Earth and Planetary Science Letters 152Ž. 1997 59±73

Cosmogenic dating of fluvial terraces, Fremont River,

James L. Repka a, Robert S. Anderson a,), Robert C. Finkel b a Department of Earth Sciences and Institute for Tectonics, UniÕersity of California, Santa Cruz, CA 95064, USA b Geosciences and EnÕironmental Technology DiÕision and Center for Accelerator Mass Spectrometry, Lawrence LiÕermore National Laboratory, LiÕermore, CA 94550, USA Received 19 March 1997; accepted 25 August 1997

Abstract

Absolute dating of river terraces can yield long-term incision rates, clarify the role of climate in setting times of aggradation and incision, and establish the rates of pedogenic processes. While surface exposure dating using cosmogenic 10Be and 26Al would seem to be an ideal dating method, the surfaces are composed of individual clasts, each with its own complex history of exposure and burial. The stochastic nature of burial depth and hence in nuclide production in these clasts during exhumation and fluvial transport, and during post-depositional stirring, results in great variability in clast nuclide concentrations. We present a method for dealing with the problem of pre-depositional inheritance of cosmogenic nuclides. We generate samples by amalgamating many individual clasts in order to average over their widely different exposure histories. Depth profiles of such amalgamated samples allow us to constrain the mean inheritance, to test for the possible importance of stirring, and to estimate the age of the surface. Working with samples from terraces of the Fremont River, we demonstrate that samples amalgamated from 30 clasts represent well the mean concentration. Depth profiles show the expected shifted exponential concentration profile that we attribute to the sum of uniform mean inheritance and depth-depen- dent post-depositional nuclide production. That the depth-dependent parts of the profiles are exponential argues against significant post-depositional displacement of clasts within the deposit. Our technique yields 10Be age estimates of 60"9, 102"16 and 151"24 ka for the three highest terraces, corresponding to isotope stages 4, 5d and 6, respectively. The mean inheritance is similar from terrace to terrace and would correspond to an error of ;30±40 ka if not taken into account. The inheritance likely reflects primarily the mean exhumation rates in the headwaters, of order 30 mrMa. q 1997 Elsevier Science B.V.

Keywords: cosmogenic elements; terraces; fluvial features; absolute age

1. Introduction begins to incise again, abandoning its old flood plain. Accurate dating of strath terrace deposits can Strath terraces represent ancient floodplains of a constrain the rate of incision of the stream system as river that is incising into bedrock. They are formed well as the timing of the events that control incision when, after remaining at the same elevation and rates. This information can be used to determine the widening its floodplain for a period of time, a river response time of a fluvial system to base level lowering or to rock uplift within the drainage basin. ) Corresponding author. Fax: q1 408 459 3074. E-mail: jl- River terraces are excellent chronosequence sites for [email protected] the study of soil profile developmentŽ e.g.,wx 3. , clast

0012-821Xr97r$17.00 q 1997 Elsevier Science B.V. All rights reserved. PII S0012-821XŽ. 97 00149-0 60 J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 weathering, and eolian inflation, the rates of which We describe here an amalgamation technique may be constrained by dating the deposits. which greatly reduces the effort required to deter- Dating fluvial terraces is often difficult, however. mine mean cosmogenic nuclide concentrations of If suitable organic remains can be located, 14C may surfaces composed of many clasts with disparate be employed to date material deposited in the most exposure histories. Anderson et al.wx 15 outlined this recent of the late Pleistocene climatic cycles. Some method and presented preliminary results. In this terraces sequences have been dated using tephra paper we further illustrate the inheritance problem wx3,4 , or by correlation with moraines wx 5,6 . UrTh and our proposed dating method with a set of numer- dating of soil carbonate coatings on subsurface clasts ical simulations, and then employ the method to has been employed in some settingsŽ e.g.,wx 3,7. , estimate the ages of the Fremont River terraces. although this technique suffers from an unknown lag between deposition of the clast and accumulation of the innermost carbonate coating. 2. Description of Fremont River terraces While in situ produced cosmogenic radionuclides 10 Be, 26Al, and 36 Cl are now widely used for dating The Fremont River drains the basalt-capped of bedrock surfaceswx 8±11 , this method has seen Aquarius and plateaus and cuts through limited use in depositional environments. Phillips two monoclines, the Waterpocket Fold and Caineville and co-workerswx 3,8 have used36 Cl to date large Reef, before joining the Muddy River near Han- boulders on moraines and associated outwash ter- ksville to form the , a tributary to races. We argue that a significant source of uncer- the Colorado RiverŽ. Fig. 1 . Tills and outwash de- tainty in this and all other depositional systems arises posits found in the drainages skirting the Aquarius from the accumulation of nuclides in the sampled Plateau indicate that the Fremont has at times been clasts prior to depositionŽ. i.e. `inheritance' . glacier fed. The Blind Lake and Donkey Creek tills Previous studies on depositional surfaces, includ- have been correlated to the Pinedale glaciation and ing our own work, have revealed significant scatter the Carcass Creek to the Bull Lake glaciation, based in the effective ages derived from individual clasts on morphology and extent of the deposits and on the sampled from such surfaces. At issue is how to degree of weathering of the clastswx 16 . We have interpret this scatter. One possible source of scatter is focused on the reach near North and South Caineville post-depositional turbation of clasts within the de- MesasŽ. Fig. 2 , 15±20 km east of Capitol Reef posit, resulting in a mean nuclide production rate that National Park. is lower than that of the surface. In this case the Four terraces can be traced through the field area, surface clast with the largest effective age would ranging in elevation from 20 to 135 m above the provide a lower bound on the age of the deposit. If modern floodplainwx 5,17 . Each terrace consists of a the scatter reflects instead the stochastic nature of several meter thick gravel veneer capping a strath pre-depositional nuclide inheritance, the surface clast surface eroded into shale and sandstone bedrock. We with the lowest effective age provides an upper have labeled them FR0±FR4 from lowest to highest bound on the age of the deposit. The resolution Ž.Fig. 2 , FR0 referring to the modern floodplain. requires sampling of the subsurface and averaging FR1 and FR4 exist within the field as isolated rem- over many clasts to deal with the stochastic nature of nants, with maximum widths of a few tens of meters. inheritance. If the deposit has been static since depo- We have designated sublevels of FR2 as FR2A-C, sition, all clasts will have experienced post-deposi- from lowest to highest. FR2A and FR3 are the most tional nuclide production rates determined only by extensively preserved of the terraces. their depth within the deposit. The mean concentra- No absolute dates for either the moraines or the tion profile should be an exponential profile that fluvial terraces of the Fremont River exist. Howard asymptotically approaches the mean inheritance at wx5,17 relied upon careful correlation of the terraces depth. If there has been significant post-depositional with the morainal sequence on stirring of clastsŽ. bio-, pedo- or cryoturbation , this Ž.Fig. 1 , correlating terrace FR2A Ž his 4A . with the expected profile will be disrupted in some way. Pinedale glaciation, and terrace FR3Ž. his FR3A with J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 61 the Bull Lake glaciation. Recent dating of glacial deposit makes a relatively planar contact with under- deposits in the type localities adjacent to the Wind lying shale and sandstone bedrock. River Range, Wyoming, reveals that each glaciation Clasts range in diameter up to tens of centimeters. consists of numerous advanceswx 3,25 . Clearly, abso- The deposits are dominated by locally derived sand- lute dating is needed to unravel the timing within stone and shale, quartzite and chalcedony from the any particular system. Triassic Shinarump conglomerateŽ presently exposed All of the Fremont terraces have in common a 50 km upstream in Capitol Reef National Park. , and single layer of varnished clasts, forming an interlock- basalts from lava flows that mantle Boulder Moun- ing desert pavement surface capping a thin Ž-10 tain and Fish Lake Plateauwx 18 . In our cosmogenic cm. layer of nearly clast-free silt, probably reflecting radionuclide dating we have used only the well eolian inflation of the surface. The silt layer caps preserved and ubiquitous quartzite clasts on these several meters of river gravels and sands. The terraces. The sizes of the quartzite clasts are similar stratigraphy is consistent with that of a braided chan- from terrace to terrace. Their extreme resistance to nel. The total gravel thickness varies from less than 2 weathering minimizes the problem of erosional loss to roughly 10 m. Where visible, the base of the from the sampled clasts. Finally, as the age of the

