Introduction

Data on world-wide glacier fluctuations are being compared and interpreted in an increasing number of publications (Reference PatzeltPatzelt, 1985; Reference Kislov and KoryakinKislov and Koryakin, 1986; Reference Makarevich and RototayevaMakarevich and Rototayeya. 1986; Reference Vallon, Reynaud and LetréguillyVallon and others. 1986; Reference Haeberli, Müller, Alean, Bösch and OerlemansHaeberli and others. 1989a; Reference Kich and OerlemansKick, 1989; Reference OerlemansOerlemans and others. 1993; Reference Williams and FerrignoWilliams and Ferrigno, 1993; Reference OerlemansOerlemans, 1994). The surface areas of glaciers in China account for about 10% of the total surface area covered by ice caps and mountain glaciers on Earth, existing outside the large polar ice sheets (Reference Haeberli, Bösch, Scherler, Østrem and WallénHaeberli and others, 1989b. With the exception of a few individual glaciers, however, most Chinese glaciers have been rarely monitored and measured. Consequently, data on glacier fluctuations in China are limited. The Chinese glacier monitored and measured in most detail is Ürümqi Glacier No. 1 within the Ürümqi River source region in the central Tien Shan. Since 1980. variations in the positions of glacier fronts (length change) during various time periods have been reported for anout 200 glaciers (Reference Xiang-SongZhang, 1980a. Reference Xiang-Songb: Karakoram; Reference ZhenshuanZhang and Mi. 1981: Qilianshan, Kunlunshan and Tien Shan; Reference Guanghe, Shunying and ZhongxiangWu and Others, 1983: Tien Shan; Wu and others, 1985: Qilianshan, Kunlunshan and Tien Shan; Reference Guanghe, Shunying and ZhongxiangWu and others, 1983: Tien Shan; Reference Xie Zichu and WangXie and Others, 1985: Qilianshan;; Ren, 1987: Kunlunshan; cf. also Reference Yafeng, Maohuan, Binghui, Shi, Huang and RenShi and others, 1988. cf. Fig. 1). Glaciers in the Swiss Alps were chosen for a main comparison, because they have been monitored in considerable detail during the past century. In Switzerland, mass-balance information exists for five glaciers and extensive records on length change are available for more than 100 glaciers.

Fig. 1 Location of areas studied (inset) and of mountain ranges in China. The number of glaciers studied is: 55 in Tien Shan. 42 in Qilianshan, 40 in Kunlunshan, nine in the Himalaya, seven in Hengduanshan, six in the Karakoram. three, in Tanggulashan and Iwu in lite Altai.

Glacier Mass Balances

Although Ürümqi Glacier No. 1 has been measured and studied in detail, mass-balance observations were interrupted between 1967 and 1979. The mass-balance data used here for the time interval 1967–79 were reconstructed by J. H. Reference Xie Zichu and WangZhang (1981), Reference Jinhua, Xiaojun and JijunZhang and others (1984) and Reference Yafeng, Maohuan, Binghui, Shi, Huang and RenShi and others (1988) on the hasis of precipitation and air temperature recorded by a meteorological station situated 2.5km from the glacier and at an altitude of 3589ma.s.l. The mass-balance record of another glacier. Qiyi Glacier in the central Qilianshan, was reconstructed (Reference Chaohai and ZichuLiu and Xie, 1987) after 5 years of field measurements using a relationship between accumulation, ablation and meteorological parameters (altitude of the 0°C isotherm, summer air temperature, etc.). Both glaciers are in areas of a continental climate. In Switzerland, the mass balance of Aletschgletscher has been estimated from a hydro-logical balance model calibrated by precision mapping in 1927 and 1957. The model uses predpitadon and run-off measurements and assumes that annual evaporation remains constant.

Figure 2 Compares annual mass balances since 1959 for glaciers in China and Switzerland. Annual variations in mass balance are quite synchronous for the Swiss glaciers, which are located in the same climatic region and are close together. On the other hand, short-term variations are not synchronous in Switzerland and China. Reference Letréguilly and TeynaudLetrèguilly and Reyuaud (1990) discussed similar spatio-temporal patterns of mass-balance variability and pointed out that the identity of annual mass-balance variations cannot be recognized beyond individual mountain ranges. They further concluded that, beyond a distance of 500 m, synchroncity cannot be found at decadal time-scales but dial the main secular trends seem to be common in Europe and High Asia. Close inspection of the records indicates that common negative balances during 1962–64 in Switzerland, for instance, correspond to highly positive balances of Ürümqi Glacier No. 1. An abrupt change to positive balances on Swiss glaciers since 1965. and continuing until about 1970. is accompanied by strong mass losses of Ürümqi Glacier No. 1.

