Climate-Controlled Variations in Scree Production, Southern Alps, New Zealand

Climate-controlled variations in scree production, Southern Alps, New Zealand

T.C. Hales*   Department of Geological Sciences, University of Oregon, Eugene, Oregon 97403, USA Joshua J. Roering 

ABSTRACT production. Systematic variations in rock up- The interaction of ¯uvial, glacial, and hillslope processes controls the development of lift rate, precipitation, and temperature across mountain belts and their response to tectonic and climatic forcing. Studies on the contri- our study area enable us to relate scree pro- bution of hillslope processes to mountain erosion have focused on bedrock landslides, as duction to climatic- and tectonic-driven they have a profound and readily observed impact on sediment yield and slope morphol- factors. ogy. Despite the ubiquity of scree (or talus) mantled slopes in mountainous terrain, the role of frequent, low-magnitude (Ͻ100 m3) rockfall events is seldom addressed in the STUDY AREA context of landscape evolution. Here we quantify the contribution of rockfall erosion The Southern Alps are formed by oblique across an 80 by 40 km transect in the Southern Alps, New Zealand, by analyzing the collision of the Paci®c and Australian plates spatial extent of scree slopes mapped from aerial photographs and estimating long-term along the Alpine fault (Fig. 1). Uplift rate de- (10±15 k.y.) rockfall erosion rates from the accumulation of slope deposits below bedrock cays from ϳ11 mm/yr at the Alpine fault to headwalls and in debris and alluvial fans. Along the rapidly uplifting, high-rainfall west- Ͻ1 mm/yr at the eastern range front (Adams, ern margin, where high rates of bedrock landsliding have been previously documented, 1980). A prevailing westerly air¯ow creates a scree-mantled slopes are sparse. Rainfall decreases exponentially east of the Main Divide, strong orographic effect with maximum pre- and the proportion of slopes mantled by scree increases monotonically, attaining a max- cipitation of 15 m/yr mid-slope on the western imum value of 70%. The systematic distribution of scree deposits cannot be attributed to range front, decreasing to Ͻ1 m/yr in the east lithologic variation, seismicity, or the legacy of glaciation. Instead, climate may serve as (Grif®ths and McSaveney, 1983) (Fig. 2A). a primary control on scree production, as nearly 70% of the mapped scree deposits in The Southern Alps have been extensively gla- our transect are con®ned to a narrow elevation range of 1200±1600 m above sea level ciated (Suggate, 1990). In the west, U-shaped (masl). Our analysis of altitudinal controls on annual temperature variations indicates valleys have been ¯uvially dissected during that scree production via frost-cracking processes may be maximized between elevations the current interglacial, generating steep veg- of 1600 and 2300 masl, as higher elevations are subject to persistent permafrost which etated slopes prone to bedrock landsliding. obviates the frost-cracking process. Rates of rockfall erosion near the rapidly uplifting East of the Main Divide, broad glacial valleys Main Divide are low (Ͻ0.1 mm/yr), whereas rates in the scree-dominated eastern areas have been in®lled by active braided streams average 0.6 mm/yr and may approximately balance rock uplift. and debris and alluvial fans (Adams, 1980). Currently, glaciers are found only at relatively Keywords: rockfall, periglacial processes, Southern Alps, scree slopes. high elevations along the Main Divide. West of the Main Divide, erosion rates via INTRODUCTION of erodable bedrock, freeze-thaw cycles, frost bedrock landsliding and suspended sediment Bedrock erosion by hillslope processes such cracking, frost wedging (Matsuoka, 2001), analyses yield estimates of ϳ2±18 mm/yr and as bedrock landsliding, rock avalanches, and vegetation (which both adds cohesive strength ϳ6±12 mm/yr, respectively (Grif®ths, 1981; debris ¯ows signi®cantly affects the size and to bedrock through root systems and reduces