Introduction
In recent years, satellite remote sensing has increasingly been applied to various glaciological studies, using a variety of satellite platforms and sensors, including those with improved spatial, spectral, radiometric and temporal resolutions. The advantages of satellite remote sensing include periodic and large area coverage under the uniform condition. Because the swath widths are at least 60 km, satellite data are best suited for studying large to mediumsized glaciers with areas exceeding a few hundred km2. For small glaciers, conventional aerial photographs are still an important source of remotely sensed data, the coverage of which can be easily varied by changing the flying height and/or focal length of a camera lens.
In remote areas, a metric camera and a light aircraft able to accommodate it are often not available. Hielo Patagónico Norte (the northern Patagonia icefield) is such an area. Aerial photographs taken by the U. S., Chilean and Argentine governments between the 1960s and 1980s, with scales of about 1:70 000 to 1: 80 000, cover the area. These photographs are generally still of a scale too small for the examination of glaciers 550 km2. For such small glaciers, near-vertical aerial photographs taken with a 666 cm format camera are easy to acquire, because a hand-held camera is simple to operate and economical to use (Reference Aniya and NakajimaAniya and Naruse, 1985, Reference Aniya and Naruse1986, Reference Aniya and Naruse1987). Also an opening in the aircraft fuselage can be made inexpensively, without the need for substantial structural modification.
The purpose of this study is to analyze morphological changes on the surface of Glaciar Soler, Chile, based on the morphological maps produced from mosaics of 666 cm aerial photographs taken in 1984,1986 and 1999.
Study Area
The northern and southern Patagonia icefields are located in the southern part of South America, between 46°30’ and 51°30’ S along longitude 73°30’W, extending for 550 km along the axis of the Andes, with widths of about 8–45km. The two icefields, Hielo Patagónico Norte (HPN; northern Patagonia icefield; 4200km2) and Hielo Patagónico Sur (HPS; southern Patagonia icefield; 13 000 km2) are the largest temperate glaciers in the Southern Hemisphere, with a total combined area of 17200km2. HPN has 28 outlet glaciers (Reference AniyaAniya, 1988, fig. 1) flowing out from the margin of the icefield. In the past we have carried out extensive fieldwork in HPN, in particular at Glaciares Soler and San Rafael in 1984–86 (Reference NakajimaNakajima, 1985, Reference Nakajima1987) and in 1998 (Reference AniyaAniya and Naruse, 2001). As part of these projects, Aniya acquired vertical aerial photographs over Glaciar Soler in 1984,1986 and 1999 (Reference Aniya, Naruse and NakajimaAniya, 1985, Reference Aniya1987, Reference Aniya and Aniya2000).
The study area: Glaciar Soler and HPN.
Glaciar Soler is located at 46°54’ S, 73°11’W, on the east side of HPN (Fig. 1). It is a small glacier with a total area of 50.9 km2 and a length of 16.6 km. The elevation ranges from 3078m (top of Cerro Hyades) to about 350 m, but much of the accumulation area in the icefield lies at 2000–2200m. The accumulation–area ratio (AAR) is 0.73, with the equilibrium-line altitude (ELA) located around 1350m (Reference Aniya and NaruseAniya and Naruse, 1987). From estimation of the ELA in the field in 1985 (Cassasa, 1987), and interpretation of vertical aerial photographs taken in 1975, the AAR was estimated (Reference AniyaAniya and Naruse, 1987). Because of subsequent rapid recession, the ELA may have lowered and the AAR may have become smaller, but we do not have the new estimates. The ablation area lies mostly below icefalls, from 850m elevation down, with an area of 14.5 km2, a length of about 7 km and a width of about 2 km. The ablation area consists of two distinct ice bodies, debris-free and debris-covered. The debris-free part, which is fed with ice from the icefield, is composed of two different sections, while the debris-covered part, which is fed by ice avalanches from the flank of Cerro Hyades, contains five different sections (Reference AniyaAniya and others, 1988; Fig. 2 inset).
Surface morphology maps: A, 1984 (after Reference Aniya, Naruse and NakajimaAniya and Naruse, 1985); B, 1986 (after Reference AniyaAniya and others, 1988); and C, 1999 (extent of Fig. 3 indicated). Inset map is based on the 1986 map, showing the debris-free and debris-covered bodies and glacier sections.
