Rapid Glacial Erosion at 1.8 Ma Revealed by 4He/3He Thermochronometry

See allHide authors and affiliations

Science  09 Dec 2005:
Vol. 310, Issue 5754, pp. 1668-1670
DOI: 10.1126/science.1118519


Alpine glaciation and river incision control the topography of mountain ranges, but their relative contributions have been debated for years. Apatite 4He/3He thermochronometry tightly constrains the timing and rate of glacial erosion within one of the largest valleys in the southern Coast Mountains of British Columbia, Canada. Five proximate samples require accelerated denudation of the Klinaklini Valley initiating 1.8 ± 0.2 million years ago (Ma). At least 2 kilometers of overlying rock were removed from the valley at ≥5 millimeters per year, indicating that glacial valley deepening proceeded ≥6 times as fast as erosion rates before ∼1.8 Ma. This intense erosion may be related to a global transition to enhanced climate instability ∼1.9 Ma.

The hypotheses that alpine glaciers erode at higher rates than rivers (1, 2) and control mountain topographic relief (3, 4) can be tested if pre- and syn-glacial erosion rates can be measured. Debate over the relative contributions of fluvial and glacial erosion to the development of relief has persisted in part because glacial erosion rates have proven difficult to measure, and because it is not well known when and how fast relief develops. Recent advances in quantifying glacial erosion processes include studies of sediment accumulation rates (5), suspended sediment loads (6), geochronology (79), and numerical modeling of subglacial processes (1012). These studies have led to an apparent discrepancy between erosion rates determined from sediment yields (5) (on time scales <103 years) and erosion rates inferred from low-temperature cooling ages (8). The latter are sensitive to mean erosion of mountain ranges over much longer time scales (>106 years) and usually indicate slower rates of erosion than those determined from sediment yields. A critical evaluation of the long-term consequences of glaciation on topography has been hampered by the lack of data to constrain erosion rates before, and during, past glaciations. We present an application of 4He/3He thermochronometry (13) to compare long-term erosion rates before and during alpine glaciation in the Klinaklini valley of the Coast Mountains, British Columbia.

The thermal field of the uppermost few kilometers of Earth's crust responds to changes in surface topography (14). Consequently, the history of mountain relief is recorded by the cooling history of rocks as they approach the surface via denudation. (U-Th)/He ages of apatite provide the most sensitive indicator of such near-surface cooling (14). This sensitivity arises from the temperature dependence of helium diffusion: The 4He concentration in a cooling apatite reflects radiogenic production and diffusive loss integrated along the cooling path from ∼80°C, where 4He retention effectively begins, to ∼20°C, where retention is nearly quantitative. The (U-Th)/He age of an entire crystal is usually interpreted as the time elapsed since the apatite crossed the ∼70°C isotherm (15). However, far more detailed information on the cooling path, essential to resolving the role of glacial incision, potentially resides in the spatial distribution of 4He within a grain.

In a sample containing a uniform distribution of proton-induced 3He (16), sequentially measured 4He/3He ratios during stepped-extraction document the spatial distribution of radiogenic 4He (17, 18). When combined with a (U-Th)/He age and helium diffusion kinetics, the 4He distribution quantitatively constrains the sample's cooling path (13, 19). We applied this method in the Coast Mountains, which extend for ∼1000 km along the western margin of North America and have up to 4 km of topographic relief. Previous thermochronology results suggest that mountain building and exhumation >0.5 mm/year began ∼10 Ma in this region (20). Although this mountain range was heavily incised by glaciers, the timing and magnitude of glacial erosion are only poorly known (9, 20, 21). We report 4He/3He thermochronometry results for five apatite samples separated from granitic rocks collected along an approximately vertical transect in the ∼2.5-km-deep, U-shaped Klinaklini valley (Fig. 1). This valley is one of many similarly shaped glacial valleys and fjords draining the western side of the southern Coast Mountains. Samples were collected 10 km downstream from the confluence of the present-day Heiltskuk Icefield and Klinaklini glacier (∼800 km2) and a former alpine ice sheet of equal or greater size present in the main Klinaklini river valley during the last glacial maximum (22) (Fig. 1). From our observations of glacial trim lines, we estimate maximum ice thicknesses of ∼2.5 km at this location.

Fig. 1.