Hovius et al., 1997). In contrast, erosion rates shape of mountains (Schmidt and Montgom- rock strength via chemical and mechanical in the eastern region of our transect have val- ery, 1995; Burbank et al., 1996), and rates of weathering), glacial loading (which creates ues Ͻ1 mm/yr (Grif®ths, 1981) (Fig. 2D). sediment production by bedrock landslides oversteepened headwalls and fractures rock) and debris ¯ows are often signi®cant relative (Miller and Dunne, 1996), and seismic activity METHODS to rock uplift and ¯uvial incision (e.g., Hovius (Matsuoka, 2001). The spatial distribution of scree deposits in et al., 1997). In mountainous areas, scree- Although it has been proposed that funda- the Southern Alps was quanti®ed by aerial mantled slopes are ubiquitous, and localized mental morphologic characteristics of moun- photograph mapping within a 40 ϫ 80 km studies of headwall retreat show that rockfall tain ranges, such as relief, slope, and drainage transect (Fig. 1). Our study transect is bound- occurs at geomorphically signi®cant rates (as density, are controlled by ¯uvial and glacial ed by the Alpine fault at the western margin much as 4.5 mm/yr) (see reviews in Sass and incision (Whipple et al., 1999) and large land- and the Canterbury Plains to the east. We gen- Wollny, 2001). However, the contribution of slides (Schmidt and Montgomery, 1995), the erated two estimates for the extent of scree- small-scale (Ͻ100 m3) rockfalls is seldom ad- topographic imprint of scree-generating pro- mantled slopes: the ®rst represents areas cov- dressed in regional assessments of mountain cesses is poorly constrained. In contrast to in- ered by active scree, and the second includes erosion, because such small events are dif®- tensive ®eld-based studies of headwall dy- scree slopes that are currently mantled by veg- cult to document across large areas. namics, we use the spatially and temporally etation. These methods provide minimum and The regional pattern of rockfall and scree averaged signature of rockfall activity to maximum estimates of scree distribution in production re¯ects the integrated effects of quantify the contribution of scree production our transect, respectively. Active scree was processes typically studied at the scale of in- across the Southern Alps of New Zealand. The characterized as areas of light color on the ae- dividual headwalls. Controls on rockfall activ- distribution of scree slopes re¯ects the accu- rial photo that showed a slope-parallel fabric ity in alpine environments include availability mulation of rockfall events, and debris and al- and exhibited consistently moderate (ϳ30Њ) luvial fans emanating from scree-dominated slope angles (determined from digital eleva- *E-mail: [email protected]. basins provide a postglacial record of scree tion model [DEM] analysis). We identi®ed ar-

᭧ 2005 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. Geology; September 2005; v. 33; no. 9; p. 701±704; doi: 10.1130/G21528.1; 3 ®gures; Data Repository item 2005129. 701 Figure 1. Map of tectonic setting of New Zealand, with subduction in north and south margins, and continental collision occurring along Alpine fault. Inset is shaded relief map of transect across Southern Alps analyzed in this study. Superimposed on map is location of vegetated (red) and unvegetated (orange) scree slopes estimated from aerial photo mapping. eas of vegetated scree as having similarly western sites, we used ground-based surveys moderate slopes combined with evidence for to estimate the area of scree deposits and ap- exposed scree where vegetation has been lo- plied a catenary curve to valley pro®les to es- cally removed. Because our analysis focuses timate the geometry of underlying bedrock on hillslopes, we removed valleys in our cal- slopes (Hirano and Aniya, 1988). Scree pro- culation of the proportion of slopes mantled duction rate was calculated by dividing the by scree. To characterize how the distribution volume of each deposit by the time since de- of scree varies with tectonic and climatic var- glaciation and basin source area. We generated iables, we subdivided our transect into 40 bedrock erosion rate estimates by adjusting for swaths, 2 ϫ 40 km, oriented with the long the density of fan or scree material (1600 Ϯ axis parallel to the Alpine fault. 