Fig. 1. Map of the Fremont River drainage upstream of the field area. Drainage areaŽ. dashed line is just over 3700 km2 . The solid line surrounding the high plateaus represents the equilibrium line altitudeŽ. ELA during the last glacial maximum, estimated to be 3170±3260 m wx11 . The Caineville area within which the Fremont terraces are studiedŽ. Fig. 2 lies in the box east of Capitol Reef National Park. 62 J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73

Fig. 2. Detailed topography adjacent to the Fremont River as it passes eastward through the gap between North Cainville Plateau Ž.NCP and South Caineville Plateau Ž.SCP . The sampled terraces are depicted. Note the single small scraps of FR1 and FR4 surfaces, and the extensive preservation of FR2 and FR3 surfaces. conglomerate from which they are derived is much storage at any time step, and its depth of burial if it greater than the half-lives of the radionuclides, we is in storage. We specify that fluvial transport takes ignore inheritance from exposure in the ancient geo- place only in short episodes during which the clast is morphic system. at the surface; the rest of the time the clast is buried at a depth that is some fraction of a specified maxi- mum. At the end of the transport cycle, clasts are 3. Numerical simulation of our technique instantaneously deposited on a terrace at some speci- fied depth. Clasts deposited on the surface experi- We present a numerical model of cosmogenic ence steady production, while clasts deposited be- radionuclide accumulation histories for clasts within neath the surface experience production that de- a hillsloperfluvial transport system that both illus- creases with time due to eolian inflation of the trates the problem of constraining the terrace ages depositwx 26 . At each time step, the model determines and provides a theoretical backdrop for our dating where the clast is within the systemŽ on the hillslope, strategy. in the river, or on the final terrace. , determines the Our model numerically integrates the differential clast depth and the associated cosmogenic radionu- equation for production and decay through time. clide production rate, and updates the concentration. Once the clasts are in the fluvial system, a series of This model incorporates several assumptions, three random numbers dictate how long the clast will some of which are applicable to most fluvial sys- spend within the system, whether it is in transport or tems, while others should be considered specific to J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 63

the Fremont River: We assume a constant cosmic ray and Ns is the nuclide accumulation at the surface flux and therefore production rate. Cosmic ray irradi- since deposition. This approximation is valid for ation is assumed to be negligible before hillslope times short relative to the decay time of the nuclide exhumation brings the clast near the surface. We measured, and for shallow depths within which pro- neglect variability in the hillslope exhumation pro- duction by muons and by other less important mech- cess, assuming it to be spatially and temporally anisms is negligible. Since there are two unknowns, uniform within the catchment. We assume that topo- Nin and N s , two measurements suffice to solve the graphic shielding from cosmic rays by snow and equation. Using this `pairs technique' we can calcu- vegetation is insignificant. late the age of the surface: 1 DP 3.1. Theoretical background ts lnŽ. 4 l ž/DPylDN Cosmogenic nuclides are produced in situ from interactions between secondary cosmic ray particles where DP is the difference in production rates be- and material at the earth's surface. Production de- tween the samples and DN refers to the difference creases exponentially with depth: between the concentrations of the two samples. For ) young terraces, where decay may be neglected, this PzsPeyz r z 1 Ž.o Ž. becomes tsDNrDP. ) L where z s rr is the e-folding length, or the ratio In the case of a surface inflating by eolian deposi- of the absorption mean-free path Ž.L of ;145±160 tion, the assumption of a constant production rate y2 wx g cm and the material density Ž.r 20,21 . In within the subsurface is incorrect. In this case our quartz, which is an ideal mineral for measurement measured depthŽ. z overestimates the mean depth wx26 s s 9,10 , AlŽ. t1r2 0.705 Ma is produced primarily since deposition and thus underestimates the mean from silicon, while oxygen is the primary target for nuclide production rate for the subsurface sample. In 10 s production of BeŽ. t1r2 1.5 Ma . addition, the production rate difference, DP, is a For exposure at a constant depth: function of time. If we make the simplest assumption

P0 ) that silt thickness begins at zero and increases at a Nz,tŽ.sN Ž z,0 . eylt q e1yz r z Žye2ylt . Ž. l steady rate to its measured depth of L at the time of s r Ž. Here the first term represents the decay of the inher- sampling, the correction factor R P P zs for the ited concentration, NŽ. z,0 , while the second term mean production rate for the subsurface sample is: ) represents the net increase in concentration due to z ) RLŽ.s wxeL r z y15 Ž. production and subsequent decay. L The mean concentration of a cosmogenic nuclide Ž. at any particular depth, N, is approximated by mea- Eq. 4 should therefore be calculated with the dif- PzŽ. z) suring `amalgamated' samples consisting of equal s y s = ference in production rate of DP P0 mass aliquots of a large number of single clastswx 15 . ) L weL r z y1x . While this correction can become large Eq.Ž. 2 may now be rewritten as an equation for the Žof the order of 30% for a silt thickness Ls0.5 z) . , mean concentration. The difference in the concentra- the 10 cm thick silt caps of the Fremont River tion between two amalgamated samples, one from terracesŽ and z ) s80 cm. result in only a 6% in- the surface and one from a known depth within the crease of the age estimate. deposit, then reflects the expected difference in the deterministic post-depositional grow-in of cosmo- genic radionuclides represented by the second term 3.2. Model results in Eq.Ž. 2 . Ignoring decay, a full profile of the mean Fig. 3 illustrates the stochastic nature of the inher- concentrations should reveal a simple shifted expo- 10 nential: itance problem, showing simulated Be concentra- tion histories for a large number of clasts for two s q yz r z ) NzŽ.NinN s e Ž.3 end-member scenarios. The terrace age was pre- where Nin is the mean pre-depositional inheritance scribed to be 100 ka, the surface production rate 64 J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73

s r r P0 6 atom gm yr. In Fig. 3A, the exhumation spread in the final concentrations of each group rate is 500 mmryr, the maximum fluvial transport reflects the stochastic nature of the pre-depositional time is 200 ka, and the post-depositional eolian geomorphic processes. In the top run, Fig. 3A, the inflation rate is 10y3 mrka. In Fig. 3B the exhuma- age estimated using Eq.Ž. 4 is 113 ka, reasonably tion rate varies randomly among the clasts, ranging close to the prescribed 100 ka age of the surface. The from 17.5 to 42.5 mmryr, while fluvial transport is inset in Fig. 3B shows the dependence of the age zero, and the eolian inflation rate is zero. The large estimate on the number of clasts used in the amalga-

