1. Introduction
With mean summit elevations in excess of 4000 m (Garreaud 2009), the tropical Andes form a significant barrier to atmospheric flow and play a large role in modulating the weather and climate of western South America (Insel et al. 2010). In southern Peru and northern Bolivia (12°–18°S) the Andes reach their widest point, the Altiplano: a high plateau with a northwest–southeast orientation, bordered by cordilleras on both the eastern and western flanks. This study focuses on two heavily glacierized ranges on the northeastern margin of the Altiplano, the Cordillera Vilcanota and Cordillera Real (Fig. 1). The Cordillera Vilcanota is home to the Quelccaya Ice Cap, the world’s largest tropical glacier and the site of one of the most important ice core records in the tropical Andes. Ice cores extracted from Quelccaya in 1983 and again in 2003 are annually resolvable for up to 1500 years and have been used to reconstruct past changes in Pacific sea surface temperatures, migration of the intertropical convergence zone, and conditions during the Little Ice Age (Thompson et al. 2013). In the Cordillera Real, an ice core extracted from Nevado Illimani in 1999 dates back 18 000 years and contains information about tropical Andean climate changes that occurred during the transition from the Last Glacial Stage to the Holocene (Ramirez et al. 2003). Long-term climate records derived from ice cores such as these can improve understanding of climate dynamics and enable the evaluation of possible hydroclimatic responses to past and modern-day climate forcings. In the tropical Andes, these records are particularly valuable because of their potential to record the El Niño–Southern Oscillation (ENSO)—the leading mode of South American climate variability—and because they provide historical climate data in an otherwise data-sparse region (Vimeux et al. 2009). However, significant uncertainties remain in the interpretation of the climate signals recorded in these ice cores (Vimeux et al. 2009), and this data archive is now in danger of being lost forever due to rapid ablation (Rabatel et al. 2013; Vuille et al. 2018; Thompson et al. 2017).
Study area. (left) The locations of the Cordilleras Vilcanota and Real, cities (white stars), and mountains (blue triangles) discussed in this study. (right) Key features of South American circulation discussed in this study. SACZ = South Atlantic convergence zone, SALLJ = South American low-level jet.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
The key uncertainty addressed in the present study relates to the interpretation of stable water isotopes preserved in tropical Andean ice cores. Stable water isotopes in ice cores are applied in paleoclimate reconstructions based on the assumption that they are closely related to the isotopic composition of precipitation at the time, after taking into account possible postdepositional modification. In polar regions, a robust relationship between surface temperature and the 18O concentration in precipitation allows for reconstructions of historical temperature. Thompson (2000) suggested that a similar relationship might be used to make temperature reconstructions from tropical ice cores. However, in tropical regions, different meteorological regimes and moisture transport processes complicate this relationship, and it is not supported by observations. In fact, observations in the tropics suggest a stronger relationship between the isotopic composition of precipitation and precipitation amount, known as the “amount effect” (Dansgaard 1964; Rozanski et al. 1993; Vimeux et al. 2005). Nonetheless, the isotope record from Quelccaya has been incorporated into multiproxy global temperature reconstructions (Mann and Jones 2003). Hoffmann et al. (2003) show that coherent isotopic profiles from four Andean glaciers, including Quelccaya, appear to relate to precipitation variability over the Amazon basin, however, understanding of the dynamics of this relationship remains limited because of the complexity of the hydrological cycle in the Andes (Vimeux et al. 2009). To fully understand the hydroclimate signal preserved by isotopic tracers in ice cores, a better understanding of modern-day, location-specific controls on the isotopic composition of precipitation, and how this signal is retained in snow and ice, is needed.
In addition, with recent advances in ice core sampling technology, there are emerging opportunities to analyze new cores at an unprecedented resolution. For example, laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) enables sampling of major and trace elements at micrometer resolution (e.g., Mayewski et al. 2014; Sneed et al. 2015), and a novel sampling technique allows millimeter-resolution slices of ice to be analyzed for isotopic composition (M. Potocki 2018, personal communication). These innovations create the potential to generate paleoclimate reconstructions of subseasonal climate variability from tropical Andean ice cores, allowing the assessment of the timing and precursors to abrupt climate changes, and for the reconstruction of leading modes of subseasonal climate variability at both continental scale (such as the South American seesaw; González and Vera 2009), and global scale (such as the Madden–Julian oscillation; Madden and Julian 1971). There are ongoing projects to extract new cores from both of the cordilleras discussed in this study that will be sampled using this new technology. To be able to interpret these data, it is essential to understand the controls on modern-day subseasonal spatiotemporal variability of precipitation isotopes, and whether or not these signals are preserved in the ice.