Topography and sample location maps. (A) Regional physiography. Black box represents area shown in (B). (B) Shaded relief topography, vegetation cover (green), present-day glacial extent (white to light gray), sample locations (red), and topographic profile location across the Klinaklini valley (A-A′). (C) Topographic profile and sample elevations projected onto profile A-A′.

The (U-Th)/He ages of the five samples range from 1.7 to 6.4 Ma. Sample 01MR-59, collected near the Klinaklini valley floor, has the youngest (U-Th)/He age (1.7 ± 0.1 Ma), so its 4He distribution constrains recent thermal perturbations with the highest resolution. Figure 2 is an example of the 4He/3He thermochronometry results for this sample. Details of the other four samples are presented in (18) and discussed later. Collectively, these results quantify the helium diffusion kinetics and reveal the spatial distribution of 4He within each sample.

Fig. 2.

4He/3He thermochronometry for apatite sample 01MR-59. (A) 3He diffusion Arrhenius plot; circles are the diffusion coefficients, D, normalized to the diffusive length scale, a, calculated (38) from release fractions of proton-induced 3He (16). Solid black line is the inferred helium diffusion kinetics for 01MR-59 used to construct models shown in (B) and (C), determined by linear regression to a subset array (includes the 175° to 295°C steps; n = 22). Steps deviate from linearity above 300°C because of a change in diffusion mechanism (36), which may not apply to lower temperatures relevant to our models and are therefore excluded from the regression. (B) Model cooling paths. Each of these models yields the observed (U-Th)/He age for 01MR-59 (1.7 ± 0.1 Ma). Dashed lines are unconstrained by this sample. (C) Ratio evolution diagram. Shown are measured isotope ratios for each release step, Rstep(R = 4He/3He), normalized to the bulk ratio Rbulk, plotted versus the cumulative 3He-release fraction, ΣF3He. Three models are shown that correspond to the cooling paths in (B). Error bars (1σ) are specified by a vertical line through each point. The best-fitting model is shown in red; the blue curve estimates the minimum cooling rate allowed by the 4He/3He data and provides a confidence limit on the rapid cooling event at 1.8 Ma.

Using the diffusion kinetics quantified from Fig. 2A (23), Fig. 2B shows three model cooling paths that produce the (U-Th)/He age observed for sample 01MR-59, yet different spatial distributions of 4He (Fig. 2C). Ejection of α particles from the outer ∼20 μm of a mineral confers a predictable shape to the radiogenic 4He distribution, which is independent of diffusion (13, 24). The 4He distribution inferred from the data shown in Fig. 2C is almost entirely explained by α-ejection, contains little diffusive signal (red curve), and requires rapid cooling. In contrast, the 4He/3He measurements clearly preclude thermal histories involving monotonic cooling below 80°C throughout the Pleistocene (green curve).

Of the models in Fig. 2, the best fit (red curve) implies that rapid cooling began ∼1.8 Ma. Because the 4He/3He data are consistent with an end-member 4He distribution (i.e., no diffusion), they place a stringent restriction on the possible cooling path of the sample. To illustrate this limit, and to estimate a confidence interval for the solution, we sought a lower bound on the cooling rate the sample could have experienced. The blue curve in Fig. 2C shows a lower limit at which point the model is clearly no longer in agreement with the observations, requiring cooling from at least ∼70°C at 1.8 Ma to below 20°C by ∼1.4 Ma. Therefore, for 01MR-59, the red and blue curves in Fig. 2 define a confidence interval on the cooling path; both demand very rapid cooling denudation beginning 1.8 ± 0.2 Ma.

Widespread gradual climate change to ice-house conditions initiated ∼2.7 Ma (25), whereas between 1.9 and 1.7 Ma clear evidence exists for a global transition to increased climate variability (26), major changes in the provenance of Northern Hemisphere erosion products (27, 28), and the establishment of coldwater upwelling conditions in the tropical and subtropical eastern Pacific ocean (26). These global trends in climate are consistent with intense erosion ∼1.8 Ma, a time when large ice sheets with a high potential for erosion were likely pervasive across the Coast Mountains and major increases in ice-rafted debris are observed in sea-sediment records (29). Thus, glacial erosion within the valley is the most plausible way to explain such rapid cooling of the rocks. Measured geothermal gradients in the region range between ∼20° and 30°C/km (30, 31). Assuming these gradients are representative of the thermal field at the time sample 01MR-59 cooled below 80°C, our results imply that ∼1700 to 2200 m of overlying rock was rapidly eroded. Adopting the blue curve in Fig. 2B as a lower bound on the cooling rate of 01MR-59, the implied erosion rate was ≥5 mm/year between 1.8 and 1.4 Ma. These results also imply ≤300 m of erosion within the valley since 1.4 Ma.