Fig. 3. Simulated 10 Be concentration histories of many clasts from hillslope exhumation through fluvial transport to terrace deposition. A 10 s r r steady sea level surface Be production rate of P0 6 atoms gm yr is assumed. At the final site of deposition, half of the clasts are deposited in the subsurface at an original depth of 1.0 m, the other half are deposited on the surface. The terrace age is ascribed to 100 ka. Ž.A Fluvial inheritance dominates. Twenty-four clast histories are shown. Exhumation of all clasts on hillslopes is rapidŽ. 500 mmryr . Clasts spend a random time between 0 and 200 ka in fluvial transport. Eolian silt inflates the terrace surface by 0.1 m over 100 ka. An age " s " s " = 6 estimate for the terrace of 113 21 ka is obtained using the mean concentrations of Nsss0.68 0.10 and N 0.25 0.12 10 atomrgm and Eq.Ž.Ž 4 illustrated as the intersection of gray lines back-tracking concentration histories from mean concentrations.Ž. . B Hillslope exhumation dominates inheritance. Twenty clast histories are shown. Exhumation rate is chosen randomly from between 17.5 and r 42.5 mm yr. Transport time within the fluvial system is negligible. The mean concentrations of the 10 surface clastsŽ. Ns and of the 10 subsurface clastsŽ Nss are shown, from which DN is calculated. Inset depicts dependence of such age estimates on the number of clasts used in estimating the surface and subsurface mean concentrations, shown for 5 separate runs of 100 total clasts eachŽ 50 each in surface and subsurface. . J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 65 mated samples. The age estimate improves as more 3.3. RelatiÕe importance of exhumation and fluÕial and more clasts are amalgamated. transport In Fig. 4 we depict similar simulations for four hypothetical terracesŽ. A±D whose height above the Nuclide inheritance accrues during hillslope ex- floodplain is dictated by a linear incision rate of 1.2 humation and fluvial transport. Ignoring for the mo- mrka. We show the effective individual clast ages ment the decay of the cosmogenic radionuclides, the r Ž.N P0 , as well as the effective mean age, for all inheritance during exhumation by steady surface surface clasts on each terrace. We then use Eq.Ž. 4 to lowering, eÇ, on hillslopes is the product of the estimate a true surface age for each terrace, using 30 surface production rate with the time it takes the randomly selected clasts both from this surface popu- sample to traverse the final production length scale lation and from the corresponding subsurface popula- z) to the surfaceŽ e.g.,wx 9. : tion at 1 m depth. The ages are depicted by the ) Pz0 circles. It should be obvious that the mean concentra- N s Ž.6 hill e tion obtained from a single amalgamated sample of Ç surface clasts would over-estimate the age of the Variation in Nhill arises from several factors:Ž. 1 surface. If all of the assumptions made in our model differences in surface production rates in the catch- are correct, then Eq.Ž. 4 yields a reasonable approxi- ment due to altitudinal dependence of production mation for the age of the surface. Correction of Eq. rate;Ž. 2 variation in topographic shielding within the Ž.4 for steady eolian inflation Ž with Ls10 cm . catchment, locally reducing the cosmic ray flux be- increases the age estimate by roughly 6%. low that on an unshielded surface; andŽ. 3 variation in the local erosion rate in time and space. Production within the fluvial system is compli- cated by the stochastic nature of the transport and burial processes. Clast depth, zŽ. t , falls within the rangewx O,H , where O represents the surface and H is the depth of the gravel fill. Substituting bsHrz ), the cosmogenic radionuclide accumulation history can be rewritten: s bz ) y r ) NfluvialPT 0H 0 expŽ.Ž.Ž.z z F z dz 7 where T is the total transport time in the fluvial system and F Ž.z dz is the probability of finding the clast within the intervalwx z, zqdz at any time. For the simplest case of uniform probability between the surface and depth H, the integral yields: y yb Fig. 4. Results of numerical simulations, resulting in calculated 1 e 10 N sPTŽ. 8 Be ages for 2000 clasts from four hypothetical terraces in a fluvial 0 ž/b system in which a linear incision rateŽ. 1.2 mmryr; gray line is imposed. The cloud of points at each elevation represents the wide This reduces to the yet simpler expression Nfluvial spread of clast concentrations converted to effective ages through sPTrb in thick depositsŽ H)2z) . . As H in- s r 0 t N P0 , reflecting the variable inheritance of cosmogenic ra- dionuclides during transport to the depositional site. Note that the creases the clast spends less time near the surface youngest single clast age for each terrace is close to the specified and accumulates fewer cosmogenic radionuclides. age of the terraceŽ. intersection of the gray line with that elevation . The simulated accumulation histories shown in Fig. Boxes represent age estimates based upon samples amalgamated 3A suggest that difference in total transport time is from 30 surface clasts selected at random from each of the four the principal factor in clast-to-clast variation in the terraces. The large circles represent our model terrace ageŽ Eq. Ž..4 using the pairs technique, based upon amalgamated surface fluvial component of nuclide inheritance. In addition, and amalgamated subsurface samples. These model ages fall the mean rate of accumulation is significantly lower within 5% of the specified ages for each of the terraces. than the surface rate, P0 . The ratio of these mean 66 J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 production rates, as deduced from the slopes in Fig. ing on the surface can be characterized well by 3A, is roughly 4.6:1, meaning the effective produc- averaging the concentrations of a relatively small s r s tion rate PeffP 0 b, with b 4.6. Input parameters number of clastsŽ. 25±40 . The number of clasts for the model were Hs5 m and z ) s0.8 m; there- necessary to constrain the mean concentration well fore the expected value of bsHrz ) is 5.3. The enough to extract the inheritance signal depends on simulation both illustrates and verifies this simple the width of the distribution of clast inheritance. In statistical treatment. addition, inheritance becomes a smaller portion of The relative contributions from hillslope and flu- the total nuclide concentration on older terraces, vial processes is the ratio of the effective times spent making them more immune to the problems we are in the production zone in each of the geomorphic attempting to solve. We assume that there is no settings: post-depositional `stirring' of the terrace deposit. y yb Finally, we assume that there has been no erosional Nfluvial eÇÇT1Ž.e e T sf Ž.9 loss from the horizontal portions of the terrace sur- b ) Nhill zH faces. This is consistent with the inflationary model wx When the exhumation rate is low and the transport of desert pavement formation 26 . time short, or the thickness of the floodplain storage system is great, the contributions from the hillslope 4.2. Field methods system will dominate the inheritance signal, and vice versa. In the numerical simulations displayed in Fig. Samples were selected from the flattest surfaces, s r s s well away from terrace edges, in order to avoid areas 3A Ž.eÇ 0.5 m ka, Tmax 200 ka, and H 5m, for instance, the expected relative contributions to subject to gravitational creep. Where possible, sub- the inheritance should therefore be roughly 10:1 surface samples were obtained from pits dug in the fluvial:hillslope. A more realistic estimate for the interior of the terrace; otherwise they were from hillslope exhumation rate might be a few tens of areas exposed by roadcuts. The location and eleva- microns per year, and the transport times in the tion of each sample was noted. Topographic shield- fluvial system may not exceed several thousand years ing was determined by measuring the vertical angle during times of high glacial discharge. Hillslope to the horizon measured in eight directions. We exhumation may therefore become an equal or domi- collected 30±70 clasts for amalgamation of surface nant player in producing inheritance, the spread in and subsurface samples. On FR2 and FR3 we col- single-clast cosmogenic radionuclide values reflect- lected additional samples every 0.4±0.5 m, to a ing nonuniformity in the erosion rates within the depth of 1.5±2.0 m, to obtain the concentration catchment. profiles. Clast sizes ranged from 5 cm to 20 cm.

