A morphometric analysis of terrace gravels in Santa Ynez basin,Santa Barbara County,California

An analysis of 6,300 pebbles from eighteen river terraces within the five subareas of the Santa Ynez River basin has revealed significant differences in particle sphericity and roundness of the various terrace gravels.Using the Cailleux technique of pebble measurement, Krumbein's sphericity ratio formula (Y) and the author's own roundness ratio or the I/L¹ ratio (R) the computed values derived for $\text{Y}$ range from 0.92 to 1.11, and for $\text{R}$ from 0.66 to 0.79 on these benches.An analysis of variance, based on the F-test applied to samples has established that: (a) gravel sphericity and roundness increase significantly from upstream to the river mouth; (b) in the subareas themselves pebble sphericity and roundness vary significantly from bench to bench indicating varied energy conditions of deposition; (c) all terrace gravels measured show higher sphericity and roundness than the underlying Orcutt gravels. This suggests a reworking of the latter by the Santa Ynez River.     

希夏邦马峰东南富曲河谷的冰川沉积和冰川构造

在希夏邦马峰（海拔８０１２ｍ）东南富曲河谷，中更新世以来有三次冰期；即聂拉木、富曲和普罗冰期。它们均可再分为两个亚阶段。聂拉木南的高冰碛平台长３.５ｍ，宽１.５ｋｍ，厚２００ｍ。属于中更新世聂拉木冰期（聂聂雄拉冰期）的巨大山谷冰川沉积，中尼公路从高冰碛平台尾端通过，形成数公里长的冰碛剖面，呈现出美丽多姿的冰川成因类型沉积和冰川构造现象，包括冰下，冰上融出碛，冰内.冰下河道沉积，冰湖沉积，坠碛，流磺等。冰川运动时造成的冰川构造，如断层、滑动面-…等也很清楚，代表了海洋型（暖冰川）冰川沉积和冰川构造特征，是中国目前研究冰川构造最理想的场所。     

New evidence of the total glaciation of the Isla Grande de Tierra Del Fuego

The stratigraphy and preliminary interpretation of a new locality with glacial deposits in a previously interpreted ice-free area in Tierra del Fuego are presented. These deposits consist of basal proglacial gravels (at least 4–5 m thick) covered by the Drift Cabo Campo del Medio. The drift consists of a lower, 2 m thick, lodgment till and a upper, 7 m thick, glaciotectonized composite till, reflecting basal glacier shearing and upthrusting of glaciofluvial deposits and lodgment till, with superimposed deformation by gelifluction. These two main drift packages are interpreted as ground moraines, reflecting two corresponding glacial advances with an intervening recess of the glacier front. The altitude of the Drift Campo del Medio, well above younger fluvioglacial deposits related to the Last Glaciation, the degree of erosional modifications, and its external position, to the NE of the Last Glaciation moraine system, are evidence suggesting that the Drift Cabo Campo del Medio was related to a glaciation older than the Penultimate Glaciation. The finding of this locality sheds new light on the old controversy of total vs partial ice coverage of the island.     

东天山黑沟流域冰川沉积序列及OSL测年研究

黑沟源于东天山最大现代冰川作用中心博格达峰的南坡. 在第四纪冰期与间冰期旋回中,该流域的冰川均发生过多次规模较大的进退,在谷中留下了较为完整的冰川沉积序列. 这些冰川地形包含有重要的古气候变化信息,对其研究可重建黑沟流域的冰川演化史. 应用OSL对该流域的冰川沉积物进行定年,测定结果表明冰水沉积物（沙质透镜体）比冰碛物更适宜应用单片再生剂量（SAR）测年技术进行测定. 基于测得的年龄并结合地貌地层学原理可初步得出：晚第四纪期间,黑沟流域共发生了5次规模较大的冰川作用,分别为全新世期间的小冰期（16世纪以来冷期的冰进）与新冰期（距今3~4 ka的冰进）,末次冰期晚冰阶（MIS 2）与早冰阶（MIS 4）以及倒数第二次冰期（MIS 6）.     