Fig. 2 Specific net mass balance (mm w.e.) for glaciers in Switzerland and China. (a) Swiss glaciers: Silveretta-gletscher (solid line), Limmerngletscher (dotted line), Griesgletscher (dashed line) and Aletschgletscher (dashed) dotted line); (b) Ürümqi Glacier No. 1 in China.

Figure 3 shows time series of cumulative-balance variations. There is considerable variation among individual glaciers with respect to overall gain or loss in mass. Even though the reconstructed mass balance for Qiyi Glacier is very approximate, evidence of overall mass gain can be found. Four of the five annual mass balances determined in the field were positive and the average equilibrium-line altitude on the glacier decreased by about 100m between the decade 1957–66 to the decade 1967–77 (Reference Liangfu and XianchengDing and Kang, 1985). The tendency towards a positive cumulalive balance for Qiyi Glacier could therefore be real. In Switzerland, cumulative balances since 1960 were slightly positive on Silvrettagletscher and markedly negative on Griesgletscher. The phenomenon of highly variable cumulative mass-balance series and significantly different conditions of health appears to be common for glaciers in the Nothern Hemisphere (Reference Letréguilly and TeynaudLetrèguilly and Reynaud, 1990). Amongst the reasons for this are the orientation of the glaciers and the distribution of glacier area with altitude (hypsometry). In Switzerland, mass balances a ppear to be more negative on glaciers that are exposed to the northeast (there appears to be a similar phenomenon for most glaciers in France and Austria; cf, the data given in Reference HaeberilHaeberli. (1985), Reference Haeberli and MüllerHaeberli and Müller (1988); Reference Haeberli and BenistonHaeberli and others (1994); Reference Haeberli and HoelzleHaeberli and Hoelzle (1995)). Unfavourable orientations of glaciers in the Tien Shan with respect to the predominant humidity source are the northeastern and southeastern quadrants (Table 1), Qjyi Glacier, however, has an unfavourable exposure but a positive mass balance. Glacier orientation, therefore, cannot be the only or even lhe most important factor influencing the variability in cumulative mass balances. The highly positive balance of Qiyi Glacier could result from its hypsometry (Fig. 4); the wide firn basin is favourable for the accumulation of snow. In a comparable way, a relatively large firm basin seems to remedy the unfavourable northern exposure of Silvrettagletscher. Plallalva-gletscher and Limmerngletsehcr are probably typical examples of the combined elfecls of orientation and hypsometry. These two small glaciers are immediately adjacent to each other but their mass balances are different. Their combined exposure and hypsometry may explain the different balances. A similar phenomenon can also be observed with Hinlereis- and Kesselwandferner in Austria (Reference Kuhn, Markl, Kaser, Nickus, Obleitner and SchneiderKuhn and others, 1985; Reference GreuellGreuell, 1992).

Fig. 3 Cumulative mass balance (CMB) vs time for glaciers in (a) Switzerland and (b) China.

Table. 1. Relationship between orientation of accumulation area and cumulative balance

Fig. 4 Hypsometry of glaciers with negative (left) and positive (right) cumulative mass balance. Ordinate gives altitude interval in m a.s.l.

In strong contrast to the remarkable scatter in annual and cumulative mass balance, changes in cumulative mass balance are similar over the time period considered (1959–90): during the lalesi decade (1980–90), an accelerating trend towards more negative mass balances appears in both regions (Fig. 3). Short-term local to regional variability is nevertheless superimposed on this general trend. A change towards markedly negative balance for Ürümqi Glacier No. 1 occurred in 1978 with the strongest loss in 1981 (Fig. 2), while Swiss glaciers at the same time maintained positive or zero balances. Rates of mass loss Ibr the Swiss glaciers, on the other hand, have accelerated strongly since 1981. The trend towards negative balances then remains obvious for the entire decade of the 1980s in both regions. This observation may indicate that accelerated warming of the 1980s, as reflected by glacier and permafrost changes in the Alps (Reference Haeberli and BenistonHaeberli, 1994), not only appears in mountain ranges with transitional to maritime climatic conditions but also affects areas of strong continentality.