4.3. Estimation of production rates 4. Application to dating of the Fremont River terraces We have used the long-term average high latitude sea level production rate of 5.8 atomrgmryr sug- 4.1. Assumptions gested in the most recent work of Nishiizumi et al. wx24 . This is approximately 13% higher than that of We assume that the deposition rate of the terrace Clark et al.wx 23 and 4% lower than the that of gravels is sufficiently rapid that there is no age Nishiizumiwx 22 . The surface production rate for the structure within the deposit. There is no stratigraphic given latitude and altitude is then adjusted using evidence for significant time spent during deposition coefficients reported in Lalwx 9 . Variations in mag- of the terrace gravels, and rapid deposition is consis- netic field strength introduce uncertainties into pro- tent with our interpretation of these gravels as braided duction rate estimates. Our calculations include a stream deposits. We assume that the cosmogenic production rate uncertainty of 10%. The topographic radionuclide inheritance signal is truly random. We shielding factorwx 22 was very near 1.0 in all cases. assume that the mean inheritance of the clasts arriv- The soil density was estimated using a measured J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 67 1E11 . 12 and s 38.3 33.5 13.1 16.1 32.7 37.4 y 12.1 9.9 9.3 15.0 28.6 9.9 12.1 8.1 6.1 8.0 12.1 2.0 3.1 " " " " " " 1.0 " " " " " " " " " " " " " " 9.86E ion of the three. 8% . Errors were uction rate is then s Ž. " Ž. easured at the CAMS 2 to be 35%, to arrive at a cm 17.8 21.5 233.6 12.8 134.9 14.1 106.3 11.3 104.8 13.4 119.8 r 12.9 12.1 9.3 8.6 87.4 10.0 106.9 7.5 72.1 6.5 60.0 8.8 79.5 10.0 86.8 7.5 69.0 6.9 66.3 6.9 65.5 5.5 39.3 2.6 23.7 5.7 50.9 2.7 16.3 2.4 17.8 " " " " " " 1.3 6.4 Ž. " " " " " " " " " " " " " " " " " " 156 g s

L gm ka ka 2.321 201.8 2.766 121.1 0.633 129.9 1.069 94.2 3.116 124.8 0.748 81.4 0.629 70.1 0.677 56.3 1.195 79.0 2.572 85.3 1.511 68.6 0.962 60.4 0.427 64.4 0.442 46.1 0.745 24.0 1.039 52.4 2.769 106.8 0.107 20.0 0.244 17.9 0.081 6.2 m " " " " " r 5% and " " " " " " " " " " " " " " " " Ž. 2 cm r 14. y 145 g gm atom s 0.057 7.875 0.160 83.5 0.107 99.1 0.102 159.9 0.060 84.0 0.058 21.841 0.045 13.219 0.072 10.418 0.037 8.641 0.041 10.469 0.037 7.186 0.050 6.016 0.080 8.587 0.043 6.805 0.052 6.578 0.034 6.438 0.047 3.851 0.015 2.343 0.035 4.957 0.045 9.937 0.058 11.283 0.028 1.600 0.025 1.744 0.018 0.653 m Be Al r " " " " " " " " " " " " " " " " " " " " " " " "

L 14. Beryllium samples were normalized to LLNL STD10000 10Be:9Be 37 3.306 45 2.020 21 2.135 20 1.352 20 1.561 27 1.167 19 0.941 36 1.311 25 1.130 32 1.002 12 1.056 32 0.747 22 0.847 16 0.323 31 0.289 29 0.105 22 1.416 20 0.390 19 1.707 19 1.985 wx w x y " " " " " " " " " " " " " " " " " " " " wx 1.8 1.401 0.4 1.409 0.7 1.658 0.3 2.638 0.3 363 0.3 216 0.3 442 0.3 260 0.3 207 0.3 344 0.3 184 0.3 244 0.2 77 0.3 113 0.3 229 0.3 211 0.3 295 0.3 62 0.3 105 0.3 70 0.3 71 0.3 331 0.3 234 0.3 487 " " " " " " " " " " " " " " " " " " " " " " " " wx wx B and were uniformly less than 6.0E 10 10 26 10 26 10 26 Ž. Ž. Ž.Ž. Ž. Ž.Ž . Ž . Ž. Ž. 12 . Aluminum blank ratios were uniformly less than 1.0E y 9.92E 3 s clasts km cm ppm ppm atom cm ; we then calculate P z using Eq. 1 and Brown et al.'s 20 values of r 0.1 g " Ž. Al data: single clasts 26 Be and 4155-1.7 FR3 1 38.4 1.43 0 17.1 104 170.5 4155-1.6 FR3 1 38.4 1.44 0 17.1 104 65.5 4155-1.3 FR3 1 38.4 1.44 0 17.1 104 25.3 4155-1.1 FR3 1 38.4 1.44 0 17.1 104 41.4 9073-5 FR3 1 38.4 1.44 0 17.2 105 24.3 9073-4 FR3 1 38.4 1.44 0 17.2 105 22.7 9244-7.6 FR2C 1 38.4 1.42 0 16.9 103 26.5 9244-7.5 FR2C 1 38.4 1.42 0 16.9 103 24.9 9244-7.3 FR2C 1 38.4 1.42 0 16.9 103 24.8 9244-7.2 FR2C 1 38.4 1.42 0 16.9 103 24.4 9244-7.1 FR2C 1 38.4 1.42 0 16.9 103 25.0 9172-5B FR2C 1 38.3 1.42 0 16.9 103 25.0 6202-2E FR2C 1 38.4 1.42 0 16.9 103 16.9 9053-4 FR2B 1 38.3 1.41 0 16.7 102 23.4 3314-3A FR2B 1 38.4 1.41 0 16.8 102 25.1 3314-2A FR2A 1 38.4 1.40 0 16.6 101 25.1 9234-2.13 FR1 1 38.4 1.38 0 16.4 100 24.8 9234-2.8 FR1 1 38.4 1.38 0 16.4 100 24.1 9234-2.3 FR1 1 38.4 1.38 0 16.4 100 25.2 9234-2.2 FR1 1 38.4 1.38 0 16.4 100 25.2 9234-2.1 FR1 1 38.4 1.38 0 16.4 100 25.7 3314-1C flood plain 1 38.4 1.36 0 16.2 99 25.8 Beryllium blank corrections take into account interference by the isobar 3314-1A flood plain 1 38.4 1.36 0 16.2 99 24.7 Sample Terrace No. of Latitude Altitude Depth P Be P Al Be Al Be concentration Al concentration Be model age Al model age adjusted to account fordensity the depth of to 2.1 the subsurface samples: we estimatefacility the terrace at gravels Lawrence to LivermoreKNSTD9919 contain 30% National 26Al:27Al large Laboratory. clasts, Aluminum and samples the porosity were of normalized the to matrix standards prepared by Nishiizumi, KNSTD9860 26Al:27Al Table 1 10 3294-1A flood plain 1 38.3 1.42 0 16.9 103 25.0 Altitudes and latitudes for samples were determined using: hand held GPS; laser EDM;propagated triangulation using the with uncertainties compass given, and with topographic a 10% maps; uncertainty and assigned a to the combinat surface production rate and 7% to the sample depth. All samples were m Production as a function of latitude and altitude is calculated from equations given by Lal 9 and calibrations of Nishiizumi et al. 24 . The surface prod 68 J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73

