Using 13 samples collected from a 4.1 meter profile in a well-dated and stable New Zealand fluvial terrace, we present the first long-term accumulation rate for meteoric 10Be in soil (1.68 to 1.72 × 106 at/(cm2·yr)) integrated over the past ∼18 ka. Site-specific accumulation data, such as these, are prerequisite to the application of meteoric 10Be in surface process studies. Our data begin the process of calibrating long-term meteoric 10Be delivery rates across latitude and precipitation gradients. Our integrated rate is lower than contemporary meteoric 10Be fluxes measured in New Zealand rainfall, suggesting that long-term average precipitation, dust flux, or both, at this site were less than modern values. With accurately calibrated long-term delivery rates, such as this, meteoric 10Be will be a powerful tool for studying rates of landscape change in environments where other cosmogenic nuclides, such as in situ 10Be, cannot be used.
 Because of documented long-term changes in primary meteoric 10Be production [Frank et al., 1997], climate (primarily precipitation) [Dore, 2005], and the source and volume of allochthonous dust [Baumgartner et al., 1997], there are differences between long- and short-term meteoric 10Be delivery rates. These complexities suggest the importance of calibrating site-specific, long-term delivery rates by measuring the accumulation of meteoric 10Be in geologic archives. Such work has been done in lake deposits, deep-sea sediments, and glacial ice [e.g., Finkel and Nishiizumi, 1997; Frank et al., 1997] but not in soils, the basis for most geomorphic studies. Here, we quantify the meteoric 10Be inventory in a 4.1 m depth profile collected from a stable and well-dated alluvial surface on New Zealand's North Island and estimate a long-term accumulation rate for meteoric 10Be in soil.
2. Behavior of Meteoric 10Be
 Meteoric 10Be is a valuable tool for studying surface process rates because, once deposited, it adsorbs tenaciously to near-surface materials in all but the most acidic soils [You et al., 1989]. Unlike shorter-lived radionuclides, such as 210Pb and 137Cs [e.g., Walling et al., 2003], the longer half-life of 10Be (1.36 Myr [Nishiizumi et al., 2007]) increases the period of time over which the nuclide accumulates in soils and penetrates to depth before decay, thus extending the timeframe over which the method is applicable. Because measurements of meteoric 10Be are made on bulk samples, the presence or absence of a specific mineral phase is irrelevant, making the isotope useful across a wide variety of landscapes.
 The flux of meteoric 10Be to terrestrial environments comes from two sources: 10Be produced in the atmosphere by spallation of nitrogen and oxygen and delivered to earth's surface by precipitation and dryfall (primary component), and 10Be adhered to airborne dust (recycled component) [Monaghan et al., 1986].
 Primary production of meteoric 10Be is controlled by solar activity and magnetic field intensity [Masarik and Beer, 2009], both of which vary over time [Beer, 1994; Frank et al., 1997]. The subsequent distribution of primary meteoric 10Be is controlled by atmospheric circulation, with annual precipitation being a strong predictor of total meteoric 10Be fallout at any one location [Heikkilä et al., 2009].
 Because geochemical processes in soils rapidly meld primary and recycled meteoric 10Be, constraining the spatial and temporal variation in the rate of accumulation of both components is required when measurements of meteoric 10Be are used for modeling surface processes. Most contemporary 10Be flux measurements exclude dust influence to determine the primary 10Be flux. In this study, both components are critical and not explicitly separable.
3. Geologic Setting
 We sampled a soil profile within the Waipaoa River Basin, a 2,200 km2 catchment draining the eastern margin of New Zealand's North Island (Figure 1) [Mazengarb and Speden, 2000]. At ∼38°S Latitude, this site receives ∼110 cm of rain annually [Hessell, 1980].
 Within the basin, an extensive flat-lying fluvial terrace (termed Waipaoa-1) stands up to ∼100 m above the mainstem and many of the tributary channels of the Waipaoa River. This terrace surface is capped by ∼10 m of coarse fluvial gravel deposited during the last glacial maximum [Berryman et al., 2000]. Atop the gravel, lie several meters of overbank silty clay-rich flood deposits laid down as this river level was rapidly abandoned in response to a combination of tectonic uplift and a switch in the fluvial system from aggradation to rapid incision, most likely in response to changing climate following the glacial maximum at ∼18 ka [Berryman et al., 2000; Eden et al., 2001]. Evidence from other dated terrace surfaces suggests that the cessation of aggradation at ∼18 ka was a regional event across the eastern and southern North Island [Eden et al., 2001]. Where we sampled the Waipaoa-1 terrace, it stands ∼50 m above the modern channel, is extensive, flat, far from any nearby slopes, well-preserved, and lacks any surface drainage, indicating that little net erosion or deposition have occurred since the emplacement of the overbank deposits shortly after ∼18 ka. Land clearance and agriculture have at most reworked the upper several dm of the sampled site.