The glacier has been retreating, and between 1986 and 1999 part of the snout retreated about 200–500m (Reference Aniya and NaruseAniya, 2001). The retreat has been accompanied by considerable thinning, averaging 5.2 ma–1 for 1983–85 and 3.2 ma–1 for 1985–98 at 0.5–2.5 km from the snout (Reference Naruse and SkvarcaNaruse and others, 2000). Consequently, two proglacial lakes have formed, though calving is not yet active.
Aerial-Photographic Survey Operations
In order to study the morphology of Glaciar Soler, we flew over the ablation area of the glacier in January 1984, November 1986 and November 1999. The camera used was a Zenza Bronica SQ-Am with a format of 666 cm, equipped with an 80,100 or 110 mm lens. The film used was Kodak Ektachrome Professional EPD 220 Reversal (ASA 200 for 1984, and 64 for 1986) and ENP 220 Reversal (ASA 100 for 1999). The aircraft employed were a Cessna 182 in 1984 and 1986, and a Piper Seneca in 1999. A small hole of about 15 cm diameter was cut through the fuselage under the co-pilot’s seat. The camera was placed in the opening and held by hand. Using a gridded focusing screen, photographs were taken manually so that forward overlap was >50% to achieve complete stereoscopic coverage. The flying height was about 3200ma.s.l. in 1984 and 3000 ma.s.l. in 1986 in order to avoid Cerro Hyades (3078m), which looms above the head of the ablation area. In 1999, however, it was about 2500ma.s.l. due to low cloud ceiling, which forced us to miss the icefall area of the mountainside.
Surface Morphology
The uncontrolled aerial-photographic mosaic of Glaciar Soler was assembled from 40–60 aerial photographs acquired each year for mapping morphology (Fig. 2). The approximate scale was determined using a 1:50 000 scale topographic map published by Instituto Geográfico Militar of Chile. Because of the uncontrolled nature of the mosaic, all distance measurements are approximate, with an estimated error of up to 10%.
The 666 cm quasi-vertical aerial photographs were enlarged to 24624 cm (scales of roughly 1:4500 for 1999 and 1: 8000 for 1984) for detailed stereoscopic interpretation. The mapped features include debris-covered and debris-free glacier surfaces, medial moraines, band ogives, wave ogives, major individual crevasses, general crevasse patterns, supraglacial streams and ponds, and ice-cored moraines. Special features are depressions, grooves, troughs, moulins and debris mounds for 1984, 1986 and/or 1999 maps. Grooves are wide U-shaped troughs which have developed nearly perpendicular to the crevasses on the right margin of the middle reach of the glacier.
Morphological changes
Between 1984 and 1986, the most notable change in the surface morphology was the formation of thrust-moraine ridges located across the glacier about 1km from the snout (Fig. 2b). They were covered by rounded sand. Since then, this part of the glacier has become a proglacial lake, with residuals of the thrust moraines bordering the lower end of the lake. The formation of a proglacial lake in front of the debris-covered glacier ice (north or lefthand side) is a notable change between 1986 and 1999. From the aerial-photographic survey itwas inferred that the lake was formed around 1991–92 (Reference AniyaAniya, 1992; Reference Wada and AniyaWada and Aniya, 1995). The proglacial lake on the south (righthand) side of the snout has grown considerably since 1986.
On the 1999 map (Fig. 2c), there is a semicircular debris-covered ice feature, part of which stands out slightly from the surrounding ice, on the righthand side of the snout (Fig. 3). When Aniya visited here in November 1995, it was a depression a few hundred meters in diameter, with 10–15m deep cliffs. There was a channel (or narrow proglacial lake) between the depression and the valley wall. The edge of the depression on the valley-wall side was then covered by a large volume of rock-avalanche debris on which some vegetation survived. On the valley wall there was a fresh scar of a landslide (or rock avalanche). According to a local guide, a flood had occurred in May 1994 in Río Cacho. Indeed, the area immediately below the snout had been scoured up to several meters above the normal river level.
Semicircular debris-covered surface and circular crevasse pattern on the southern (righthand) margin of the glacier, about 1.4 km from the snout (photo taken by Aniya on 29 November 1999).
From the above information, the event that occurred is interpreted as follows. It is assumed that there was a sub-glacial cave below the depression since on the 1993 oblique aerial photographs taken by Aniya there was a trace of the circular depression. The cave may have been formed on the lee side of a large protruding bedrock hump. Probably it was in the process of subsidence. Then a rock avalanche slid into the proglacial lake and onto the glacier surface; a large volume of water was displaced, removing the pressure that supported the ice above, and the surface collapsed, in turn displacing another large volume of water.