30% clasts in the terrace deposit, and an estimated porosity of 0.35 within the fines, yielding a density of 2100 kgrm3. We assume no vertical density structure, an assumption that results in negligible error. 4.4. Laboratory methods Clasts were individually crushed and sieved to a uniform size of 200±500 mm, after which amalga- mated samples, each representing a single depth, are constructed from equal mass aliquots of 25±40 indi- vidual clasts. Clean quartz separates were isolated by chemical leachingwx 28 . We added 0.5 mg of Be carrier to a 20 g sample of quartz and dissolved the sample in 3:1 HF:HNO3 . Aluminum concentrations were determined by atomic absorption spectroscopy on an aliquot of the dissolved sample. Al and Be were separated by ion chromatography and precipi- tated as metal hydroxides which were then oxidized to Al23 O or BeO. The ratio of the radionuclide to the stable isotope was determined by accelerator mass spectrometry at Lawrence Livermore National Labo- ratorywx 29 .10 Be concentrations were determined relative to an ICN standard and 26Al to an NBS standard, both prepared by K. Nishiizumi. 10 BerBe ratios were corrected for interference from 10 B. The measured ratios of both 10 Be and 26Al were corrected for process blanks treated in the same way as the samples. The uncertainties quoted are the 1s AMS errors only.

5. Results and discussion 10 26 5.1. Single clasts Fig. 5. Summary of results from Be and Al analyses of Fremont River terrace quartzite clastsŽ data shown in Table 1Ta- 26 10 The spread of effective ages derived from single ble 2. . `s Al results; ls Be results. Ages were calculated Ž. using production rates determined using nuclear disintegration clasts is very large Table 1; Fig. 5A , demonstrating rates reported in Lalwx 9 , and scaled to fit the production calibra- that the problem of inheritance is significant in this tions reported in Nishiizumi et al.wx 24 .Ž. A Results for whole 26 10 geomorphic system. In all cases, Al and Be re- single rock samples. Note the wide spread in effective ages. These sults yield similar ages, implying that processing are all maximum age estimates if the chief error is associated with errors or long burial are not important. Even the inheritance.Ž. B Results for amalgamated surface samples repre- senting at least 25 clasts. Note considerable reduction in age cobbles sampled from the modern stream system scatter from results from single clasts shown inŽ. A . These ages show significant nuclide concentrations. correspond to the boxes in Fig. 4, and should be considered maximum ages. The length of the arrows show downward correc- 5.2. Amalgamated surface samples tion of mean ages from these amalgamated surface samples when Ž.Ž. Results from the surface amalgamated samples inheritance is taken into account see C . C Results of the pairs technique employing two amalgamated samples. Triangles repre- Žnumbers of clasts vary from 25±40; Table 2; Fig. sent the 26Al and 10 Be model ages determined using the profile 5B. show significantly less scatter than those for techniqueŽ. see Fig. 4 . J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 69 single clasts. Most repeat samples using different sets of clasts show some overlap, both in 26Al and in 10 Be. If the spread in effective ages from single clasts represents variable inheritance, then age esti- mates from these surface amalgamated samples are to be considered maxima. 5.3. Paired±amalgamated sample technique In Fig. 5C we plot age estimates of the terraces adjusted for inheritanceŽŽ.. Eq. 4 . The model ages for the higher terraces increase with elevation, al- though the lowest terraceŽ. FR1 appears to be slightly older than the next highest terrace. The arrows on Fig. 5B depict the downward shift in the age esti- mate when inheritance is taken into account. 5.4. Profile technique Fig. 6 shows vertical concentration profiles of four terraces. Two of the terraces, FR2C and FR3, were sampled at several depths. The profile of FR2C consists of two amalgamated samples collected from the surface, one from 0.4 m, two from 0.6 m, and one from 1.0 m depth. The FR3 profile consists of four surface samples, one from 0.5 m, and two each from 1.0 and 1.5 m. The best-fitting shifted exponen- tialsŽŽ.. Eq. 3 fit the shapes of the full profiles quite well. For the higher surfaces, the shift representing pre-depositional inheritance is similar, about 24"8 ka. The inheritance shift for FR1 is minimal. The age estimates based upon these profiles are shown as triangles on Fig. 5C. At this location there is no well expressed terrace that corresponds to the last glacial maximumŽ. LGM . The surface in the proper elevational range is FR1. The pairs technique suggests that FR1 is slightly older than FR2C, which is geomorphically impossi- ble. It is likely that the gravel deposits on this small terrace remnant consist entirely of material quarried from the adjacent older terrace either at the end of the depositional phase, or during the post-deposi- Fig. 6. Full cosmogenic radionuclide concentration profiles for FR2C and FR3, and the amalgamated sample results used in the tional evolution of the scarp separating the two. This 26 10 pairs techniques on all terraces. Note AlŽ. top axis and Be would result in an age estimate for FR1 similar to Ž.bottom axis are plotted on scales that differ by a factor of 6.0. that determined for FR2C, which indeed is what we Curves are best fits to Eq.Ž. 3 , with z) fixed at 0.8 mŽ solid line observe. An alternative explanation is that the LGM 10Be; dashed 26 Al. . The gray vertical lines represent best fits for Ž is represented here by the current floodplainŽ. FR0 . the inheritance, Nin , width indicates error in estimate of inheri- . In either case, given the minor, narrow exposures in tance for FR2C, FR3 and FR4. Inheritance on FR1 appears to be minimalŽ. see text for discussion . The best fit for the post-deposi- this reach of the Fremont River, we are not confident tional component of the radionuclide concentration at the surface, in our dating of FR1. Ns , is then used to estimate the age of the surface. 70 J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 ) ) ) ) ) ) ) ) ) ) ) ) 1E11 . 12 and s 16.0 22.5 17.8 26.4 21.3 21.7 21.2 21.8 23.4 23.7 2.9 1.7 13.1 9.3 9.1 23.1 25.2 10.9 8.3 12.9 11.1 27.3 20.6 20.5 14.4 23.3 y " " " " " " " " " " " " " " " " " " " " " " " " " " 9.86E ion of the three. ) en for subsurface 8% . Errors were uction rate is then s Ž. " at the CAMS facility Ž. 2 to be 35%, to arrive at a cm 12.015.1 132.2 14.1 141.1 14.9 133.3 13.9 16.7 117.3 13.717.5 115.1 19.3 174.4 17.7 171.4 25.9 158.0 149.6 r 4.82.07.4 24.3 7.1 14.4 8.5 66.2 16.5 60.9 17.0 71.3 8.4 60.6 6.7 74.4 9.3 89.8 9.1 66.4 14.2 82.3 13.8 80.8 12.7 45.6 9.9 56.2 60.8 54.4 22.1 13.712.8 85.0 111.2 " " " " " " " " " " " Ž. " " " " " " " " " " " " " " " " " " 156 g s