In response to these current limitations and opportunities in the analysis of tropical Andean ice cores, the goals of this study are to 1) to identify subseasonal signals of isotopes in precipitation on the northeastern flank of the Altiplano, and their spatiotemporal characteristics, 2) to determine the dominant meteorological controls on these signals, and 3) to investigate how they are retained in the annual-layer snowpack on tropical Andean glaciers. This work will be applicable in the analysis and interpretation of subseasonal isotope signals from new and existing ice cores in this region, and in isotope hydrology studies.
2. Background
a. Overview of isotope systematics
b. Weather and climate of the northeastern Altiplano
To understand the controls on the isotopic composition of precipitation in the Cordillera Vilcanota and Cordillera Real it is therefore essential to understand the local and regional meteorology. These ranges experience distinct wet (November–April) and dry (May–October) seasons. Water vapor that precipitates over the ranges primarily originates over the Atlantic Ocean and undergoes recycling due to convective precipitation and evapotranspiration over the Amazon (Grootes et al. 1989). Precipitation is typically associated with upstream air trajectories from the north-northwest and weak (approximately 1 m s−1) low-level moisture advection (Vimeux et al. 2005; Insel et al. 2013; Perry et al. 2014). These trajectories are broadly parallel with the South American low-level jet (SALLJ; Fig. 1), a southeast-flowing low-level air current that reaches maximum velocity close to the 850-hPa level east of the Andes in Bolivia (Vera et al. 2006). An active SALLJ enhances moist advection up the east Andean slope, particularly along northeast–southwest–oriented valleys (Junquas et al. 2018), to the Altiplano where moisture influx from the Amazon basin is the limiting factor to precipitation (Garreaud 1999; Garreaud et al. 2003). During the wet season, precipitation is episodic, with alternating periods of wet and dry conditions lasting 5–20 days (Lenters and Cook 1997; Garreaud 1999, 2000). Despite heterogeneous terrain, these episodes are regionally coherent (Garreaud 2000; Perry et al. 2014; Hurley et al. 2015) and appear to relate to a number of different forcing mechanisms including propagating extratropical cyclones, changes in upper-level zonal flow, and cold air incursions (CAIs) (Lenters and Cook 1999). On a diurnal scale, precipitation during the wet season is typically bimodal, featuring precipitation peaks around local midnight and in the afternoon around 1600 LST (Perry et al. 2014, 2017; Chavez and Takahashi 2017; Junquas et al. 2018). The late afternoon events are convective in nature and result from instability due to daytime heating of the lower atmosphere. The meteorological forcing associated with the nocturnal peak is an area of current research, but observations suggest that these events are regionally coherent and stratiform in structure (Perry et al. 2017; Endries et al. 2018), indicating a greater degree of organization.
c. Current understanding of the controls on isotopes in tropical Andean precipitation
On both interannual and seasonal time scales, δ18O in tropical precipitation is observed to correlate with changes in precipitation amount, referred to as the “amount effect” (Dansgaard 1964; Rozanski et al. 1993). However, the nature of this amount effect and its physical mechanism remains a topic of ongoing research. One interpretation of the physical mechanism for the amount effect is that the successive isotopic depletion of precipitation results from a feedback mechanism, whereby the initial isotopic composition of the vapor depends on the isotopic composition of precipitation from the previous storm, reintroduced into the atmosphere by re-evaporation (Risi et al. 2008; Field et al. 2010). More frequent and more intense convective events will thus cause progressive depletion of low-level moisture. Another interpretation is that this effect relates to the degree of organization of storms (Kurita 2013). In the tropical Andes, several studies find the relationship between δ18O and local precipitation amount to be weak (Vimeux et al. 2005; Insel et al. 2013; Fiorella et al. 2015), with some studies finding more robust relationships with upwind precipitation amounts (e.g., Vuille and Werner 2005; Vimeux et al. 2005; Fiorella et al. 2015). Insel et al. (2013) related central Andean δ18O variability to the strength of the SALLJ, which plays an active role in transporting moisture to the study area (Junquas et al. 2018). Although there is clearly a relationship between isotopic composition and precipitation amount, a limitation shared by each of the tropical Andean isotope studies discussed above is that they use monthly precipitation samples that are insufficient to capture isotopic variability associated with the episodic 5–20-day wet and dry periods. In contrast, Vimeux et al. (2011) use one year of event-based precipitation samples from the Zongo valley in the Cordillera Real and identified subseasonal δD variability that is consistent between observation sites despite large differences in elevation and local precipitation amounts. These variations appear to coincide with variations in the position of the South Atlantic convergence zone (SACZ; a band of convection in eastern Brazil associated with the South American summer monsoon; Fig. 1), supporting the findings of an earlier modeling study (Sturm et al. 2007). However, because these data come from a single valley, this result cannot be generalized for the broader region. The present study builds on the findings of Vimeux et al. (2011) by using daily precipitation samples collected from a broader area along the northeastern edge of the Altiplano over a multiyear period.