Variations in the Positions of Glacier Fronts (Glacier-Length Changes)

There are no continuous data on annual variations in the positions ol glacier fronts in China. Information on glacier-length changes over various longer lime intervals, however, is available in a number of publications (Reference Xiang-SongZhang, 1980a, Reference Xiang-Songb; Reference ZhenshuanZhang and others. 1981; Reference Guanghe, Shunying and ZhongxiangWu and others, 1983; Reference HaeberilHaeberli, 1985; Reference Xie Zichu and WangXie and others, 1985; Ren, 1987; Reference Haeberli and MüllerHaeberli. 1988; Reference Yafeng, Maohuan, Binghui, Shi, Huang and RenShi and others. 1988; Reference Haeberli and HoelzleHaeberli and Hoelzle. 1993). With the exception of a few direct measurements, most of those data were obtained from topographic maps at scales of 1 :50000 and 1 : 100000, from aerial photographs and from satellite imagery. Data on glacier-length changes in Switzerland were mainly extracted from the annual reports prepared by VAW/ETHZ for the Swiss Glacier Commission (Reference Kasser, Aellen and SiegenthalerKasser and others, 1986; Reference AellenAellen, 1988; Reference Aellen and HerrenAellen and Herren, 1991, Reference Aellen and Herren1992a, Reference Aellen and Herrenb, Reference Aellen and Herren1993). Most of the time intervals considered are shorter than die dynamic response time of the glaciers involved. It is, therefore, necessary to analyse cumulative length changes of glaciers with comparable geometry especially total length) in order to avoid comparing glaciers with highly different response characteristics (cf. Reference KuhnKuhn, 1978; Reference Haeberli, Müller, Alean, Bösch and OerlemansHaeberli and others, 1989a; Reference Haeberli and HoelzleHaeberli, 1995).

Figure 5 compares cumulative length changes of three Chinese glaciers with average cumulative length changes determined for Swiss glaciers of more or less equal length. Tuergangou Glacier (Tien Shan) and Qiyi Glacier (Qilianshan) are continental glaciers but Hailuogou Glacier (Mount Gonga in the Hengduanshan) is some-whal maritime. These three glaciers are among the few in China which have been documented by repeated surveys over variable time intervals and hence may be compared with detailed Swiss records. A general trend towards retreat has obviously predominated during the past 30 years for the investigated glacier sizes in China as well as in Switzerland. It is also noteworthy that the continental Qiyi Glacier reacts less and the maritime Hailuogou Glacier more sensitively than Alpine glaciers with their transitional climate.

Fig. 5 Comparison between average change in length of Swiss glaciers and changes in length of individual Chinese glaciers.

The similarity of long-term glacier-length changes in China and Switzerland also appears in the statistics for different classes of glacier length (Table 2). Increases and decreases in the percentage of retreating glacier snouts during different time intervals roughly follow the same pattern in both countries, especially with respect to glaciers shorter than 10km. Glacier retreat clearly predominates, with the exception of the intermittent advance of long Chinese glaciers after the middle of the present century and during the period around 1980 when 2–5 km long glaciers advanced in both countries. In general, percentages and rates of retreat for glaciers shorter than 10km were higher during the 1950s but decelerated since the 1960s, leading to a tendency towards İntermittent advance in the 1970s with a peak from the middle of the 1970s to the beginning of the 1980s.

Table. 2. Comparism of length changes for comparable length classes (km) of glaciers in China and Switzerland (% is percentage of retreating glaciers with number of glaciers in brackets; rate is average rate of length change in ma ⁻¹ for the entire sample of the size category and the considered time interval). The locations of the Chinese glaciers for the individual time intervals are as follows: ⁽¹⁰ = Qilianshan (eight glaciers); ⁽²⁾ = Tien Shan (two glaciers) and Qilianshan (one glacier; ⁽³⁾ Qilianshan (nine glaciers); ⁽⁴⁾ Qilianshan (four glaciers) and Tien Shan (three glaciers); ⁽⁵⁾Qilianshan (six glaciers), Tien Shan (one glacier) and Kunlunshan (seven glaciers); ⁽⁶⁾Qilianshan (two glaciers) and Nianqingtanggulashan (one glacier); ⁽⁷⁾ Qilianshan (four glaciers) and Altaishan (one glacier); ⁽⁸⁾ Qilianshan (11 glaciers), Tien Shan (11 glaciers) andKunlunshan (six glaciers); ⁽⁹⁾ Tien Shan (one glacier), Pamir (one glacier) and Kuntunshan (four glaciers); ⁽¹⁰⁾ Tien Shan (two glaciers) and Hengduanshan (one glacier); ⁽¹¹⁾ Qilianshan (one glacier). Karakoram (four glaciers), Himalaya (one glacier) and Henduanshan (two glaciers); ⁽¹²⁾ Tien Shan (11 glaciers); ⁽¹³⁾ Qilianshan (one glacier), Tien Shan (11 glaciers), Kunlunshan (17 glaciers), Karakuram (two glaciers) and Himalaya (one glacier); ⁽¹⁴⁾ Qilianshan (one glacier), Tanggulashan (one glacier). Kunlunshan (two glaciers) and Hengduanshan (four glaciers); ⁽¹⁵⁾ Qilianshan (one glacier), Karakoram (one glacier). Hengduanshan (two glaciers); ⁽¹⁶⁾ Tien Shan (six glaciers); ⁽¹⁷⁾ Qilianshan (one glacier). Tien Shan (two glaciers). Kunlunshan (ten glaciers), Himalaya (one glacier) and Karakoram (one glacier); ⁽¹⁸⁾ Qilianshan (one glacier). Tanggulashan (one glacier), Hengduanshan (three glaciers) and Kunlunshan (one glacier)