4. Age of Sampled Profile
 The overbank deposits contain age-constrained tephras used to estimate the timing of the Waipaoa-1 terrace abandonment and emplacement of the sediment we sampled. The Rerewhakaaitu Tephra is located at or near the base of the Waipaoa-1 overbank deposits that cap the fluvial gravels [Berryman et al., 2000; Eden et al., 2001; Froggatt and Lowe, 1990]. The stratigraphic position of this tephra indicates that it fell coincidently with the initiation of rapid incision [Berryman et al., 2000; Eden et al., 2001]. The overlying flood deposits were emplaced relatively quickly (perhaps over the course of decades; [Eden et al., 2001]) until the river had incised far enough to isolate the terrace surface from further aggradation. The age of the Rerewhakaaitu Tephra is constrained with multiple radiocarbon ages (n = 4) of organic material directly overlying the tephra in a bog core collected nearby [Lowe et al., 1999]. We calibrated the radiocarbon age of 14,700 ± 95 14C yrs with CALIB REV6.0 [Stuiver and Reimer, 1993], yielding a 1σ age range of 17,659 to 18,030 cal. yr.
 The Waipaoa-1 terrace is ideal for constraining the long-term delivery rate of meteoric 10Be because: 1) the airfall deposition of the Rerewhakaaitu Tephra within the overbank deposits constrains the integration time of 10Be accumulation, 2) an intact younger capping tephra bed argues against either surface erosion or deposition, 3) the fine texture of the soil and the buffering capacity of the carbonate-bearing source rocks [Black, 1980; Mazengarb and Speden, 2000] ensure retention of meteoric 10Be and, 4) the ∼5 m of overbank deposits above the basal tephra at the location we sampled is thick enough to retain the inventory of meteoric 10Be delivered since 18 ka.
5. Sampling and Analysis Techniques
 We sampled the Waipaoa-1 overbank sequence from a recent excavation at 2931760 E, 6297492 N (NZ Grid 1949 (Figure 1)). The sequence consists of fluvial silty clay-rich sediment containing small amounts of reworked tephra. The overbank sediment is capped by a discrete younger tephra bed (presumably the widespread ∼3500 cal. ybp Waimihia Tephra) [Eden et al., 2001], the upper ∼15 cm of which has developed an organic-rich A/O-horizon. We collected a total of thirteen, 15 to 37 cm thick amalgamated samples. In addition, we collected several undisturbed samples of profile sediment for dry density determination.
 We dried and milled samples and isolated meteoric 10Be from ∼0.5 g aliquots using a modification of the method of Stone , then calculated meteoric 10Be concentrations from 10Be/9Be ratios measured at Lawrence Livermore National Laboratory. Data were normalized to the 07KNSTD3110 standard with an assumed ratio of 2850·10−15 [Nishiizumi et al., 2007]. All measured sample isotopic ratios were corrected using process blanks prepared from acid-leached fluvial sediment collected in the Waipaoa Basin; blank corrections ranged from 2.1 to 0.3% of measured ratios.
6. Long-Term Meteoric 10Be Delivery Rate
 In general, meteoric 10Be concentrations decrease regularly down section (Figure 2 and Table S1 of the auxiliary material), with a maximum concentration of 16.27 ± 0.40 × 107 atoms/g in the uppermost sample, and a minimum concentration of 3.12 ± 0.07 × 107 atoms/g near the bottom of the profile. When deposited, the overbank sediment carried some meteoric 10Be, its inherited concentration. Following the abandonment of the Waipaoa-1 terrace and the emplacement of the overbank sequence, additional atmospherically-derived meteoric 10Be accumulated, adsorbed to fine sediment, was bioturbated, and translocated downward through macropores, resulting in the profile shape we see today (Figure 2). We consider the relatively uniform and low concentration of meteoric 10Be in the bottom ∼0.6 m of the profile (samples WA102l and n (Figure 2)) as representative of the inherited component of the total inventory of meteoric 10Be in the profile, and subtract the thickness-weighted average concentration of these two samples from all others, except WA102a and b. Because these two uppermost samples were sourced primarily from airfall tephra, we assume they contained no meteoric 10Be when deposited.
 We use equation (1) to calculate a total inventory of meteoric 10Be (N; 3.02 ± 0.05 × 1010 atoms/cm2) deposited and adsorbed since the abandonment of the Waipaoa-1 terrace.
where, ntot = the measured concentration of meteoric 10Be (atoms/g), ninh = the inherited component of the total concentration (3.21 ± 0.06 ·107 atoms/g), ρ = the dry density of the depth increment (g/cm3), and l = the increment thickness (cm). The dry density of the overbank silt and clay (WA102c to n) is 1.68 ± 0.03 g/cm3 based on repeat measurements (n = 4) of undisturbed samples we collected. We use a literature value for the dry density of tephra (1.05 ± 0.12 g/cm3 [Houlbrooke et al., 1997]) for the uppermost tephritic increments (WA102a and b).