3. Data and methods
a. Precipitation isotope data
Citizen scientist observers, recruited for the project in the Cordillera Vilcanota and Cordillera Real study regions (Fig. 2), measured daily precipitation amount and collected samples for isotopic analysis. Sampling began at two sites in 2013, with others added in 2016 and 2017 (see table inset in Fig. 2). Each site was equipped with a 10-cm diameter measuring cylinder rain gauge with a funnel to limit evaporation. Trained observers took daily liquid-equivalent precipitation measurements at 0700 LST (1200 UTC in Peru, 1100 UTC in Bolivia) using established protocols (Cifelli et al. 2005), and when there was more than 0.5 mm of precipitation, a sample was directly transferred and sealed in an 8-mm glass vial for isotopic analysis. In the event of snow, the funnel was removed from the gauge and the snow was melted in a basin of warm water before sampling. Samples were retrieved annually, and the University of Arkansas Stable Isotope Laboratory (UASIL) analyzed them for isotopic composition (δD, δ18O) using a thermal conversion elemental analyzer (TC/EA) coupled to a continuous flow isotope ratio mass spectrometer (CF-IRMS; see Gehre et al. 2004). The data were normalized using USGS 48, UASIL L, and UASIL R standards (Nelson 2000) and the approximate uncertainty in laboratory results is ±1‰ for δD and ±0.5‰ for δ18O. This complete dataset is available online (Appalachian State University 2018). The relationship between δD and δ18O is described by the deuterium excess parameter (d-excess = δD − 8δ18O; Dansgaard 1964). The lower diffusivity of 1H218O compared to 2H216O means that δ18O is relatively more sensitive to kinetic effects than δD, resulting in changes in d-excess. D-excess can therefore be used to detect, and to some extent diagnose, nonequilibrium processes and help to delineate the factors that contribute to differences in the isotopic composition of precipitation. Small or negative d-excess can result from below-cloud evaporation, and higher d-excess values can result from nonequilibrium condensation during the growth of ice crystals (Jouzel and Merlivat 1984) or continental moisture recycling (Gat and Matsui 1991). The combined measurement uncertainties in δD and δ18O (above) result in a measurement uncertainty of ±4.1‰ for d-excess. Tests performed during fieldwork indicated that samples could become enriched in heavy isotopologues after small afternoon precipitation events, due to evaporation in a rain gauge prior to collection. Precipitation samples potentially modified in this manner are identifiable by low (<0‰) d-excess values, and were excluded from analysis in the present study.
Locations of precipitation collection sites. The inset lists station names, elevations (m MSL), and the start date of each record.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
b. Isotope dataset processing
To assess the spatial variability of δ18O, each site is compared pairwise by correlating δ18O measurements on common days with measured precipitation (Table 1). To assess temporal variability in δ18O a 15-day moving average of δ18O is calculated. A 15-day window is chosen because it is short enough to capture subseasonal fluctuations in δ18O but long enough to fill the majority of data gaps associated with zero-precipitation dates. The seasonal depletion cycle (90-day moving average) is subtracted from the 15-day moving average each year to identify subseasonal δ18O anomalies. We define positive (negative) δ18O anomalies as periods during the detrended subseasonal δ18O signal is above the 75th percentile (below the 25th percentile) for at least five consecutive days. All δ18O averages referred to in this study are weighted by precipitation amount.
