snow fields on the south face of Cerro Tunari (about 5050 m asl), as seen on a field trip end of May 2005 indicate that a modern ELA would probably be not much higher than the highest peaks in the research area.Mark et al. ( 2005) calculates recent ELAs in the Cordillera Oriental in Bolivia at altiludes around 4980-5200 m asl (based on data from 1991-1995), the geographically dosest being at 5050 m asl (Quimsa Cruz, 17.06°S.67.24°W) and 5100 m asl (Laguna Kollpa Kkota, 17.43°S, 67.88°W).This information served to test the modeis ability to estimate present glacio-cli¬ matological conditions in the research area.Based on local climatic data (Table 1,1960-80) modeled modern ELAs varied between 5150 and 5250 m asl, depending on exposition, amount and annual distribution of pre¬ eipitation.The modeled results correspond with those observed in the field.The calculated present annual and seasonal balances are shown in Fig. 4. The ability of the model to calculate present values correctly made it possible to use the same model for assessing past gla¬ ciation events.

2D-Results
In the 2D-modeling approach, all 17 glaciers were investigated.Results derived with the higher lapse rate scenario (Table 1) are summarized in Fig.  Uncertainties: 1 +5 m due to moraine erosion 2 ± 5% (uncertainties in surface reconstruction) 3 uncertainties in ", 2,and ± 20% flow parameters Tab.4: Geometry and mass flux in the cross sections (ablation zone) of the paleoglacier Wara Wara (Fig. 1) Kennzahlen der Querschnitte im Ablationsgebiet des ehemaligen Gletschers Wara Wara (Fig. 1) Geometrie et flux de glace dans les coupes transversales (zone d'ablation) du paleoglacier Wara Ware (Fig. 1) between the glaciers in the east (ELAs at 4250 m), west (ELAs at 4550 m) and middle (ELAs at 4350 m) of the research area.Assuming synchronous glacier advances, these results can be interpreted as follows: ff preeipitation is taken as a constant over the whole study area, the temperature depression is highest in the western part and lowest in the eastern part (dif¬ ferences of up to 2.5°C).In this small area, this type of temperature pattern seems very unlikely.
-If a uniform temperature decrease over the study area is assumed, the eastern part would need a much higher increase in preeipitation than the middle and western parts (plus 300-800 mm annual preeipita¬ tion).
Modeling results with a lower lapse rate lead to simi¬ lar conclusions.Temperature-precipitation-«solution areas» (Fig. 5) generally move downward by about 0.5°C It is thus clear that there is no linear relation¬ ship between T and P. The Solution scenario lines become less inclined with the change from colddry conditions to warm-wet conditions.Therefore, at [dT= -9°C; dP= -200 mm], a further decrease in preeipitation of 100 mm requires a decrease in tem¬ perature of about 3°C.In contrast, at [dT= -6°C; dP= +200 mm], an increase in temperature of about 3°C has to be compensated by an increase in preeipitation of more than 700 mm! Glaciers in the lower left of Fig. 5 reacted very sensitively to changes in preeipita¬ tion because it was cold enough.Glaciers in the upper right, on the other hand, reacted especially sensitively to changes in temperature because the climate was wet enough.This pattern can be explained by the fact that under cold climatic conditions.preeipitation falls exclusively as snow: in a generally warm climate.small changes in temperature may have a huge effect on accumulation as preeipitation falls either as snow or rain.