Push moraines in the upper valley of Santa Cruz river,southwest Argentina. Structural analysis and relationship with Late Pleistocene paleoclimate

The upper cliff of the Santa Cruz River was used to assess the proglacial environments of the Argentino Glacier outlet of Late Pleistocene age. These cliffs show glaciolacustrine, fluvioglacial and till deposits, where only the first one are deformed. Glacial landforms in the area and these structures suggest that the ice mass advanced, topographically controlled, towards the east from the Patagonian Ice Sheet pushing up the proglacial sediments.The spatial arrangement of thrusts and overturned folds, the drumlins-flutes moraine directions and the end moraines shape, allow inferring the dynamic and the Argentino glacier profile. Detailed analyses of the glaciotectonic structures indicate that these have two origins: load in the north with stress transfer to the southeast, and push from the west. Through the analysis of deformed sediments, their thickness and their sedimentary and structural features, three zones of deformations were recognized. Each of these zones was associated to glacial advances because of changes of the regional climate conditions.     

The stratotype and facies of the glacial Lower Vendian Nichatka Formation,Chara River basin,Central Siberia

Sediments of the Nichatka Formation are facially studied and thoroughly described, the sections are correlated, and the subformations are recognized. The formation represents a key stratigraphic unit to reveal the origin of the Central Siberian glacial horizon and to correlate it with glacial horizons in other regions of the world, namely, with the Laplandian Horizon of the Lower Vendian, Nantou and Marino tillites, etc. The Nichatka Formation is correlated with the glacial Bolshoi Patom (Dzhemkukan) Formation of the Vendian reference section at the Ura Uplift. Unlike the latter, it is mainly composed of continental glacial deposits and is marked by a complex facies composition. The glacial origin of the Nichatka Formation is reliably determined on the basis of a set of diagnostic characters. This permits a more complete reconstruction of the Early Vendian depositional environments. In addition to typical basal tillites and marginal moraine deposits, the formation includes glaciolacustrine and fluvioglacial sediments along with aquatillites, allotillites, and the glacial fan, including subaqueous, deposits.     

End moraine construction by incremental till deposition below the Laurentide Ice Sheet: Southern Ontario,Canada

Eyles, N., Eyles, C., Menzies, J. & Boyce, J. 2010: End moraine construction by incremental till deposition below the Laurentide Ice Sheet: Southern Ontario, Canada. Boreas, 10.1111/j.1502‐3885.2010.00171.x. ISSN 0300‐9483. Just after 13 300 ¹⁴C a BP in central Canada, the retreating Ontario lobe of the Laurentide Ice Sheet briefly re‐advanced westwards through the Lake Ontario basin to build a large end moraine. The Trafalgar Moraine (27 km long, 4 km wide) is composed of a distinctly red‐coloured silt‐rich till (Wildfield Till, up to 16.5 m thick) formed by the reworking of proglacial lake deposits and soft shale bedrock. The moraine has a pronounced ramp‐like longitudinal form passing upglacier into fluted till resting on exposed shale. Analysis of water well stratigraphic data, drilled sediment cores, downhole gamma‐ray logs and exposures in deep test pits shows that within the moraine the Wildfield Till is built of superposed beds up to 7 m in thickness. These are inferred to result from the repeated incremental deposition of fine‐grained debris being moved towards the ice margin as a deforming bed such as identified at modern glaciers. A total till volume of 0.81 km³ was produced in a very brief time‐span along a transport path probably no greater than 10 km in length. Subglacial mixing of pre‐existing sediment and soft shale was clearly a very effective process for generating and moving large volumes of till to the ice margin. Similar till‐dominated end moraines occur widely around the margins of the Great Lake basins, where the markedly lobate margin of the retreating Laurentide Ice Sheet re‐advanced repeatedly into proglacial lakes and over fine‐grained sediment. This suggests the wider applicability of the till transport and incremental depositional model presented here.     

Ferromanganese oxide deposits from the Central Pacific Ocean,I. Encrustations from the Line Islands Archipelago