L gm ka ka 0.681.510.97 113.4 141.4 131.7 0.861.03 160.1 178.4 m 0.150.091.040.64 23.1 0.49 13.6 0.36 69.5 0.26 66.5 0.51 70.9 0.45 66.1 0.92 68.2 0.70 78.6 0.55 62.9 0.39 87.2 0.25 85.5 0.27 38.5 62.5 57.2 59.3 0.530.300.260.28 83.9 107.2 98.9 0.37 106.7 0.50 149.1 143.8 5% and r " " " " " " " " " " " " " " " " " " " " " " " " " " " Ž. 2 cm r 14. y 145 g gm atom s 0.0710.0230.036 2.47 0.035 1.41 0.070 6.39 0.026 5.90 0.025 6.87 0.041 3.52 0.034 2.89 0.044 8.72 0.044 6.58 0.054 8.16 0.035 8.02 0.032 6.16 0.046 4.96 0.050 4.71 0.063 4.04 0.063 12.99 0.102 13.72 0.076 13.01 0.0650.0440.0360.027 7.10 0.026 4.85 0.088 3.88 0.086 3.56 0.048 131.4 17.40 0.082 127.5 17.12 83.0 2.47 3.20 m Be Al r " " " " " " " " " " " " " " " " " " " " " " " " " " " " "

L 14. Beryllium samples were normalized to LLNL STD10000 10Be:9Be 711 0.388 0.219 23 1.119 28 1.073 28 1.143 31 0.563 28 0.546 810 1.285 1.043 22 1.448 20 1.420 18 1.149 44 0.828 20 0.879 18 0.676 12 1.898 16 2.338 14 2.182 22 1.144 27 0.853 27 0.756 27 0.643 15 2.747 17 3.048 15 0.460 31 0.544 wx w x y " " " " " " " " " " " " " " " " " " " " " " " " " " wx 0.3 146 0.3 346 0.3 346 0.3 215 0.3 653 0.3 738 0.3 459 0.3 335 0.3 575 0.3 522 0.3 446 0.3 103 0.3 222 0.3 145 0.3 276 0.3 508 0.3 150 0.3 177 0.3 201 0.3 253 0.3 251 0.3 150 0.3 204 0.60.60.4 2.182 2.119 1.509 0.3 371 0.3 338 0.5 214 " " " " " " " " " " " " " " " " " " " " " " " " " " " " " wx wx B and were uniformly less than 6.0E 10 10 26 10 26 10 26 Ž. Ž. Ž.Ž. Ž. Ž.Ž . Ž . Ž. Ž. 12 . Aluminum blank ratios were uniformly less than 1.0E y 9.92E 3 .Ž. s ) cm ; we then calculate P z using Eq. 1 and Brown et al.'s 20 values of clasts km cm ppm ppm atom r 0.1 g " Ž. Al data: amalgamated samples 26 Ž Be and Beryllium blank corrections take into account interference by the isobar Production as a function of latitude and altitude is calculated from equations given by Lal 9at and Lawrence calibrations Livermore of Nishiizumi National et al. Laboratory. Aluminum 24 . samples The surface were prod normalized to standards prepared by Nishiizumi, KNSTD9860 26Al:27Al adjusted to account fordensity the depth of to 2.1 the subsurface samples: we estimatesamples the terrace labeled gravels with to a contain 30% large clasts, and are the calculated porosity with Eq. of the 4 matrix , using the mean surface production rate and concentration for that terrace. All samples were measured Table 2 10 3294-1B flood plain 30 38.39234-6A 1.42 FR19234-6B FR1 0 30 30 16.9 38.4 1034175-2A 38.4 FR2C9244-8A 28.0 1.38 FR2C9244-8B 1.38 FR2C 334175-3A 50 FR2C 25 50 38.3 25 7.9 38.4 35 7.9 51 1.41 38.4 1.41 51 38.3 24.9 40 1.41 24.5 60 1.414155-3A 9.5 FR3 604155-3B 7.1 FR3 1004155-4A 60 7.1 FR34155-4B 4.0 46 30 FR3 24.8 46 23 27 25.4 38.4 409234-4A 23.9 38.4 40 24.4 FR4 1.43 38.4 1.43 38.4Altitudes and 100 1.43 45 latitudes for samples 100 1.43 were 4.0 determined using: 150 38.3 handheld 4.0 GPS; laser 150 27 EDM; 1.9propagated triangulation using with 1.49 27 the compass 1.9 uncertainties and given, 25.6 topographic 14 with maps; a and 10% 25.1 200 a uncertainty 14 combinat assigned to 25.3 the surface 1.0 production 25.2 rate and 7% to 7 the sample depth. The model ages giv 26.5 3314-1B flood plain9234-2A 30 FR19234-2B FR19234-2C 38.4 FR1 30 1.36 303314-2B 38.4 32 FR2A3314-3B 0 38.4 FR2B9244-7A 1.38 38.4 FR2C 369244-7B 1.38 16.2 FR2C 31 0 1.38 38.4 30 99 0 38.4 30 0 16.4 25.9 1.39 38.4 16.4 100 1.40 38.43314-4B 16.4 FR3 0 1.41 1009244-9A 25.3 FR3 0 100 1.419244-9B 25.2 FR3 0 16.64155-1A 25.0 32 FR3 04155-1B 16.8 101 37 FR3 16.94155-2A 102 38.4 34 24.9 FR3 16.9 103 38.3 32 25.3 103 1.44 38.3 32 25.2 1.44 38.4 40 25.3 0 1.44 38.49234-3A 0 1.44 38.4 FR49234-3B 0 1.44 17.2 FR4 0 1.44 17.1 1059234-4B 0 45 17.1 FR4 104 24.7 50 17.1 37 104 24.8 38.3 17.1 104 8.3 25.3 38.3 45 104 53.0 1.49 53 54.2 1.49 38.3 0 36.6 0 1.49 17.8 200 17.8 109 1.0 109 25.5 25.4 7 47.7 Sample Terrace No. of Latitude Altitude Depth P Be P Al Be Al Be concentration Al concentration Be model age Al model age KNSTD9919 26Al:27Al J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 71