An erosion event of this magnitude must have affected samples in a broader region, although the magnitude of erosion could be spatially variable. Application of 4He/3He thermochronometry on multiple, neighboring samples provides a test for internal consistency and a way to bootstrap solutions and develop a three-dimensional denudation record through time. For example, assuming that thermal gradients between samples were always ≤30°C/km, a best-fit cooling path for one sample restricts the acceptable cooling path of a second sample located 500 m (vertically) to be offset by no more than 15°C at all times. We tested the internal consistency of our results using a vertical profile of samples collected ∼6 km from sample 01MR-59 (Fig. 1).

Figure 3 shows a set of cooling paths for four apatite samples collected from the western wall of the valley (Fig. 1) along with the solution for 01MR-59. All of the 4He/3He thermochronometry results (18) support the cooling history inferred from 01MR-59 and suggest that samples collected deep within the Klinaklini valley (at present elevations ≤2000 m) experienced rapid cooling, starting ∼1.8 Ma. The four samples from the western wall require an earlier period of slower cooling (∼20°C/106 years), between 5 and 2 Ma, and the two highest samples suggest even slower cooling before 5 Ma. As expected, the data require that samples at higher elevations were at lower temperatures, and therefore closer to the paleo-surface, as the Klinaklini valley deepened.

Fig. 3.

Internally consistent set of cooling paths for a near-vertical array of samples collected from the west wall of Klinaklini valley. Listed in the legend are the name, (U-Th)/He age, and elevation of each sample, respectively. The corresponding 4He/3He thermochronometry results are shown in the supporting online material. Confidence intervals for each cooling path are approximately the same as that indicated by the difference between the red and blue curves in Fig. 2.

For instance, the data require that by 1.8 Ma, sample TEKI-23 was already at temperatures <20°C, and so was located very near the paleo-surface. Therefore, at that location, much less erosion occurred since 1.8 Ma than in the bottom of the valley. The present elevation of TEKI-23 is ∼2100 m above sea level and ∼1900 m above sample 01MR-59. The difference in elevation between TEKI-23 and 01MR-59 is in excellent agreement with the minimum amount of erosion inferred solely from 01MR-59. Intense glacial erosion ∼1.8 Ma greatly deepened the Klinaklini valley and enhanced topographic relief between the locations of these two samples, although the region's relief before 1.8 Ma cannot be quantified solely from this data set. Further, because TEKI-38 is located at lower elevation than 01MR-59 yet was at lower temperatures, the comparative 4He/3He thermochronometry of these samples implies that glacial valley widening progressed toward the east.

Our results indicate acceleration in rock cooling and erosion rates at a time when widespread glaciation was active in North America, but nearly 900,000 years after the onset of Northern Hemisphere glaciation (25). We find that between 1.8 and 1.4 Ma, glacial erosion rates in the Klinaklini valley were significantly higher (by a factor of ≥6) than erosion rates before 1.8 Ma (between ∼5 and 1.8 Ma), which more likely involved fluvial processes. The coincidence of this event with other erosion and climate events occurring in both the Northern and Southern hemispheres between 1.9 and 1.7 Ma (2629, 32, 33) suggests that the intense erosion may be related to a global transition to increased climate variability at that time (26, 33, 34). This supports the notion that transitions out of periods of climate stability to high-frequency changes in temperature and precipitation enhance erosion by preventing fluvial and glacial systems from establishing new equilibrium states (34).

Although glacier ice likely existed within major valleys during more recent glaciations (i.e., after 1.4 Ma), 4He/3He thermochronometry indicates that most of the present relief of the Klinaklini valley developed shortly after continental ice sheets did, and that recent glacial advances resulted in considerably less net erosion within preexisting valleys. Thermochronometric evidence from the northern Coast Mountains revealed an average erosion rate of only 0.22 mm/year between 10 and 4 Ma and suggested a substantial increase in exhumation rate sometime after 4 Ma (9). If representative of the entire range, the Klinaklini data set suggests that the Coast Mountains experienced a major topographic modification after 2 Ma.

Supporting Online Material

SOM Text

Figs. S1 to S5

Tables S1 to S5

References and Notes

References and Notes

Stay Connected to Science

Navigate This Article