100 kg mϪ3) and bedrock (2700 Ϯ 100 kg We calculated long-term rockfall erosion mϪ3) (Sass and Wollny, 2001). Because the rates in the eastern part of our transect by es- debris and alluvial fan calculations do not ac- timating the volume of debris and alluvial fans count for scree stored with basins above each accumulated on surfaces of known age. We fan, erosion rate estimates for the eastern sites analyzed fans formed at the mouths of small re¯ect minimum values. (Ͻ1km2) unglaciated drainage basins, where ®eld observations indicated that rockfall activ- RESULTS ity dominates sediment production. During the The distribution of scree, which re¯ects Last Glacial Maximum (LGM) these small rates of production and removal by ¯uvial Figure 2. A: Graph showing minimum (light gray), mean (dark gray), and maximum steep drainage basins drained directly onto processes, varies systematically across our (black) elevations calculated for 2 km large valley glaciers, such that most sediment transect (Fig. 2B). In the western Southern swaths across transect. Black squares rep- produced was likely evacuated. The fans that Alps, scree deposits are small, sparse, and ac- resent mean annual precipitation (Grif®ths developed within the Cass Valley and Lake count for Ͻ10% of hillslope area (Fig. 1). and McSaveney, 1983). B: Minimum (light Coleridge sites accumulated after retreat of These deposits are typically found beneath gray squares) and maximum (dark gray squares) estimates of fraction of hillslopes large valley glaciers ca. 13.5 ka (Suggate, isolated unvegetated headwalls at high eleva- mantled by scree across our transect. Note 1990) (Fig. 1). The Cass Valley was isolated tions. Sparse evidence for relict (vegetated) rapid increase in scree mantling in eastern from further glacial and ¯uvial erosion by a scree slopes exists west of the Main Divide. part of range. C: Variation in elevation of ac- bedrock ridge at the head of the valley such The fraction of scree increases to 10%±20% tive scree deposits (orange zones in Fig. 1). Solid black line and shaded region represent that the fans provide a complete record of at higher elevations along the range axis, mean and standard deviation of elevation post-LGM sediment production from the where scree deposits form along the margins values. Note relatively consistent elevation small basins. In the western half of our tran- of U-shaped valleys and are disconnected of scree slopes. D: Distribution of rock uplift sect (ϳ20 km from the Alpine fault), we es- from the channel network, suggesting that the and erosion rate estimates. Uplift estimates, timated rockfall erosion rates using scree de- preservation potential is relatively high. Bed- using geologic indicators and tilt of paleo- lake terraces (black squares; Adams, 1980), posits situated along the margin of valley rock exposure is common in this area because are approximately equal to estimates of walls that were occupied by glaciers at 10±12 much of the landscape is above the tree limit. scree production in eastern region (this ka (Ivy-Ochs et al., 1999). East of the Main Divide the fraction of total study; black triangles) and bedrock landslid- For the eastern sites, we estimated the vol- (vegetated and active) scree-mantled slopes ing in west (light gray circles) (Hovius et al., 1997). Other estimates of bedrock erosion ume of each debris or alluvial fan using a 25 increases monotonically (Fig. 2B), reaching a from suspended sediment are shown (dark m DEM, interpolating proximal valley wall maximum value (70%) between 40 and 65 km gray circles) (Grif®ths, 1981). MAPÐmean and ¯oor elevations as basal datums. In the from the Alpine fault. The proportion of active annual precipitation.