3D-Results
As mentioned above, only one paleoglacier fulfilled all requirements, enabling the exaet computation of its flow behavior (Wara Wara, Fig. l).Fig. 6 shows the bed and modeled surface topography along its central flowline.The 3D-modeling approach (Table 5) shows simi¬ lar results for both lapse rate scenarios.It is evident that a massive temperature decrease with annual preeipita¬ tion reflecting present values led to the observed local last glaciation maximum at Wara Wara.The calculations respond very sensitively to small changes in slope and flow parameters.These uncertainties are listed in Table 5. Fig. 7 shows the annual mass balances of the paleo¬ glacier Wara Wara with regards to mean.maximum and minimum ice flow through the calculated cross sections (±20% flow parameters, ±2.5 m thickness,±5% slope) with the lower lapse rate of -0.53 ± 0.106 °C/100 m.The maximum ice flow scenario is characteristic of a «cool-wet» climate (dT= -5°C; dP= +910 mm) whereas the minimum ice flow scenario reflects «cold-dry» cli¬ matic conditions (dT= -7.65°C; dP= +60 mm).There are clear seasonal differences in the computed mass balance (Fig. 8).During winter (April-September), net ablation is calculated over the whole glacier, even at elevations higher than 5000 m asl.Intensive solar radia¬ tion in combination with a dry windy climate leads to intensive Sublimation.Respective values are higher than the accumulation of about 120 mm (most preeipi¬ tation is deposited as snow).The low balance gradient in the ablation area reflects well the cold.dry climate.During summer (October-March).net accumulation dominates the glacier.In the annual mean the ELA is situated at 4250 m asl.
Basal shear stress was calculated for the upper (Ql. Table 4) and lower cross section.the values being between 0.773-0.855bar, and 0.856-0.946bar respec¬ tively.These values are not typical for either maritime  The grey vertical line shows a mass balance of zero.The solid horizontal line represents the current ELA at -5190 m asl. the dashed horizontal line the paleo-ELA at -4250 m asl.Present climatic conditions at 3500 m asl are 7.0°C and 780-910 mm annual preeipitation.
After application of these results to the whole test area (Fig. l), taking isochronous and iso-thermal conditions for LLGM into account, the conclusion can be made that in comparison to today, preeipitation was similar in the west¬ ern part of the study area.In the middle part of the study area preeipitation was greater by about 100 mm. and in the eastem part.the difference was close to 200 mm.This reflects a uniform temperature depression of about 6.5°C and con- firms a NE-SW preeipitation gradient during the LLGM.
5 Paleoclimatic implications -discussion Kull et al. (2003) show that glaciers in areas with annual preeipitation values above about 800 mm are mainly con¬ trolled by temperature.In areas with less preeipitation, humidity becomes the dominant parameter.Consequently.the varying chronology of glacier advances in the South American Andes can mostly be explained by their geo¬ graphical location with respect to the atmospheric circula¬ tion Systems.For example. in the dry western cordillera of northern Chile and southern Peru, glacier advances were caused by an increase of moisture supplied by the easterly tropical circulation during late glacial times (around 15-10 kyr B.E).During the LGM (around 20 kyr B.P.). the cli¬ mate in the Atacama Desert was too dry to allow massive glacier advances (Grosjean et al. 2001: Kull & Grosjean 2000: Kull et al. 2002).In contrast.in the eastern Cordill¬ era Oriental.massive glacier advances occurred during the LGM.Therefore.even under cold LGM conditions, the moisture supply was high enough to enable glacier advances (Seltzer et al. 2002).
The conclusions made here are in agreement with recent exposure age dates of moraines in the same study area (Zech et al. 2006b).the results of Kull et al. (2003)   for north-western Argentina.glaciers substantially advanced at about 20 kyr.minor advances occurred around 12 and 10 kyr B.P.. Glacier advances at about 20 kyr B.P. (LGM) point to a tem¬ perature sensitivity of glaciers on the eastern slope of the Cordillera Oriental.Moisture supply from the east¬ ern lowlands seems to have been sufficient even during cold phases.The modeled climate scenario with a mas¬ sive cooling of about -5.0 to -7.65CC also suggests that __. with a lapse rate of -0.7 +/-0.15°C/100 m The areas show possible temperature-preeipilalion combina- tions that lead to zero mass balance for the paleoglaciers.The different grey scales match with the three zones Wesl(bright).
Middle (dark).and East (medium) inlo which the research area has been divided (Fig. 1) and represent average results.
It shows that the warmer the climate.the higher the preeipi¬ tation has to be to keep the glacier at zero balance and to produce an ELA at the observed altitude.