Ogives
Of the seven surface ice-flow features of the ablation area, all of which are fed through an icefall or by avalanches, wave ogives develop only on the B2 feature, while band ogives develop on the B1, B4, B5 and A1 features (Fig. 2 inset; Reference Aniya, Casassa and NaruseAniya and others, 1988). Among these band ogives, those that developed on the B5 feature are most distinctive, followed by those developed on the A1 feature. Band ogives developed on the B5 and A1 features were delineated and marked on the mosaic, and the width of a pair of dark and light bands was measured at the contact of the B5 and A1 features. Between 1984 and 1999, the retreat of the snout is substantial, so the distance was measured from the base of the large icefall (A1 feature) for comparison, rather than from the snout.
In 1986, development of aerial photographs that covered the middle part of the A1 and A2 features revealed that neither the light ogive bands nor the individual crevasses could be identified. However, light ogive bands were observed on the B5 feature, and the measurements were performed there.
On the 1999 mosaic, light ogive bands have become obscure on all the ice-flow features. Wave ogives on the B2 feature have also become less distinctive. The distinction between the B4 and B5 features has become less clear because of poor development of light ogive bands on the B4 feature. Although debris-covered, the surface of the B4 and B5 features is flat and smooth, unlike that of the B1, B2 and B3 features which is quite rough because of a thick debris cover. Light ogive bands developed on the B5 feature are recognizable about halfway down the icefall, but farther down they become indistinct. Those on the A1 feature have also become much less distinct, and few of the arcuate light bands could be identified. Consequently, the moving average of the width could not be plotted for many of the ogives, and a dummy figure had to be employed. The poor recognition of light ogive bands in the upper part of the ablation area may be attributed to the increased exposure of bedrock in the icefall. This causes an increase in the supply of supraglacial debris, which, combined with thinning of ice in the lower half of the glacier, leads to greater accumulation of debris on the surface.
The width of a pair of light and dark ogive bands can be taken as the annual flow velocity at that point in a stable glacier (Reference King and LewisKing and Lewis, 1961; Reference HaefeliHaefeli, 1966; Reference Aniya and NaruseAniya and Naruse, 1987). Although Glaciar Soler has been retreating, the width–velocity relationship may be taken as roughly true for the study period. Figure 4 show plots of the ogive band spacing and the estimated annual flow velocity from the field measurements, with the distance from the base of the large icefall. The flow velocity was measured in the field during the periods 23–25 December 1983, 21 October–18 November 1985 and 15 November–10 December 1998; the daily average was calculated from these measurements. The estimated annual flow velocity was obtained by simply multiplying the daily average by 365. Although the field measurements were carried out 1year earlier for the 1999 ogive spacing, it is assumed that there were no substantial differences. Each year, variation patterns of the ogive spacing and field-measured flow velocities are similar, although in 1984 and 1998/99 field-measured flow velocities were greater than the ogive spacing, while in 1986 ogive spacing was greater than the field-measured flow velocity. This is attributed to the different dates used for the measurement periods, and the likelihood that the amount of subglacial water was different (Reference Aniya, Casassa and NaruseAniya and others, 1988). Indeed, the field-measured flow velocity increased from October to December, i.e. from spring to summer, although the ogive spacings are more or less similar in the three periods. At Glaciar Soler, the contribution of the basal sliding, which is largely controlled by the amount of subglacial water and development of a complex subglacial stream network at the glacier/bedrock interface, to the flow velocity is estimated to be nearly two-thirds (Reference Naruse, Fukami and AniyaNaruse and others, 1992). The large seasonal variation in flow velocity at Glaciar Soler strongly contrasts with the flow-velocity variation of Glaciar Perito Moreno, HPS, where seasonal variation is small (Reference Naruse, Yamaguchi, Aniya, Matsumoto and OhnoNaruse and Skvarca, 2000). Because Glaciar Soler terminates on land, while Glaciar Perito Moreno calves into a lake, the condition of the snout probablyaccounts for the difference in flow-velocity variations.
Plots of ogive band spacing and field-measured flowspeed.