702 GEOLOGY, September 2005 scree remains relatively constant between 20% strong variations in scree mantling (Fig. 2B). and 30% (Fig. 2B). In this region, scree often At 20 km along our transect, a discrete change mantles entire hillslopes, and some exceed 1 occurs from sandstone-argillite to schist km in length. Convex scree-mantled ridges (MacKinnon, 1983). Although it has been and small isolated bedrock headwalls domi- suggested that schists and other higher-grade nate the headwaters of small basins. Field ob- metamorphic rocks west of the Main Divide servations suggest that scree production in this may obviate scree production (Whitehouse, area is transport limited, such that active trans- 1988), we observe continuity in the distribu- port is required to expose bedrock faces. tion of scree-mantled slopes across this dis- To analyze how climatic variables may af- tinct lithologic boundary. fect rockfall initiation, we calculated the ele- Over geologic time scales, large-magnitude vation distribution of scree slopes across our earthquake events have frequently occurred transect. For this analysis we used the distri- along the Alpine fault, whereas seismic activ- bution of active scree because it allows for ity has been less persistent in the eastern part Figure 3. Annual air-temperature variations comparison with the modern climatic regime. of our transect (Pettinga et al., 2001). During for different elevations in central Southern Despite systematic variations in precipitation, large-magnitude earthquakes, rockfall activity Alps. Temperature data averaged over peri- rock uplift, and relief, the mean elevation of is concentrated near the epicenter and decays od >50 yr at The Hermitage, Mount Cook (762 m, light gray line) were ®t with repre- active scree slopes (Fig. 2C) is surprisingly with distance (Adams, 1980). Given that the ϳ sentative spline (dark line). Using calculated -consistent at 1400 m, and nearly 70% of the Alpine fault is the dominant seismic source in lapse rate of 0.6 ؇C/100 m (see text), we re deposits occur between 1200 and 1600 masl. our study area, our observed scree distribution produced this spline at 1554 m and 2300 m. Minimum erosion rates calculated from is inconsistent with earthquakes as the pri- Recreated spline at 1554 m was compared measurements of 15 debris and alluvial fans with 5 yr worth of data collected at that el- mary scree-generating mechanism. evation (light gray line) and shows faithful in the eastern region (Cass Valley and Lake Glaciation has been proposed as a control reproduction of annual variation at that ele- Coleridge) have a mean value of 0.6 Ϯ 0.4 on rockfall activity by steepening headwalls vation. At 1554 m elevation, <10 d ®t within -mm/yr (Table DR1).1 Near the Main Divide, and increasing fracture density by glacial frost-cracking window (؊3to؊8 ؇C), com our ®eld-based estimates of scree production loading and valley modi®cation (Miller and pared with >60 d at 2300 m. rates are uniformly low (0.01 mm/yr). Dunne, 1996). We recognize the importance of these mechanisms in generating rockfall in vations and a large proportion of the land- DISCUSSION AND CONCLUSIONS the Southern Alps; however, if glacial activity scape above the tree limit. Previous studies of individual headwalls is dominant in setting the stage for scree pro- The ef®cacy of periglacial processes at pro- have recognized many factors that control duction, scree deposits should be most prev- ducing scree depends on altitudinal tempera- scree production, including rock type, vege- alent along the Main Divide, coincident with ture variations. Recent studies of rock defor- tation, temperature, earthquakes, and time the maximum extent of glaciation. Instead, our mation in periglacial environments suggest since deglaciation (e.g., Sass and Wollny, analyses suggest that the zone of maximum that the accretion and subsequent freezing of 2001), although the relative importance of scree extent is offset from the axis of maxi- water in large pores may be a dominant factor these factors remains elusive. Our