The profiles and amalgamated sample pairs sug- volume record. The Pinedale moraine and associated gest 10 Be-based age estimates of 60"9 ka for FR2C, terrace sequence includes last glacial maximum dates 102"16 ka for FR3, and 151"24 ka for FR4. ranging from 17 to 21 ka. The Bull Lake glaciation While the uncertainties are large, comparison with includes a wide temporal range of morainal and the global ice volume record depicted by the d 18 O terrace features. The associated WR3 terrace has curvewx 30 supports the idea that the Fremont terraces been dated, using36 Cl, at 126±103 kawx 19 . Results were emplaced in late glacial or glacial maximum of work using 10 Be methods parallel to our own timesŽ. Fig. 7 . The amalgamated surface sample implies an age of roughly 100 ka for this terracewx 25 . resultsŽ. Fig. 5B from the FR2A and FR2B surfaces While the age the FR2C terrace does not correlate imply that they may be roughly coeval with the with any dated surface in the Wind River sequence, FR2C surface. In the absence of samples from depth the Sierran system does record such an event on these two extensive lower surfaces, we cannot ŽYounger Tahoe in the Bloody Canyon moraines: further constrain the time span that these three sur- 40±55 kawx 13. . faces represent. If they are all associated with isotope Knowledge of the ages of these terraces allows us stage 4, then the vertical range of depositional sur- to estimate various geomorphic process rates within faces produced during this glacial stage was roughly the Fremont River system. While the incision rate of 30 m. the river varies with timeŽ as the existence of ter- We note that the 102 ka age for the FR3 terrace races attests. , even the mean incision rate between corresponds well with recently reported ages for the times of terrace abandonment appears to vary widely. WR3 terrace along the Wind River. The Wind River The terrace ages suggest variation by roughly a has 15 well preserved terraceswx 14 and displays a factor of 3, from 0.30 to 0.85 mrka around a mean rich response to small fluctuations in the global ice of 0.6 mrka. We therefore caution against placing much faith in linear interpolation or extrapolation of terrace ages from any pair of known ages, as is commonly done.

The inheritance, Nin , is of the same order for all three terracesŽ 0.42x106 atomrg10 Be for FR2C, 0.48x106 atomrg for FR3, and 0.44x106 atomrg for FR4. . This corresponds to the arrival of clasts at the terrace with mean effective exposure ages of 30±40 ka. The inheritance can be used to constrain hillslope exhumation rates and fluvial transport times. If we s assume rapid fluvial transportŽ. NinN hill , we can solve Eq.Ž. 6 for the mean hillslope erosion rate to yield eÇs30 mrMa. If we assume exhumation to be s rapidŽ i. e., NinN fluvial , then N fluvial corresponds to a mean transport time of 166 kaŽŽ. using Eq. 8 with Hs5 m. . Because inheritance is the sum of these Fig. 7. Age estimates of the three best dated terracesŽ FR2C, FR3, two processes, the exhumation determined here is a and FR4. shown against the last 250 ka of the normalized global minimum and the fluvial transport time is a maxi- d 18 O recordŽ the scale on the left axis is standard deviations about the mean. wx 27 ; oxygen-isotope stages are numbered. The terrace mum. Given the 40±50 km distance from our source agesŽ. gray bars were determined using production rates scaled by area to our sampled terraces transport times of the the Nishiizumi et al.wx 24 calibrations, and post-depositional accu- order 10 5 yr seem unreasonable. The 30 mrMa mulations of cosmogenic radionuclides constrained by fitting the erosion rate, on the other hand, is similar to bedrock concentration profiles shown in Fig. 6. Also shown are the erosion rates measured elsewhere in the arid Ameri- summer and winter insolation calculations of Berger and Loutre wx wx31 for 308N. The terrace ages correspond roughly to global ice can west 12,1 . This argues that the primary source volume maxima in stages 4, 5d and 6, and to maxima in the of inheritance in the Fremont system is obtained summer insolation and minima in the winter insolation histories. during exhumation on hillslopes within the headwa- 72 J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 ters, and that clast-to-clast variability in inheritance porting the use of the pairs technique in this system is due to high variability in local rates of exhuma- for determining both the age and inheritance. We tion. That the mean inheritance is roughly consistent suggest that this is probably not a general result, and from terrace to terrace could be interpreted as signi- that full concentration profiles should be used to test fying that the mean weathering rate within the basin for problems related to post-depositional turbation of over this interval of time has been surprisingly con- clasts and for inflation or deflation of the surface stant. within any system being dated. The inheritance of cosmogenic radionuclides is significant and must be taken into account in estimating ages of depositional 6. Conclusions surfaces. Studies of the rate of pedogenesis, and of the rate of weathering of clasts on these terraces can The large spread of single clast ages implies a now proceed with the knowledge that the timing has wide scatter in the inheritance signal, raising a cau- been established to within roughly 10±20%, most of tionary flag against the use of small numbers of this error being in choice of cosmogenic production clasts in dating depositional terraces. This scatter can rate. be reduced significantly by using samples consisting of aliquots from large numbers of clasts. The repro- ducibility of the concentration values from samples Acknowledgements amalgamated from 25±40 clasts implies that these are sufficient numbers to constrain the mean concen- We gratefully acknowledge the Petroleum Re- trations in this particular geomorphic system. search Fund of the American Chemical Society, a Using the latest Nishiizumi et al. production rates, grant from the Center for Accelerator Mass Spec- we estimate 10 Be ages of roughly 60"9, 102"16 trometry at Lawrence Livermore National Labora- and 151"24 ka for three of the terraces along the tory, and a Cole award from the Geological Society Fremont River. These ages are consistent with a of America, for support of this research. We thank terrace genesis model wherein terrace gravels are the following individuals for their help in the field, laid down in braided outwash plains associated with the lab, andror for reviewing early versions of this high water and sediment flux during glacial maxima, paper: Christian Brauderick, Alex Densmore, Greg and are abandoned as the Fremont River begins to Dick, James Georgis, Joe Koning, Craig Lundstrom, incise due to high water and low sediment flux upon Greg Pratt, Eric Small, and Melissa Swartz. In addi- retreat of the glacierswx 2,6 . Our dating implies that tion, thorough and constructive reviews of an earlier climate swings that generate only small and short- version of the manuscript by K. Nishiizumi, E. Brown lived variations in global ice volume are sufficient to and an anonymous reviewer aided significantly in its generate significant glacial systems in small moun- condensation. This work was partially supported un- tain chains. The incision rate between terrace-for- der the auspices of the DOE by LLNL under contract ming events appears to vary by a factor of three, W-7405-Eng-48. []MK implying that the method of dating strath terraces by fitting them to a single, long-term mean incision rate yields at best a crude approximation. References At this point, one must still exercise considerable wx1 E.E. Small, R.S. Anderson, R.C. Finkel,10 Be and 26 Al geological experience in the interpretation of cosmo- erosion rates from summit flats in the Rocky Mountains: genic radionuclide dates from these depositional ter- Constraints on the rate of relief production, EOS Trans. Am. races. By developing a strategy to account for the Geophys. Union 76Ž. 1995 690. geomorphic delivery system of clasts to the final site wx2 G.M. Richmond, Quaternary stratigraphy of the La Sal of deposition, we have dealt with a large portion of Mountains Utah, U.S. Geol. Surv. Prof. Pap. 324Ž. 1962 135. wx3 O.A. Chadwick, R.D. Hall, J. Conel, F.M. Phillips, M. the noise in the system. The profile technique, em- Zreda, M.D. Gosse, Glacial deposits and river terraces, in: ployed here on two of the terraces, generates the Wind River Basin, Friends of the Pleistocene Rocky Moun- expected inheritance-shifted exponential profile, sup- tain Cell, Field Conference Guidebook, 1994, 141 pp. J.L. Repka et al.rEarth and Planetary Science Letters 152() 1997 59±73 73