The number of pairs of ogive bands recognized on the 1984 aerial photographs is 32; for 1986 it is 34. In 1999, only 20 pairs could be recognized. From these sets of aerial photographs an average flow velocity of about 160 ma–1 is calculated for the ablation area. According to the ogive spacing, the flow velocity ranges from about 350 ma–1 near the icefall to some 100 ma–1 near the snout. Comparison of flow-velocity distributions during these three periods reveals that variations in flow velocity in 1984 and 1999 are rather similar, and unlike those in 1986. In 1984 and 1999, the flow velocity decreased fairly regularly downstream, while in 1986 it varied considerably near the middle of the glacier. The flow-velocity pattern in 1986 (Fig. 4b) was compared with the inferred bedrock profile (Reference CasassaCasassa, 1987; Reference AniyaAniya and others, 1988; Fig. 5). About 3.5 km from the icefall, flow velocity increased in both ogive-spacing and field-measured data. In the bedrock profile there is a rise of about 3.5 km from the icefall, and measurement point A6 is located at the top of the subglacial rise. It can be inferred that the irregular flow-velocity pattern around 3–4 km from the icefall was caused by compressing and extending flows that reflect the bedrock profile. The increase in the ogive spacing near the snout in 1999 could be due to errors in delineation of light bands, which are very obscure.
Inferred subglacial topography along the border between the B5 and A1 features, where the ogive band spacing was measured (after Reference AniyaAniya and others, 1988).
Crevasse pattern
Onthe 1999 aerial-photographic mosaic, there is an interesting crevasse pattern near the semicircular debris-covered ice feature located by the proglacial lake on the south (righthand) side of the snout (Fig. 3). Its pattern is circular, with a convoluted upstream end, suggesting a complicated flow regime, probably caused by subglacial topography.
Otherwise, no significant changes in crevasse patterns were recognized between 1984 and 1999. Notable features are splaying crevasses, which can be clearly observed near the southern margin of the middle reach of the glacier. The crevasses meet the margin at angles of about 45°. They are curved up-valley towards the center line and are aligned roughly parallel to the flow direction near the center line. They are formed by composition of shear stress due to lateral drag and longitudinal compressive stress (Reference NyeNye, 1952). This stress field is verified from flow measurements in December 1998 (Reference Yamaguchi and AniyaYamaguchi, 2000), in which flow velocity decreased monotonically from 0.59 md–1 (215ma–1) at Y7 to 0.3 md–1 (110ma–1) at Y3 (see Fig. 3c), which corresponds to a compressive strain rate of –0.04 a–1.
Reference Harper, Humphrey and PfefferHarper and others (1998) concluded from an analysis of crevasse patterns at Worthington Glacier, Alaska, U.S.A., that the splaying and transverse crevasses are translated by the flow from several tens of meters to 100 m, and last no more than 1–2 years. On Glaciar Soler, most of the old, inactive crevasses have quickly disappeared after moving only a few hundred meters due mainly to the high rate of surface melting. This may be a reason why there are no clear crevasses except in the marginal areas of the middle and lower reaches of the ablation area.
Surface lowering
In general, surface lowering between 1984 and 1999 is notable everywhere. Reference Naruse, Yamaguchi, Aniya, Matsumoto and OhnoNaruse and others (2000) reported that between 1985 and 1998 the mean thinning of the surface of the lower half of the ablation area was 42 ±5m. The debris-covered part of the B1 feature along the valley wall has since become an ice-cored moraine, the result of the surface lowering. The white ice of the B2 feature has become shorter by about 400 m between 1984 and 1999. The wave ogives developed here have become less pronounced, with a diminished amplitude. The B3 feature, a medial moraine, has also become shorter by about 500 m.
Concluding Remarks
Using uncontrolled aerial-photographic mosaics of Glaciar Soler, which were produced from 666 cm quasi-vertical aerial photographs taken manually from small aircraft in 1984, 1986 and 1999, the surface morphology and changes over time were analyzed, thereby demonstrating the advantage of using such photographs for studies of small glaciers. From the number of ogive bands, the average flow velocity was deduced to be about 160 ma–1 for the ablation area, and from the ogive-bandspacing a flow velocity ranging from about 350 to 100ma–1 was deduced. Comparison with estimates from short-term field measurements (daily mean multiplied by 365) indicates that the seasonal variation in flow velocity is large, probably due to variation in the amount of basal sliding, and care must be taken when extrapolating these values into annual values.
Acknowledgements
We would like to thank the pilots, C. Leon of Don Carlos Aeroportes Ltda, and F. Mayr of Transportes San Rafael of Coyhaique, Chile, who made an opening in the fuselage of his plane for aerial-photographic surveys. Figure 4 was produced by H. Mizukoshi, Geographical Survey Institute of Japan. The projects were supported by the Ministry of Education, Science, Sports and Culture, Japan.