method mum glaciation by ϳ40 km. in rock fracture or displacement (this process uses regional patterns in scree distribution, The concentration of active scree within a which increases systematically from west to is commonly referred to as frost cracking) narrow and distinct altitudinal zone (Fig. 2C) (Walder and Hallet, 1985). Theoretical (Wald- east, to infer how climatic and tectonic factors suggests that rockfall activity is controlled by in¯uence rockfall activity across the Southern er and Hallet, 1985), experimental (Hallet et the integrated effects of precipitation and tem- al., 1991), and ®eld-based studies (Anderson, Alps. Variations in lithology and metamorphic perature (which vary with elevation and gov- grade do not account for the observed scree 1998) suggest that frost cracking is ef®cient ern vegetation patterns). West of the Main Di- Ϫ Ϫ distribution. Jointed sandstones and argillites within a limited temperature range ( 3to 8 vide, high pore pressures associated with high Њ of the Torlesse Supergroup are dominant from C) and requires available water. This suggests rainfall rates have been proposed to enhance that where water is readily available, the rate 20 to 80 km along our study transect. These bedrock landsliding on steep, heavily vegetat- rocks have been deformed such that no pref- of frost cracking depends on the proportion of ed slopes, where Hovius et al. (1997) docu- Ϫ erential trends in sandstone/shale ratio or joint time that rock is at temperatures between 3 mented rapid rates of landslide-driven sedi- Ϫ Њ orientation exist (MacKinnon, 1983). As a re- and 8 C (Fig. 3). As mean annual temper- ment ¯ux (1.2 ϫ 105 to 1.1 ϫ 106 m3/yrϪ1). Ϫ Њ sult, systematic changes in mechanical rock- ature (MAT) falls below 1 C, however, a Between 0 and 10 km along our transect a fall susceptibility appear unlikely, as estimates permafrost condition ensues (Anderson, small fraction of the landscape has elevation of both rock mass strength (Augustinus, 1995) 1998), limiting the amount of available water values above the tree limit (ϳ1400 m) (Fig. and rock quality designation (Deere and and reducing the effectiveness of the frost- 2A), and dense root systems may suppress sig- Deere, 1988) (Table DR2; see footnote 1) gen- cracking process. This ``high and frozen'' con- ni®cant rockfall erosion in favor of deeper erate consistent values near the Main Divide dition may exist at elevations above 2300 m bedrock failures. Approaching the Main Di- and in the east region of our transect, despite in the Southern Alps, consistent with obser- vide, mean and maximum elevations increase vations of sparse scree and frequent deep-seated Ͼ 1GSA Data Repository item 2005129, Table markedly, such that a large proportion of the bedrock landslides at high elevations ( 2500 DR1, calculated erosion rates from fan volumes, landscape is above the tree limit and scree- m) (McSaveney, 2002). At MAT above Ϫ1 ЊC and Table DR2, values for different sites within mantled slopes become more prevalent (Fig. (or elevations below 2300 m), frost cracking the Southern Alps, is available online at 2B). East of the Main Divide, the proportion may be a viable scree-production mechanism. www.geosociety.org/pubs/ft2005.htm, or on request of the landscape covered with active scree re- from [email protected] or Documents Secre- Analysis of annual air-temperature ¯uctua- tary, GSA, P.O. Box 9140, Boulder, CO 80301- mains approximately constant (20%±30%), tions enables us to quantify the frequency with 9140, USA. consistent with relatively high maximum ele- which rocks of a particular elevation may oc-