Furthermore. the results show that moisture and con- vection still reached the eastemmost parts of the Cor¬ dillera Oriental during cold periods.This is also shown by Placzek et al. (2006).They dated a shallow lake phase between 24-20.5 kyr B.P. in the southern Alti¬ plano of Bolivia.The west-east gradient in preeipita¬ tion was stronger than today and suggests a dominant north-eastern advection of moisture to the research area.This implies that summer humidity.like today.was the main source of preeipitation.The steeper gra¬ dient may point to lower condensation levels and/or a weakened easterly upper tropospheric circulation preventing convection and moisture from flowing as often as today over the first mountain ridges.This would explain why only the western part of the study area is influenced (see Fig. 1, 2D-results).A weaker season- ality at 10 kyr B.P. (Bush & Silman 2004) could also have softened easterly summer moisture advection.A higher frequency of south-easterly cold air outbreaks («surazo» in Bolivia) with a subsequent enhancement of winter preeipitation in the lowlands is not expected to have significantly influenced the research area due to the fact that these cold air masses do not reach the east Andean highlands.Uncertainties: T(°C) P(mm) Ice thickness + 2.5 m1 Flow parameters ±5% «Mean» is taken as the average of maximum (5 m) and minimum cross section that was measured.-Considering Single uncertainties. the effect on the Solution scenario is Symmetrie.Sum- marizing all values to a total uncertainty makes the Solution scenario asymmetric due to the non-linear temperature-precipitation-relationships (see Fig 5).
Tab. 5: Results of the dynamical ice flow (3D-modeling) of the paleoglacier Wara Wara Resultate der 3D-Modellierung eles ehemaligen Gletschers Wara Wara Resultats de la modelisation 3D des flux de glace du paleoglacier Wara Wara The three graphs show modeled climate scenarios with mean (solid), maximum (dashed) and minimum (dotted) ice flow through the defined cross sections.All scenarios lead to the same ELA.The horizontal grey line is the paleo-ELA at 4250 m asl.
Modellierte Bilanz-Höhen-Verteilung des ehemaligen Gletschers Wara Wäret Modelisation de Tequilibre-aliitucle du paleoglacier Wara Wara ern slope of the Cordillera Oriental in Bolivia.Their results indicate a relatively dry LGM. a wet late glacial period and drier climatic conditions at the onset of the Holocene.This climatic history may explain why dif¬ ferent glacier advances are associated with different glacier geometries.Mapping of former glacier extents (Fig. 1) points to a broad ränge of observed A AR's and various advances indicated by different moraine stages.This agrees with maximum advances under «cold-dry» (low AAR) conditions during the LGM, followed by late glacial «warm-wet» advances (high AAR) as also observed in north-western Argentina (Kull et al 2003).
Chronology. geomelry and moraine stratigraphy in the region therefore need further attention in order to be able to discern which glacier advances were driven by what climatic conditions and when.

Conclusions
A glacier-climate model was used to reconstruct paleoclimatic conditions that triggered the local last glaciation maximum on the eastern slope of the Cordillera Oriental near the city of Cochabamba (17°15'S/66°15'W).Bolivia.The results point primarily to a temperature sensitivity of the glaciers in the study area.In order to reach the maximum extent.the most likely paleoclimate scenario suggests that annual mean temperatures were lower by 6.5 (+1.4. -1.3)°C while annual preeipitation was about 300 mm higher than today.Humidity was delivered as today by north-east- erly advection of moist air in summer to the research area.The steeper west-east gradient in the past prob¬ ably suggests a lower condensation level in the atmos¬ phere. in agreement with the calculated massive tem¬ perature depression.Low shear stress values below 1 bar and an AAR of 0.59 confirm that the maximum glacial advance was not primarily driven by a preeipi¬ tation increase under a «warm-wet» climate scenario.
On the contrary, the results suggest a «cold-humid» cli¬ mate as most likely for the observed maximum glacier advances in the region.The solid line is the calculated annual balance with mean ice flow (see Fig. 7).The differences between the winter and summer balances are substantial.This is an indication lhat during winter.the whole glaciated area is influenced by ablation.Further. the low winter balance gradients of -1.5 kg*m :*m ' indicate «cold-dry» (Sublimation dominated) climatic conditions.whereas the summer season is characterized by «wel-warm» (melt intensive) climatic conditions.