Correlation coefficients between δ18O at each station on days with common precipitation (the number of days used for each correlation is in parentheses). All correlations are significant at the 99% confidence interval or greater.
c. Relationship with synoptic circulations
To examine associations with synoptic circulations, we use data from ERA-Interim, a global atmospheric reanalysis model with 80-km horizontal resolution (Dee et al. 2011). We compare the subseasonal δ18O signals and selected variables from ERA-Interim averaged over the region shown in Fig. 2 (12.0°–18.5°S, 65.0°–74.0°W). This region is selected to encompass all observer sites; it should be noted that a disadvantage of this region is that it includes parts of the Amazon basin and coastal Altiplano with highly dissimilar climatological regimes, and these areas are not considered independently in this study. The following variables from ERA-Interim are tested at both 500 hPa (near surface) and 250 hPa (upper level): relative humidity r, zonal wind u, meridional wind υ, geopotential height z, and temperature t. For each variable significantly correlated (p < 0.01) with the subseasonal δ18O signal for at least three wet seasons, spatial difference plots between the positive and negative δ18O anomalies provide further insight into the relationship between isotopes and synoptic-scale meteorological conditions. Additionally, the possible impact of CAIs on the subseasonal δ18O signal is assessed. CAI event dates are determined using data from ERA-Interim and the method described in Garreaud (2000); when the sea level pressure in a 5° × 5° grid box centered at 25°S, 57.5°W is greater than 1015 hPa, and the sea level pressure tendency is within the top 10% of the seasonal frequency distribution.
d. Signal preservation in the annual-layer snowpack
The propensity of deposited snow to retain the subseasonal isotopic signal observed in precipitation is assessed using isotope samples from a 2.16-m-deep snowpit at 5670 m above mean sea level (MSL) on Quelccaya, sampled on 16 July 2017 (location shown in Fig. 1). A vertical profile created using samples of 8 mm of liquid water equivalent (LWE) were collected until a visible dust horizon and ice layer demarcated the base of the annual layer. Samples were allowed melt in sealed plastic bags (with air removed) and then transferred into glass vials to be processed as described in section 3a.
To compare the isotopic profile of the snowpit to isotopic signal observed in precipitation, it is necessary to develop an age model for the snowpit. To approximate an age model, we use daily snow height and precipitation measurements from an automatic weather station on Quelccaya (Appalachian State University, unpublished data), further details of this weather station are available in Perry et al. (2017). The first day of continuous wet season accumulation on Quelccaya (the start of the snowpit) is determined using snow height measurements. Then, the ratio of the total precipitation recorded by the Quelccaya pluviometer (between the estimated start date and the snowpit sampling date) to the total liquid water equivalent of the snowpit is used to estimate daily snowpit accumulation. This approach assumes no loss of precipitation from the snowpack on Quelccaya due to ablation or sublimation during the wet season. The age model allows a broad comparison between the isotopic profile of the annual-layer snowpit on Quelccaya and the isotopic signal observed in precipitation.
4. Results
a. Overview of patterns in precipitation δ18O
All raw precipitation measurements and isotope data are shown in Fig. 3.1 Because the majority of precipitation (and therefore the majority of sample collection) occurs from November to April, when comparing interannual variability we consider each hydrological year to start and finish in August; thereby capturing the entire wet season. Henceforth, when discussing the year 2017, for example, we refer to 1 August 2016–1 August 2017 (see Fig. 3). A breakdown of all data from individual sites in 2017 (when all 10 sites were active) is provided in the online supplemental material.