Chemical and mineralogical analyses of a well-controlled suite of ferromanganese encrustations from the Line Islands Archipelago (Central Pacific) suggest that they represent purely hydrogenous deposits—i.e. they have formed through the slow accumulation of trace metal-enriched oxides directly from the water-column. Mineralogically they consist predominantly of δMnO₂ and amorphous FeOOHxH₂O. Compositionally, they are similar to δMnO₂ nodules from adjoining basinal areas but are enriched in both Mn (mean = 20.4%, max = 29.3%) and Co (mean = 0.55%, max = 1.57%). δMnO₂ is the most important trace metal bearing phase; strong associations are noted between it and Co, Mo, Ni, Zn, and Cd, whilst only Be is associated specifically with FeOOH. V, Sr and Pb are partitioned between the authigenic oxide phases, whilst Ti most probably occurs as TiO₂xH₂O. Cu is contained in both aluminosilicate contaminant phases and Fe oxide phases. These relations are considered to reflect the differing scavenging behaviour of Mn and Fe oxides in the water column.Crusts from ~1–2 km are enriched in Mn and the Mn-related elements and exhibit higher $\text{Mn}\text{Fe}$ ratios than deeper crusts, which are compositionally constant. The higher $\text{Mn}\text{Fe}$ ratios may result from a supply of Mn from continental borderland sediments at these depths, which is transported horizontally by advective-diffusive processes. Since manganophile elements are enriched relative to Mn in the 1–2 km crusts, it is considered that the supply of Mn is scavenged by existing oxides, is oxidised and effectively occludes them. A higher proportion of oxide particles thus exhibit Mn oxide scavenging properties in the 1–2 km depth zone. The increased vertical flux of Mn resulting from the supply at ~1–2 km is not reflected by higher $\text{Mn}\text{Fe}$ ratios in deeper crusts, so that the vertical flux of oxides is not simply related to the standing crop. The $\text{Mn}\text{Fe}$ ratios of the crusts thus reflect the composition of suspended oxides at similar depths.     

Kiglapait geochemistry V: Strontium

The Kiglapait intrusion contains 330 ppm Sr and has $\text{Sr}\text{Ca}\text{= 5 × 10}{}^{\mathrm{?3}}$ and $\text{Rb}\text{Sr}\text{= 3 × 10}{}^{\mathrm{?3}}$, as determined by summation over the Layered Group of the intrusion. Wholerocks in the Lower Zone contain 403 F_L^0.141 ppm Sr, where F_L is the fraction of liquid remaining; Sr drops to 180 ppm at the peak of augite production (F_L = 0.11) and rises to a maximum of 430 ppm in the Upper Zone before decreasing to 172 ppm at the end of crystallization. Feldspars in the Lower Zone contain 532 F_L^0.090 ppm Sr, increasing to 680 ppm in the Upper Zone before decreasing to 310 ppm at the end. Clinopyroxenes contain 15 to 30 ppm Sr and have a mineral-melt distribution coefficient D = 0.06 except near the top of the intrusion where D = 0.10.The calculated feldspar-liquid distribution coefficient has an average value near 1.75 but shows four distinct trends when plotted against X_An of feldspar. The first two of these are strongly correlated with the modal augite content of the liquid, on average by the relation D = 1.4 + 0.02 Aug^L. The third (decreasing) trend is due to co-crystallization of apatite, and the fourth (increasing) trend can best be attributed to a triclinic-monoclinic symmetry change in the feldspar at An₂₆, 1030°C. The compound feldspar-liquid distribution coefficient K_D for $\text{Sr}\text{Ca}$ bears out these deductions in detail and yields ΔG_r for the Sr-Ca exchange ranging from nearly zero at the base of the Lower Zone to ?26 kJ/gramatom at the end of crystallization. The compound feldspar-liquid distribution coefficient K_D for $\text{Rb}\text{Sr}$ varies from 0.3 in the Lower Zone to 1.1 at the end of crystallization.The ratio $\text{Ca}{}^{\mathrm{F}}\text{Ca}{}^{\mathrm{L}}$ is about 1.45 for troctolitic liquids containing 5% augite, for which K_D (Sr-Ca) = 1.0 and D_Ca = D_Sr. For common basaltic liquids containing 20% augite, the Kiglapait data predict $\text{solSr}{}^{\mathrm{F}}\text{Sr}{}^{\mathrm{L}}\text{= 1.8}$, as commonly found elsewhere. The strong dependence of D_sr on augite content of the liquid illuminates the role of liquid composition and structure in determining the feldspar-liquid distribution coefficient. Conversely, a discontinuous change in the trend of D_Sr when apatite arrives shows that the effect is due to apatite crystallization itself, not to the continuous variation of the liquid as it becomes enriched in apatite component.     