Modellierte saisonale Bilanz-Höhen-Verteilungen für Wyrwoll, K.H., Dong, B. & R Valdes (2000): On the position of southern hemisphere westerlies at the last glacial maximum: an outline of AGCM Simulation results and evaluation of their implications.-In: Qua¬ ternary science reviews 19,9: 881-898.Zech, R., Kull  This study presents results from a glacier-climate model that reconstructed glacio-climatological conditions during the last local glaciation maxi¬ mum (LLGM) in the Cordillera to the north of Cochabamba (17°15'S, 66°15'W), Bolivia.Results emphasize the temperature-sensitivity of glaciers on the eastern slope of the Cordillera Oriental.Maxi¬ mum glacier advances appear to have been caused by a massive cooling of about 6.5°C while annual preeipitation was about 300 mm higher than today (850 mm/yr).Modeling results indicate maximum glacial advances during cold phases such as MIS 2 (25-18 kyr B.P.) and minor advances during late gla¬ cial cool events (12-10 kyr B.P.).This chronology is supported by exposure age dating results.Further evidence may be found in the low AAR-values (accu¬ mulation area ratio) which indicate low mass balance gradients and therefore cold climate conditions.Mod¬ eled basal shear stresses smaller than 1 bar exelude extremely «cold-dry» or «warm-wet» conditions.The spatial pattern of regional paleo-ELA's (equilibrium line altitude).with higher ELAs in the western part of the study area, reflects a strong east-west gradient in paleoprecipitation.Easterly summer preeipitation is suggested to be the reason for this phenomenon.
These results are in agreement with other studies of the east-Andean slope.indicating temperature as the driving factor for maximum glacier advances in northwestern Argentina.
5. All possible temperature-preeipitation scenarios that fulfill condition I of the model are shown.A clear differ¬ ence in reconstrueted mean climatic conditions exists

Fig. 4 :
Fig. 4: Present annual and seasonal balance-elevation distribution in the highlands around Cochabamba
).In the east¬ ern cordillera on the slope to the Amazon and Chaco basins, several glacial advances are mapped and dated to pre-LGM and late glacial times(Clapperton et al.  1997; Heine 2000; Schäbitz 2000;
T,i°C]:Tih= Ym + Ya (cos d) + (Da + C *) cos t + grad,, +P* gradh: lapse rate (h/100m)ts: hours in summer P*: preeipitation correction k: empirical coefficient C: cloudiness (in tenth parts) d: days in winter (Apr-Sep) [/(t)] Wm: winter mean value t: hours d,: days in summer (Oct-Mar) [/(t)] Wa: winter amplitude h: altitude d: days [/(t)] S0: solar constant (t370Wm'2) E0: excentricity correction factor C,: zenith angle of the sun to an inclined surface T: linke turbidity factor ' S.card et al. (1998); Wagnon et al. (1999) Parametres climatiques utilises dans le modele des cycles et amplitudes journalieres et annuelles Quelle: Kull et al. 2002 are investigated.Two different temperature lapse rates are used in order to counteract the sensitivity of the model to this important parameter (Table Verteilung in der Cordillera nördlich von Cochabamba Repartition annuelle et saisonalere actuelle de Tequilibre-altitude dans la Cordillere au nord de Cochabamba or Continental type glaciers given the altitudinal ränge of the glacier and those of Seltzer et al. (2002) for Bolivia.Following Zech et al. (2006b), Abbott et al. (2003)analyzed several sediment cores from lakes in the Cordillera The results presented here from the Wara Wara paleogla¬ cier suggest that maximum glacier advances on the east¬ ern slope ofthe Cordillera around Cochabamba occurred during füll glacial times of MIS 2 (25-18 kyr B.P). Subsequent minor advances occurred likely during relatively warm and especially wet periods in late glacial times.