Summary of all (top) daily precipitation, (middle) δ18O and (bottom) d-excess measurements used in this study. Lines connect data collected on consecutive days. Dashed blue lines on the δ18O plots are precipitation amount weighted means between 1 Nov and 1 May each year.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
There is a strong linear relationship between δ18O and δD in all samples and all seasons (Fig. 4); the rest of this analysis therefore focuses on δ18O for brevity. Each year, there is a gradual depletion of 18O in precipitation during the wet season, with minimum δ18O observed around March, followed by enrichment at the onset of the dry season (Fig. 3). On subseasonal time scales, there is high variability in δ18O with large amplitude oscillations (>10‰) over 1–7 days. In addition, throughout the wet season, there are periods of higher and lower δ18O that last between 6 and 30 days. Subseasonal variations in δ18O exhibit strong synchronicity at all sites despite large differences in local precipitation amount; this is especially apparent in 2017, where there are data from a larger number of stations that extend along a 500-km transect and range in elevation between 3350 and 5050 m MSL (Fig. 5). The range of δ18O values in the first three years is from −30‰ to +5‰, however in 2017 there are several samples from multiple sites recording δ18O < −35‰. During the exceptionally strong El Niño event of 2015/16 (Xue and Kumar 2017), the annually averaged δ18O in precipitation was 5.6‰ higher than the average of the other three years (Fig. 3).
The δ18O and δD of each daily precipitation sample used in this study in all seasons (2014–17). Different shapes/colors correspond to different stations (see legend). The GMWL (solid black) and local meteoric water line (LMWL; dashed black) are plotted for comparison.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
(a) The 15-day moving average δ18O (weighted by precipitation amount) at each site that collected samples in 2017. (b) As in (a), but for measurements of daily precipitation amount.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
b. D-excess
For the majority of samples at all sites, the relationship between δD and δ18O is very close to the Global Meteoric Water Line (GMWL): the globally averaged relationship between δD and δ18O in meteoric waters that have not undergone surface evaporation (Fig. 4; Craig 1961). Deviations from the GMWL can occur due to kinetic processes during precipitation (e.g., subcloud evaporation), kinetic processes along the moisture transport pathway (e.g., continental moisture recycling) or from changes in evaporation conditions at the moisture source (Pfahl and Sodemann 2014). Local meteoric water lines for each individual site, and for the dry season (JJA) samples compared to the wet season (DJF) samples are provided in the supplementary materials. Highest d-excess values (>30‰) occur during the peak of the wet season (Fig. 3), and are associated with precipitation that is highly depleted in 18O. Over several multiday periods, high d-excess is observed at certain stations in the Cordillera Vilcanota (e.g., Pucarumi and Sallayoc) but not at others that are only 25 km away (i.e., Murmurani, Chillca). This suggests that these high d-excess values may result from local effects, as discussed in section 5b.
c. Spatial variability
Despite a 1700-m difference in elevation between the highest and lowest sites, no relationship is evident between elevation and δ18O in 2017 (Fig. 5). There is also no evident relationship between local precipitation amount and δ18O (Fig. 5). The range of mean δ18O between sites is 4.7‰ and standard deviation is 1.0‰ suggesting a high degree of agreement between sites despite the differences in elevation, daily precipitation, and horizontal separation. There are strong, statistically significant (p < 0.01) correlations between δ18O measured at every site (Table 1). Particularly noteworthy is the consistency of the correlation coefficients among sites within the Cordillera Vilcanota and between the Cordillera Vilcanota and the Cordillera Real. This result clearly demonstrates that subseasonal variations of isotopes in precipitation along the northeastern margin on the Altiplano are recording a common signal.
d. Temporal variability
Since subseasonal variations of δ18O in precipitation are demonstrated to be coherent across the Cordillera Real and the Cordillera Vilcanota, a regionwide δ18O mean is calculated for the 2017 wet season by taking the mean 15-day δ18O weighted by precipitation amount between all sites (Fig. 6a). There are three key modes of temporal variability apparent in these data. The first, characterized by gradual depletion of 18O in precipitation during the wet season and enrichment at the onset of the dry season, is the seasonal cycle, well captured by a 90-day moving average (orange line, Fig. 6a). The second consists of 6–30-day cycles of relatively more-depleted and relatively more-enriched periods that are clearly delineated by the 15-day moving average; this is temporally consistent with the recognized time scale of variations between wet and dry episodes on the Altiplano during the wet season (Lenters and Cook 1997; Garreaud 1999, 2000). The remaining variability relates to short-term “storm scale” oscillations in δ18O. Although these variations may represent an interesting meteorological signal, they are less consistent between sites, and it is likely that postdepositional processes will smooth out this signal in the snowpack; this study therefore focuses on the subseasonal signal (15-day moving average), which is retained in snowpack into the subsequent dry season. Subtracting the seasonal cycle from the 15-day moving average isolates this subseasonal signal, yielding a time series of subseasonal δ18O anomalies (Fig. 6b). Due to the lack of precipitation in the dry season, there are only enough data to define the subseasonal δ18O anomalies between November and May, hence the rest of this study will focus on subseasonal variations of δ18O during this period. Assuming that the spatial coherency of the δ18O variations is true in all years, repeating this analysis for 2014–16 identifies the timing of positive and negative δ18O anomalies along the northeast margin of the Altiplano for each year, despite fewer sampling locations (Fig. 7). This method identifies two or three negative δ18O anomalies each year lasting between 9 and 24 days. In 2014 and 2016 there is one positive anomaly, with two in 2015 and four in 2017. The positive δ18O anomalies last between 9 and 27 days. The amplitude of δ18O anomalies varies from year to year, with the largest amplitude anomalies occurring in 2017.