The AD 1717 rock avalanche deposits in the upper Ferret Valley (Italy): a dating approach with cosmogenic 10Be

On 12 September AD 1717, a rock volume larger than 10 million m³ collapsed onto the Triolet Glacier, mobilized a mass composed of ice and sediment and travelled more than 7 km downvalley in the upper Ferret Valley, Mont Blanc Massif (Italy). This rock avalanche destroyed two small settlements, causing seven casualties and loss of livestock. No detailed maps were made at the time. Later investigators attributed accumulations of granitic boulders and irregular ridges on the upper valley floor to either glacial deposition, or the AD 1717 rock avalanche, or a complex mixture of glacial deposition, earlier rock avalanche and AD 1717 rock avalanche origin. In this study, we present cosmogenic ¹⁰Be exposure ages from nine boulders in the extensive chaotic boulder deposit with irregular ridges, two from Holocene glacier‐free areas, and one from a Little Ice Age moraine. Exposure ages between 330 ± 23 and 483 ± 123 a from eight of nine boulders from the chaotic deposit indicate that at least seven were deposited by the AD 1717 rock avalanche. The other three boulders yielded ¹⁰Be exposure ages of 10 900 ± 400, 9700 ± 400 and 244 ± 97 a, respectively. Our results are in good agreement with the existing chronology from dendrochronology and lichenometry, and radiocarbon analysis of wood samples, but not with older ¹⁴C ages from a peat bog in the upper part of the valley. Based on the new age control, the rock avalanche deposits cover the whole bottom of the upper Ferret valley. Copyright © 2012 John Wiley & Sons, Ltd.     

Context of relict Wisconsinan glacial ice at Angus Lake, SW Banks Island, western Canadian Arctic and stratigraphic implications

Massive ice incorporating till beds and scattered erratic clasts exposed in a bimodal thaw slide displayed an association with the surrounding fluvioglacial and flow-till facies indicative of a consanguineous relationship. The massive ice cores a low ridge interpreted as a recessional moraine feature paralleling the Last Glacial Maximum advance limit. Both ice and sand wedges penetrate the massive ice and their tops are truncated by a thaw unconformity overlain by a diamicton. In a permafrost environment, relict glacial ice can persist for many millennia, and the timing of the thaw is largely determined by the vagaries of erosional processes. The age of related kettle landforms can display considerable diachronism.     

Amino-acid racemizarion in Quaternary shell deposits at Willapa Bay,Washington

Extents of racemization ($\text{d}\text{l}$ ratios) of amino acids in fossil Saxidomus giganteus (Deshayes) and Ostrea lurida Carpenter were measured on shell deposits exposed at 21 sites on the east side of Willapa Bay, Washington. Amino acids from Saxidomus show less variability in $\text{d}\text{Spl}$ ratios and, therefore, are of greater use in correlation and age estimation than are amino acids from Ostrea. Shells of two different ages, about 120,000 ± 40,000 yr old and about 190,000 ± 40,000 yr old, are present. These ages correspond to Stages 5 and 7 of the marine isotope record defined by Shackleton and Opdyke in 1973 and hence the shell deposits likely formed during two different high stands of sea level. The stratigraphic record at Willapa Bay is consistent with this interpretation.     

Rare earth elements in Huronian (Lower Proterozoic) sedimentary rocks: Composition and evolution of the post-Kenoran upper crust

The Huronian sequence (Lower Proterozoicl. north of Lake Huron, contains tillites and abundant fine-grained sedimentary rocks. Analyses of rare earth elements (REE) in the matrix of tillite samples from the Gowganda Formation (~ 2.3 Gal is considered to be a reasonable estimate of upper crustal REE abundances for the region north of Lake Huron at the time of Gowganda deposition. The average is characterized by a moderately steep pattern (σLREEσHREE = 9.1) and a slight negative europium anomaly ($\text{Eu}\text{Eu1}\text{= 0.89}$). This pattern is similar to estimates of the composition of the surface of the Canadian Shield and is intermediate between estimates of typical Archean and post-Archean sedimentary rocks. REE patterns for framework granitoid clasts from the tillite suggest that K-rich granites, which were apparently unimportant in the formation of Archean sedimentary rocks, were abundant in the source regions of the Gowganda Formation. This may explain the intermediate nature of the Gowganda pattern.Comparison of the tillites and associated Gowganda mudstones suggests that previous estimates of upper crustal REE abundances, which were based on the analyses of fine-grained sedimentary rocks, may be systematically high. Relative distributions, however, are the same.Analyses of mudstones from the McKim. Pecors. Serpent Gowganda Lorrain and Gordon Lake Formations suggest rapid evolution in the composition of the exposed upper crust at the close of the Kenoran orogeny. REE patterns at the base of the Huronian are similar to typical Archean sedimentary rocks. REE characteristics change up section: patterns at the top resemble typical post-Archean sedimentary rocks.It is inferred that an essentially episodic change from an early exposed upper crust dominated by a tonalite-greenstone suite to one approximating granodioritic composition is recorded in Huronian sedimentary rocks. A deviation from the evolutionary trend of the Huronian, documented in the Gowganda Formation, may be related to the glacial origin of the Gowganda.     