 We arrive at a geologic delivery rate (q; atoms/(cm2·yr)), corrected for decay and inheritance, for the meteoric 10Be accumulated within the measured profile (N; atoms/cm2) over the duration of time since the abandonment of the Waipaoa-1 surface (t; yrs) and emplacement of the overbank sediment with equation (2):
We assume λ = 5.1·10−7 yr−1, the decay constant for 10Be [Nishiizumi et al., 2007]. The calibrated 1σ age range of 17,659 to 18,030 cal. yrs translates into a 1σ range of decay-corrected deposition rates for meteoric 10Be of 1.72 to 1.68 × 106 atoms/(cm2·yr).
 Our analysis accounts for all errors associated with AMS measurement, radiocarbon measurement and calibration, and density; however, several possible sources of error are difficult to quantify. If the overbank deposits we sampled were emplaced after the age-constraining basal tephra, the integration time of ∼18 ka would be an overestimate. If surface erosion over the last 18 ky removed material, the measured 10Be inventory would be an underestimate. If the radiocarbon age of the basal tephra is younger than the deposit, the period of accumulation we use would be too short.
 Using precise AMS measurements (<2%, 1σ) of a deep soil profile from a stable depositional surface of constrained age, we provide the first explicit long-term, soil-based calibration of meteoric 10Be deposition integrated over a geologically relevant time interval. The soil we sampled (Figure 2 and Table S1) contains meteoric 10Be derived from three distinct sources: 1) meteoric 10Be inherited prior to the emplacement of the overbank deposits, 2) atmospherically-derived primary meteoric 10Be, and 3) dust-derived recycled 10Be. Our approach quantifies and subtracts the inherited component from the total inventory (N; equation (1)) allowing us to estimate the temporally averaged meteoric 10Be delivery rate (q; equation (2)) since the exposure we sampled was emplaced. The delivery rate we calculate reflects contributions of both primary and recycled meteoric 10Be.
 Contemporary data suggest that meteoric 10Be deposition rates in New Zealand correlate well with precipitation (Figure 2, inset) and that the majority of meteoric 10Be accumulated in the profile we measured is atmospherically-derived (primary). Measurements of meteoric 10Be in modern precipitation collected over two years at four sites spanning New Zealand show a range in deposition rates from 1.7 to 5.2 × 106 atoms/(cm2·yr), with total flux strongly correlating to annual precipitation [Graham et al., 2003]. When these values are normalized to mean annual rainfall at each site and 700 MV of solar activity [Masarik and Beer, 2009; Usoskin et al., 2005], the between-site variability collapses to 1.4 to 2.1 × 104 atoms/cm3 of rainfall. Based on 7Be and dust concentration measurements, Graham et al.  estimate that only about 10% of the contemporary meteoric 10Be fallout is recycled from dust. If the atmospherically-produced primary component is considered separately, modern meteoric 10Be deposition rates (Figure 2, inset) in New Zealand range from ∼1.4 to ∼4.2 × 106 atoms/(cm2·yr).
 If these modern 10Be deposition values represent long-term conditions, and long-term dust flux remained ∼10% of the total meteoric 10Be deposition, then our measured long-term total meteoric 10Be deposition rate of ∼1.70 × 106 atoms/(cm2·yr) suggests that precipitation at the Waipaoa site averaged ∼77 cm/yr. This estimate is ∼30% lower than contemporary measurements [Hessell, 1980], suggesting that precipitation averaged over ∼18 ky was lower than today. Alternatively, some of the difference may be due to a recent increase in meteoric 10Be recycled from dust. Contemporary dust is primarily generated by human activities. If the long-term dust flux on the largely unglaciated North Island is negligible and meteoric 10Be concentrations in contemporary rainfall are otherwise representative of long-term conditions, paleo-precipitation would be ∼91 cm/year over 18 ky, still about 17% drier than modern climate records indicate. Regional paleoclimate records are consistent with this interpretation of the meteoric 10Be data, as they suggest that the eastern North Island was substantially drier prior to an ENSO-driven precipitation increase approximately 4 ka [Gomez et al., 2004].
 Our findings demonstrate the feasibility of calibrating long-term meteoric 10Be accumulation rates using deep, stable, well-dated soil profiles. Such soil-based calibrations are important because soils constitute the source material for most surface process studies including fluvial sediment analysis [e.g., Reusser and Bierman, 2010]. Terrestrial calibration of meteoric 10Be delivery rates compliments other methods. Polar ice cores reliably record 10Be fluxes over time at high latitudes [e.g., Finkel and Nishiizumi, 1997]; however, these fluxes can differ dramatically from those at lower latitudes because of atmospheric production and mixing processes [e.g., Heikkilä et al., 2009]. Deep-sea and most lake sediment records are filtered by drainage basin and biologic processes making delivery rates over time difficult to deconvolve accurately [e.g., Aldahan et al., 1999]. Because deposition rates of meteoric 10Be to the soil surface change over time and space as rainfall, dust flux, and geomagnetic shielding all vary, performing additional geologic calibrations at a variety of latitudes, in different precipitation regimes, and over different integration times will improve the accuracy and precision of surface process studies using this isotope system.
 We thank B. Gomez and M. Marden for introducing us to the Waipaoa, T. Brown for AMS assistance, and G. Balco and A. Heimsath for helpful reviews. Funded by NSF BCS-0317530 and NSF ARC-0713956.