GEOLOGY, September 2005 703 cupy the frost-cracking temperature window. sect, uplift estimates based on the tilting of Hovius, N., Stark, C.P., and Allen, P.A., 1997, Sedi- We averaged and smoothed (using splines) paleolakeshores reveal similar rates (Ͻ1 mm/ ment ¯ux from a mountain belt derived by land- ϩ slide mapping: Geology, v. 25, p. 231±234, 20 yr of daily temperature data for 5 sites yr), but their applicability to our study area doi: 10.1130/0091-7613(1997)025Ͻ0231: in the central Southern Alps, spanning 800 m has not been demonstrated (Adams, 1980). SFFAMBϾ2.3.CO;2. of elevation (738±1554 m) (National Institute Despite these uncertainties, available rock up- Ivy-Ochs, S., Schuluchter, C., Kubik, P.W., and Den- for Water and Atmosphere, 2004) (Fig. 3). We lift estimates in the eastern region of the ton, G.H., 1999, Moraine exposure dates im- ply synchronous Younger Dryas glacier ad- compared the difference between smoothed Southern Alps approximate estimates of long- vances in the European Alps and in the splines at each of these ®ve sites and calcu- term rockfall erosion, suggesting that rockfall Southern Alps of New Zealand: Geogra®ska lated an average lapse rate of 0.6 Ϯ 0.1 erosion may balance tectonic mass ¯ux. Annaler, v. 81A, p. 313±323. ЊC/100 m. Using the estimated lapse rate, we MacKinnon, T.C., 1983, Origin of the Torlesse ter- rane and coeval rocks, South Island, New Zea- calculated the temperature distribution for ACKNOWLEDGMENTS Funding for this work was provided by National land: Geological Society of America Bulletin, 2300 m elevation (Fig. 3). At 1554 m (the Science Foundation grant EAR-0309975. We thank v. 94, p. 967±985, doi: 10.1130/0016-7606 highest elevation of available temperature Jarg Pettinga, Ben Mackey, and Kerry Leith for (1983)94Ͻ967:OOTTTAϾ2.0.CO;2. data) Ͻ10 d fall within the frost-cracking win- ®eld assistance and discussion. Bill Dietrich, Dave Matsuoka, N., 2001, Direct observation of frost dow during a typical year. In contrast, at 2300 Montgomery, and Simon Broklehurst provided wedging in alpine bedrock: Earth Surface Pro- cesses and Landforms, v. 26, p. 601±614, doi: m, much of the winter (Ͼ60 d) is spent within thoughtful reviews that greatly improved the manuscript. We also thank Kevin McGill and the Na- 10.1002/esp.208. the frost-cracking zone, suggesting that a tional Institute for Water and Atmosphere for access McSaveney, M.J., 2002, Recent rockfalls and rock small range of elevations (1600±2300 m) has to daily temperature data. avalanches in Mount Cook National Park, the potential for sustained frost cracking. New Zealand: Reviews in Engineering Geol- ogy, v. 15, p. 35±70. Scree deposits form via rockfall erosion of REFERENCES CITED Adams, J., 1980, Contemporary uplift and erosion Miller, D.J., and Dunne, T., 1996, Topographic per- bedrock headwalls, thus the concentration of of the Southern Alps, New Zealand: Geolog- turbations of regional stresses and consequent scree within a limited elevation zone (1200± ical Society of America Bulletin, v. 91, bedrock fracturing: Journal of Geophysical 1600 masl) slightly below the frost-cracking p. 1±114. Research, v. 101, p. 25,523±25,536, doi: 10.1029/96JB02531. window is consistent with frost cracking as Anderson, R.S., 1998, Near-surface thermal pro®les in alpine bedrock: Implications for the frost National Institute for Water and Atmosphere the primary control on rockfall erosion. weathering of rock: Arctic and Alpine Re- (NIWA), 2004, Database of daily temperature The pattern of all mapped scree represents search, v. 30, p. 362±372. data for the central Southern Alps, New a temporally integrated signature of rockfall Augustinus, P., 1995, Rock mass strength and the Zealand. activity (Fig. 2B). During the last glacial max- stability of some glacial valley slopes: Zeit- Pettinga, J.R., Yetton, M.D., Van Dissen, R.J., and Downes, G., 2001, Earthquake source identi- imum, the altitudinal zone of ef®cient frost schrift fuÈr Geomorphologie, v. 39, p. 55±68. Batt, G.E., and Brandon, M.T., 2002, Lateral think- ®cation and characterisation for the Canter- cracking was considerably lower (Suggate, ing: 2-D interpretation of thermochronology in bury region, South Island, New Zealand: New 1990), such that a signi®cant fraction of ter- convergent orogen settings: Tectonophysics, Zealand Society for Earthquake Engineering rain in the eastern part of our transect may v. 349, p. 185±201, doi: 10.1016/S0040- Bulletin, v. 34, p. 282±317. Sass, O., and Wollny, K., 2001, Investigations re- have undergone frequent rockfall activity. 