Ocean chemistry during glacial time

Measurements of CO₂ to air ratios in the gas trapped in bubbles in ice of glacial age suggest that the CO₂ content of the atmosphere was considerably lower during peak glacial time than during Holocene time. The purpose of this paper is to show that such a change must in all likelihood be the result of alterations in the nutrient element chemistry of sea water. Two possible scenarios are presented. One involves alternate storage and erosion of phosphorous leaving residues from shelf sediments. The other involves changes in the $\text{C}\text{P}$ ratio in the organic debris falling to the deep sea. Means of verifying the nutrient cycle hypothesis are also given. It is shown that the ¹³C record as we know it in planktonic and benthic foraminifera, the oxygen record as inferred from benthic foraminifera species distributions, and the early post glacial CaCo₃ preservation event as recorded by aragonitic pteropods are consistent with both of the hypotheses presented. Only if an early post glacial spike in the ¹³C record for planktonic shells could be found would it be possible to eliminate one of these hypotheses (i.e., that involving shelf storage). The implications of these nutrient hypotheses to climate theory are as follows. If shelf storage is responsible for the glacial to interglacial CO₂ increase, then the CO₂ change must be considered an amplifier of some primary cause. The reason is that sea level changes are needed to drive deposition on (and erosion from) the shelves. On the other hand, if changes in the $\text{C}\text{P}$ ratio for falling debris are responsible, then the CO₂ change could either be an amplifier or a primary cause for the major glacial to interglacial climatic cycle. The latter is possible as self-sustained oscillations in ocean chemistry might be driven by interactions between ocean ecology and ocean nutrient chemistry.     

Chemistry of potassium feldspars from three zoned pegmatites,Black Hills,South Dakota: Implications concerning pegmatite evolution

An initial phase of an extensive geochemical study of pegmatites from the Black Hills, South Dakota, indicates potassium feldspar composition is useful in interpreting petrogenetic relationships among pegmatites and among pegmatite zones within a single pegmatite. The $\text{K}\text{Rb}$ and $\text{Rb}\text{Sr}$ ratios and Li and Cs contents of the feldspars within each zoned pegmatite, to a first approximation, are consistent with the simple fractional crystallization of the potassium feldspar from a silicate melt from the wall zone to the core of the pegmatites. Some trace element characteristics (i.e. Cs) have been modified by subsolidus reequilibration of the feldspars with late-stage residual fluid.$\text{K}\text{Rb}$ ratios of the potassium feldspar appear to be diagnostic of the pegmatite mineral assemblage. The relationship between $\text{K}\text{Rb}$ and mineralogy is as follows: Harney Peak Granite (barren pegmatites) > 180; Li-Fe-Mn phosphate-bearing pegmatites = 90?50; spodumene-bearing pegmatites = 60?40; pollucitebearing pegmatites < 30. Although the $\text{K}\text{Rb}$ ratios suggest that the pegmatites studied are genetically related by fractional crystallization to each other and the Harney Peak Granite, overlapping $\text{Rb}\text{Sr}$ ratios and the general increase in Sr and Ba with decreasing $\text{K}\text{Rb}$ indicate the genetic relationship is much more complex and may also be dependent upon slight variations in source (chemistry and mineralogy) material composition and degrees of partial melting.     