The retreat of the Chinese glaciers longer than 10 km during various time periods appears to be less steady and less homogenous than in Switzerland. In fact, there seems to be a distinct difference between the two countries with respect to the average rates of glacier-length changes and to percentages of advance/retreat for large glaciers. One reason may he that 51 Chinese glaciers with lengths exceeding 10 km. including 18 glaciers longer than 20 km, are used for the analysis, whereas there are only five Swiss glaciers longer than 10km and only one (Aletsch-gletscher) longer than 20 km. Another reason may be the larger errors for glaciers in China, because data on length changes of large glaciers in China are mostly estimated from satellite imagery with resolutions and accuracies of about 100 m. Comparing percentages of retreating glaciers therefore may be more representative in this special case of large glaciers and indeed gives somewhat similar results for both regions.

Figure 6 summarizes percentages of advance/retreat for all glaciers monitored in China and Switzerland over decadal time intervals. Historically, this approach has been popular. It is, however, highly problematic, because of the different response types involved and can, at best, give only a very general outline. Because of the somewhat steady retreat for large glaciers. Figure 6 shows principally decadal reactions of smaller glaciers and, as such, tends to confirm the similarity of glacier changes in both countries beyond the time-scale of one or a few years.

Fig. 6 Variation in the position of fronts for about 200 glaciers in China and 160 glaciers in Switzerland.

Fig. 7 Relationship between cumulative mass balance (CMB) and cumulative length change (CLC) for glaciers of different lengths.

Relation between Mass Balance and Glacier-Length Changes

The reactions of glaciers to climatic change involve a complex chain of processes. Mass balance is the direct, undelayed consequence of climatic forcing, while the length change is an indirect, delayed reaction. The fluctuatins described above for smaller glaciers in both countries demonstrate their sensitive reactions to climatic change. Actually, length reactions to mass-balance forcing for small glaciers are very quick (Fig. 7). Length changes for the small Plattalvagletscher, for instance, are almost synchronous with balance variations and the curve of (cumulative) glacier-length changes is not much smoother than the one of cumulative mass balance. Almost synchronous changes in length and mass balance can also be found over various time intervals at Ürümqi Glacier No. 1, though its length changes have not been continuously measured. With increasing glacier length, tongue reactions seem to be slower and the smoothing of curves from cumulative length change with respeel to mass-balance records are more pronounced. For Silvretta-gletseher, synchronous variations still exist but are less obvious. No short-term (yearly to multi-annual) relation between length change and mass balance exists for Aletsch-glctscher, a glacier more than 20km long. Its reaction time, i.e. the time lag between a marked change in cumulative mass balance and the onset of lhe corresponding advance/retreat of the glacier snout, may be 30 years or more; time for full dynamic response and adjustment to a new equilibrium length is estimated at about 70–80 years (Reference Haeberli and BenistonHaeberli, 1994; Reference HaeberliHaeberli and Hoelzlc, 1995). Such long delays will essentially smooth out shorl-lerm effects of climate and mass-balance forcing on glacier-length changes, so that yearly to decadal glacier-terminus reactions to climatic change will be unclear.

Figure 8 summarizes the relation between mean mass balance and mean annual length change four length classes of Swiss glaciers. The length changes for the shortest glaciers (≤2km) are indeed almost perfectly synchronous with variations in mass balance. Length changes for larger glaciers (total length = 2–5 km and 5–10km) appear to follow after a time lag of a few years. The length changes for 5–10 km long glaciers have a slightly longer time lag combined with more pronounced smoothing but the differences with respect to the 2–5 km long glaciers are too small to make sharp distinctions. Interpretation of the length-change signal of glaciers longer than 10km and for the time interval covered by direct mass-balance measurements is more difficult. Reaction times are most likely to be in the decade range. Most remarkably, however, and in contrast to the shorter glaciers, average rates of change remained negative (retreat) throughout the observation period. The longest glaciers were not (yet?) able to re-advance as a reaction to the positive mass balances in the late 1960s and late 1970s.