1951(02)00053-7. Brozovic, N., Burbank, D.W., and Meigs, A., 1997, garding alpine talus slopes using ground- This notion is consistent with our observation Climatic limits on landscape development in penetrating radar (GPR) in the Bavarian Alps, that as much as 70% of hillslopes in the area the northwestern Himalaya: Science, v. 276, Germany: Earth Surface Processes and Land- reveal scree mantling. By contrast, higher el- p. 571±574. forms, v. 26, p. 1071±1086, doi: 10.1002/ esp.254. evation zones near the Main Divide have had Burbank, D.W., Leland, J., Fielding, E., Anderson, R.S., Brozovic, N., Reid, M.R., and Duncan, Schmidt, K., and Montgomery, D.R., 1995, Limits less time within the altitudinally de®ned frost- C., 1996, Bedrock incision, rock uplift and to relief: Science, v. 270, p. 617±620. cracking zone (Fig. 2B). threshold hillslopes in the northwestern Him- Suggate, R.P., 1990, Late Pliocene and Quaternary In the absence of profound lithologic or alayas: Nature, v. 379, p. 505±510, doi: glaciations of New Zealand: Quaternary Sci- ence Reviews, v. 9, p. 175±197, doi: 10.1016/ seismic controls on rockfall activity, these re- 10.1038/379505a0. Deere, D.U., and Deere, D.W., 1988, The Rock 0277-3791(90)90017-5. sults highlight the potential for feedbacks be- Quality Designation (RQD) Index in practice, Tippett, J.M., and Kamp, P.J.J., 1993, Fission track tween mountain elevation and erosional pro- in Kirkaldie, L., ed., Rock classi®cation sys- analysis of the late Cenozoic vertical kine- cesses. Whereas previous studies have focused tems for engineering purposes: Philadelphia, matics of continental Paci®c crust, South Is- American Society for Testing and Materials, land, New Zealand: Journal of Geophysical on local relief and glacial equilibrium altitude Research, v. 98, p. 16,119±16,148. in assessing topographic controls on denuda- p. 91±101. Grif®ths, G.A., 1981, Some suspended sediment Walder, J., and Hallet, B., 1985, A theoretical model tion (e.g., Schmidt and Montgomery, 1995; yields from South Island catchments, New of the fracture of rock during freezing: Geo- Brozovic et al., 1997), here we illustrate the Zealand: Water Resources Bulletin, v. 17, logical Society of America Bulletin, v. 96, p. 336±346, doi: 10.1130/0016-7606(1985) p. 662±671. importance of absolute elevation in dictating 96Ͻ336:ATMOTFϾ2.0.CO;2. Grif®ths, G.A., and McSaveney, M.J., 1983, Distri- climatic controls on erosion mechanics. Whipple, K.X, Kirby, E., and Brocklehurst, S.H., bution of mean annual precipitation across 1999, Geomorphic limits to climate-induced Quanti®cation of erosion rates in the west- some steepland regions of New Zealand: New increases in topographic relief: Nature, v. 401, ern portion of the Southern Alps suggests that Zealand Journal of Science, v. 36, p. 197±209. p. 39±43. Hallet, B., Walder, J., and Stubbs, C.W., 1991, sediment ¯ux may accommodate tectonic Whitehouse, I.E., 1988, Geomorphology of the cen- Weathering by segregation ice growth in mi- mass in¯ux into the orogen (Hovius et al., tral Southern Alps, New Zealand: The inter- crocracks at sustained sub-zero temperatures: action of plate collision and atmospheric cir- 1997). Uplift rates of 0.5±1.0 mm/yr for the Veri®cation from an experimental study using eastern Southern Alps have been estimated via culation: Zeitschrift fuÈr Geomorphologie acoustic emissions: Permafrost and Periglacial Supplementeband, v. 69, p. 105±116. ®ssion-track analyses (Tippett and Kamp, Processes, v. 2, p. 283±300. 1993), although these rates remain ambiguous Hirano, M., and Aniya, M., 1988, A rational expla- Manuscript received 21 January 2005 because the effects of lateral advection on nation of cross-pro®le morphology for glacial Revised manuscript received 3 May 2005 valleys and of glacial valley development: Manuscript accepted 4 May 2005 ®ssion-track ages are dif®cult to constrain Earth Surface Processes and Landforms, v. 13, (Batt and Brandon, 2002). South of our tran- p. 707±716. Printed in USA

704 GEOLOGY, September 2005