Kiglapait Geochemistry III: Potassium and Rubidium

The regular geometry and completeness of the Kiglapait intrusion permit its bulk composition to be obtained by summation, and the composition of successive liquids to be obtained by subtraction. The summations for K and Rb give 1806 and 1.08 ppm, yielding Rfrsol|K/Rb= 1670 for the intrusion, taken as equal to the parent magma. R increases slightly from this initial value to 2000 at the end of crystallization where MgO approaches zero in the rocks. K and Rb are therefore closely coherent and their distribution coefficients can differ only by a small amount in the Kiglapait system.Apparent feldspar/liquid distribution coefficients (D^F/L) can be estimated from detailed plots of feldspar and liquid compositions against F_L. The Kiglapait data imply that these coefficients are linear 1:1 functions of plagioclase composition within experimental error, having values given by D_KF/L = 1.42? X_AnD_RbF/L = 1.13? X_An with minimum values of 0.75 and 0.49, respectively. The ratio $\text{R}{}^{\mathrm{F}}\text{R}{}^{\mathrm{L}}$ lies in the range of 1.53± 0.03 for the plagioclase composition range X_An= 0.34 to 0.67 showing that high-R rocks such as anorthosite crystallized from high-R liquids.The apparent feldspar distribution coefficients are much closer to 1.0 than common literature values. They can be reduced by assuming that the cumulate pile was continuously recharged by the circulating magma until an advanced stage of differentiation was reached, and assuming that alkalies were exchanged to the feldspars from the magma. When such an ‘aquifer recharge’ model is calibrated using olivine-liquid equilibria as a time marker for the liquid, the inferred minimum equilibrium values of the distribution coefficients are $\text{D}{}_{\mathrm{K}}\text{F}\text{L}$= 0.42, $\text{D}{}_{\mathrm{Rb}}\text{F}\text{L}$ = 0.25 at the base of the intrusion. Their variation is given by $\text{D}{}_{\mathrm{K}}\text{F}\text{L}\text{= 1.66?1.88X}{}_{\mathrm{An}}$, $\text{D}{}_{\mathrm{Rb}}\text{F}\text{L}\text{= 1.17?1.41X}{}_{\mathrm{An}}$, The equilibrium values are considered to be appropriate for deducing liquid compositions in plutonic bodies where alkali exchange can be shown or inferred to have been inhibited, such as in small intrusions. The apparent values are considered to be appropriate, even though they may be artificial, for large intrusions similar to the Kiglapait.The bulk K and Rb concentrations in the Kiglapait intrusion are consistent with a plagioclase-rich abyssal tholeiite magma. Clinopyroxene and olivine fractionation in the mantle may contribute to the production of such high-Rmagmas.     

Late Quaternary glaciation and equilibrium line altitude variations of the McKinley River region,central Alaska Range

Dortch, J. M., Owen, L. A., Caffee, M. W. & Brease, P. 2009: Late Quaternary glaciation and equilibrium line altitude variations of the McKinley River region, central Alaska Range. Boreas, 10.1111/j.1502‐3885.2009.00121.x. ISSN 0300‐9483 Glacial deposits and landforms produced by the Muldrow and Peters glaciers in the McKinley River region of Alaska were examined using geomorphic and ¹⁰Be terrestrial cosmogenic nuclide (TCN) surface exposure dating (SED) methods to assess the timing and nature of late Quaternary glaciation and moraine stabilization. In addition to the oldest glacial deposits (McLeod Creek Drift), a group of four late Pleistocene moraines (MP‐I, II, III and IV) and three late Holocene till deposits (‘X’, ‘Y’ and ‘Z’ drifts) are present in the region, representing at least eight glacial advances. The ¹⁰Be TCN ages for the MP‐I moraine ranged from 2.5 kyr to 146 kyr, which highlights the problems of defining the ages of late Quaternary moraines using SED methods in central Alaska. The Muldrow ‘X’ drift has a ¹⁰Be TCN age of ～0.54 kyr, which is ～1.3 kyr younger than the independent minimum lichen age of ～1.8 kyr. This age difference probably represents the minimum time between formation and early stabilization of the moraine. Contemporary and former equilibrium line altitudes (ELAs) were determined. The ELA depressions for the Muldrow glacial system were 560, 400, 350 and 190 m and for the Peters glacial system 560, 360, 150 and 10 m, based on MP‐I through MP‐IV moraines, respectively. The difference between ELA depressions for the Muldrow and Peters glaciers likely reflects differences in supraglacial debris‐cover, glacier hypsometry and topographic controls on glacier mass balance.