Fig. 8 Comparison between annual mass balance and annual length change for various sizes of Swiss glaciers; (a) 5 year centred moving average of mass balance; (b) 5 year centred moving average for length changes in different length (L) classes.

The time period considered in Figure 8 is about 30 years and, hence, shorter than the characteristic dynamic response time of most mountain glaciers. By looking at lime intervals, which correspond to the dynamic response time (t_a) of individual glaciers, the long-term mean mass balance (〈b〉) can be inferred from cumulative length change caused by a step change in mass balance (δb) on the basis of assumed steady-state conditions before and after response, linear adjustment of mass balance to new equilibrium conditions during response and continuity (cf. Reference Jóhannesson, Raymond and WaddingtonJohannesson and others, 1989; Reference HaeberliHaeberli, 1990, Reference Haeberli and Beniston1994; Reference HaeberliHaeberli and Hoelzle, 1995, for background and calibration):

(1)

(2)

(3)

where b_t , is the (annual) ablation at the glacier terminus, δb is the assumed Step change in mass balance leading to the length change δL over the time period t_a starting from the original glacier length L₀ and h_max is maximum glacier thickness (all values in water equivalent).

Table 3 comparts average mass balances calculated in this way for the Alpine Rhonegletscher and the Chinese Ürü mqi Glacier No. 1. The agreement between the directly measured mass balance and the mass balance inferred from cumulative length change is striking for Rhonegletscher (measured: −0.25/inferred: −0.28 m w.e. year⁻¹ as well as for Ürümqi Glacier No. 1 (measured: −0.14/inferred: −0.11 m w.e. year⁻¹). The dynamic response time of Rhonegletscher is about twice as long as that of Ürümqi Glacier No. 1. For direct comparison of the two glaciers and their secular mass changes inferred from cumulative lengih change, equally Umg time intervals should be considered. Ürümqi Glacier No. 1 began to retreat from its Little Ice Age maximum in 1876; by 1990, it had lost 0.5 km in length. With this historical retreat, the secular balance change (δb) and average secular balance (〈b〉) calculated from Equation (2) become −0.8 and −0.4 m w.e. year⁻¹, respectively. However, the glacier was probably able to adjust fully twice or even three times as a reaction to (assumed step-type) mass-balance changes since the end of the Iasi century. The calculated secular mass-balance changes and average secular mass-balance values must, therefore, be reduced correspondingly. The resulting average secular mass loss of Ürümqi Glacier No. 1 is 0.1–0.2 m w.e. year⁻¹. Such a value is roughly half the characteristic values obtained for Alpine glaciers (Rhonegletscher in Table 3; cf. also the data given by Reference Haeberli and BenistonHaeberli (1994) and Reference Haeberli and HoelzleHaeberli and Hoelzle (1995) — a fact which may be explained by the lower climatic sensitivity of the continental-type Ürümqi Glacier No. 1. It is especially important to note that the cold firn area of Ürümqi Cilacier No. 1 (Reference Haeberli and BenistonHaeberli and others, 1994) probably reacted to 20th-century atmospheric warming by increased mellwater relreezing and firn warming; mass loss was therefore restricted to the ablation area or about half the glacier area. Characteristic rates of secular glacier mass loss are in any case in the range of dm year⁻¹ and, hence, closely comparable in both regions.

Table.3. Comparison of measured and estimated glacier mass changes. Sources: Reference Aellen, Kasser and HaeberliAellen (1981), Reference JingtaiWang (1981), Reference ZhenshuanZhang. Z.S. (1981), Reference FunkFunk (1985), Reference GenxiangYou (1988), Reference Chen and FunkChen and Funk (1990), Reference Haeberli and HoelzleHaeberli and Hoeizle (1995)

Conclusions

Effects of climatic forcing on glaciers are different during various time periods and strongly depend on topographic and glaciological factors. Variations in the mass balance of individual glaciers at present depend mainly on humidity conditions and topographic factors (hypsometry and orientation with respect to the main humidity source). Comparisons between glacier evolution in China and the Swiss Alps nevertheless demonstrate that some characteristics of glacier fluctuations are similar in both regions during past decades. The overall trend is one of mass loss and glacier retreat with a modest intermittent growth and re-advance between 1970 and 1980. The similarity of variations in the positions of glacier fronts between the two countries points to comparable climatic forcing over decadal time intervals on the Eurasian continent. Inter-annual variations in mass balances, however, are different and remain partly unexplained. Further investigation of such differences is necessary and intercomparison of glacier fluctuations should be expanded to other regions of the world.