wx4 C. Jaworowski, A probable new Lava Creek ash locality: Millard and Wayne Counties, Utah, Utah Geological and implications for Quaternary geologic studies in the western Mineral Survey, Salt Lake City, 1987. Wind River basin, Wyoming, USA, Friends of the Pleis- wx19 F.M. Phillips, M.G. Zreda, J.C. Gosse, J. Klein, E.B. Even- tocene Rocky Mountain Cell, Field Conference Guidebook, son, R.D. Hall, O.A. Chadwick, P. Sharma, Cosmogenic 36 Cl 1994, pp. 135±141. and 10 Be ages of Quaternary Glacial and fluvial deposits of wx5 A.D. Howard, Study of process and history in desert land- the Wind River Range, Wyoming, Geol. Soc. Am. Bull.Ž in forms near the , Utah, Ph.D. Thesis, Johns press. . Hopkins Univ., Baltimore, MD, 1970, 198 pp. wx20 E.T. Brown, E.J. Brook, G.M. Raisbeck, F. Yiou, M.D. wx6 S. Sinnock, Glacial moraines, terraces and pediments of Kurz, Effective attenuation lengths of cosmic rays producing Grand Valley, western Colorado, New Mexico Geol. Soc. 10 Be and 26Al in quartz: Implications for exposure age 32nd Field Conference Guidebook, 1981, pp. 113±120. dating, Geophys. Res. Lett. 19Ž. 1992 369±372. wx7 J.L. Slate, W.B. Bull, T.L. Ku, M. Shafiquillah, D.J. Lynch, wx21 K. Nishiizumi, Cosmogenic production of10 Be and 26 Al on Yi-Pu Huang, Soil-carbonate genesis in the Pinacate Vol- the surface of the Earth and underground, Abstracts 8th Int. canic Field, northwestern Sonora, Mexico, Quat. Res. 35 Conf. on Geochronology, Cosmochronology, and Isotope Ž.1991 400±416. Geology, U.S. Geol. Surv. Circ. 1107Ž. 1994 234. wx8 F. Phillips, M.G. Zreda, S.S. Smith, D. Elmore, P.W. Kubik, wx22 K. Nishiizumi, Cosmic ray production rates of10 Be and 26 Al P. Sharma, Cosmogenic chlorine-36 chronology for glacial in quartz from glacially polished rocks, J. Geophys. Res. 94 deposits at Bloody Canyon, eastern Sierra Nevada, Science Ž.1989 17907±17915. 248Ž.Ž. 4962 1990 1529±1532. wx23 D.H. Clark, P.R. Bierman, P. Larsen, Improving in situ wx9 D. Lal, Cosmic ray labeling of erosion surfaces: in situ cosmogenic chronometers, Quat. Res. 44Ž. 1995 367±377. nuclide production and erosion models, Earth Planet. Sci. wx24 K. Nishiizumi, R.C. Finkel, J. Klein, C.P. Kohl, Cosmogenic Lett. 104Ž. 1991 424±439. production of 7 Be and 10 Be in water targets, J. Geophys. wx10 K. Nishiizumi, C.P. Kohl, J.R. Arnold, R.I. Dorn, J. Klein, Res. 101Ž. 1996 22225±22232. D. Fink, R. Middleton, D. Lal, Role of in situ cosmogenic wx25 G.S. Dick, R.S. Anderson, J.L. Repka, O.A. Chadwick, R.C. nuclides 10 Be and 26Al in the study of diverse geomorphic Finkel, 10 Be and 26Al dating of fluvial terraces, Wind River, processes, Earth Surf. Process. 18Ž. 1993 407±425. Wyoming, EOS Trans. Am. Geophys. Union 76Ž. 1995 690. wx11 P.R. Bierman, Using in situ produced cosmogenic isotopes to wx26 S.G. Wells, L.D. McFadden, J. Poths, C.T. Olinger, Cosmo- estimate rates of landscape evolution: A review from the genic 3 He surface-exposure dating of stone pavements Ð geomorphic perspective, J. Geophys. Res. Ð Solid Earth 99 Implications for landscape evolution in deserts, Geology 23 Ž.1994 13885±13896. Ž.1995 613±617. wx12 P.R. Bierman, How fast do rocks erode? New answers from wx27 T.E. Cerling, H. Craig, Geomorphology and in-situ cosmo- atom counting, Geol. Soc. Am. Abstr. with Programs 27 genic isotopes, Rev. Earth Planet. Sci. 22Ž. 1994 273±317. Ž.1995 44. wx28 C.P. Kohl, K. Nishiizumi, Chemical isolation of quartz for wx13 F.M. Phillips, M.G. Zreda, L.V. Benson, M.A. Plummer, D. measurement of in-situ-produced cosmogenic nuclides, Elmore, P. Sharma, Chronology for fluctuations in late Pleis- Geochim. Cosmochim. Acta 56Ž. 1992 3583±3587. tocene Sierra Nevada glaciers and lakes, Science 274Ž. 1996 wx29 J.C. Davis, I.D. Proctor, J.R. Southon, M.W. Caffee, D.W. 749±751. Heikkinen, M.L. Roberts, K.W. Moore, K.W. Turteltaub, wx14 O.A. Chadwick, R.D. Hall, F.M. Phillips, Pleistocene glacia- D.E. Nelson, D.H. Loyd, J.S. Vogel, LLNLrUC AMS facil- tions in the Rocky Mountains: Bull Lake and Sacagawea ity and research program, Nucl. Instrum. Methods Phys. Res. Ridge revisited, Geol. Soc. Am. Bull.Ž. in press . B 52Ž. 1990 269±272. wx15 R.S. Anderson, J.L. Repka, G.S. Dick, Explicit treatment of wx30 J. Imbrie, J.D. Hays, D.G. Martinson, A. McIntyre, A.C. inheritance in dating depositional surfaces using in situ 10 Be Mix, J.J. Morley, N.G. Pisias, W.L. Prell, N.J. Shackelton, and26 Al, Geology 24Ž. 1996 47±51. The orbital theory of Pleistocene climate: support from a wx16 R.F. Flint, C.S. Denny, Quaternary Geology of Boulder revised chronology of the marine d 18 O record, in: A. Berger, Mountain, Aquarius Plateau, Utah, U.S., Geol. Surv. Bull. J. Imbrie, J. Hays, G. Kukla, B. SaltzmanŽ. Eds. , Mi- 1061-DŽ. 1956 103±164. lankovitch and Climate: Understanding the Response to As- wx17 A.D. Howard, Quaternary landform evolution of the Dirty tronomical Forcing, Part 1, Reidel, Higham, MA, 1984, pp. Devil River system, Utah, Geol. Soc. Am. Abstr. with Pro- 269±305. grams 18Ž. 1986 641. wx31 A. Berger, M.F. Loutre, Insolation values for the climate of wx18 G.H. Billingsly, P.W. Huntoon, W.J. Breed, Geologic map of the last 10 million of years, Quat. Sci. Rev. 10Ž. 1991 Capitol Reef National Park and vicinity, Emery, Garfield, 297±317.