(a) The δ18O of all precipitation samples collected in 2017 (points; see legend), 3-day precipitation weighted mean of all samples where more than two data points are available (thin black line), 15-day precipitation weighted mean (green line; referred to as the subseasonal signal in this study), and the 90-day precipitation weighted mean (thick orange line; referred to as the seasonal cycle in this study). (b) Subseasonal δ18O anomalies during the 2017 wet season (subseasonal signal minus the seasonal cycle). Red shading highlights positive anomalies, and blue shading highlights negative anomalies.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
(a)–(d) Subseasonal δ18O anomaly plots (as described in Fig. 5b) for each wet season 2014–17.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
e. Relationship with regionwide meteorology and continental scale circulation
The spatiotemporal consistency of δ18O across the observational domain observed in 2017 suggests that regionally coherent meteorological conditions are responsible for the anomalies identified in section 3d. To identify the key meteorological variables that are associated with the δ18O anomaly signal we examined five domain-averaged ERA-Interim variables (geopotential height z, temperature t, zonal wind u, meridional wind υ, and relative humidity r) at two pressure levels (250 hPa and 500 hPa). Correlation coefficients calculated between the δ18O anomaly time series and each variable processed in the same way (i.e., 15-day moving averages minus the seasonal cycle) are presented in Table 2, which also includes correlations between the δ18O anomaly series and the precipitation anomaly series calculated from the mean precipitation between all observer sites. The strongest correlations are with 500- and 250-hPa relative humidity, mean observed precipitation, 250-hPa zonal wind, and 500-hPa meridional wind (Table 2). These variables were examined for spatial differences in synoptic conditions between the positive and negative δ18O anomalies each year (Fig. 8).
Correlation coefficients between δ18O anomalies and selected variables from ERA-Interim averaged over the region 12.0°–18.5°S, 65.0°–74.0°W. All variables are 15-day moving averages minus the seasonal cycle. Variables: r = relative humidity, t = temperature, z = geopotential height, u = zonal wind, υ = meridional wind. The numbers beside each variable refer to the atmospheric pressure level (hPa); P = mean observed precipitation at all sites. Correlations that are significant at the 99% confidence level are highlighted with asterisks.
Difference plots between the average conditions during all positive anomalies and all negative anomalies identified in Fig. 6 (positive minus negative): (a) 500-hPa relative humidity, (b) 250-hPa relative humidity, (c) 250-hPa zonal wind, (d) 250-hPa geopotential height, (e) 500-hPa meridional wind, and (f) 850-hPa meridional wind. The black star is the location of Quelccaya. Stippling denotes differences that are significant at the 99% confidence level.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
On average, the wet season positive δ18O anomalies that occurred in 2014–17 coincide with significantly reduced relative humidity at 500 and 250 hPa over west-central South America relative to the negative δ18O anomalies (Figs. 8a,b). At upper levels (250 hPa), the negative δ18O anomalies are typically associated with strengthened easterly winds over the Amazon basin and increased geopotential heights over southern South America, likely associated with more southward placement of the Bolivian High (Figs. 8c,d; Lenters and Cook 1997; Vuille 1999). In the 500- and 850-hPa meridional wind fields, the negative δ18O anomalies are associated with increased northerly winds, with the largest differences in 850 hPa meridional wind centered on eastern Bolivia (Fig. 8f), aligned along the axis of the SALLJ, suggesting that negative (positive) anomalies are typically associated with stronger (weaker) northerly winds to the east of the Andes.
To assess the possible influence of CAIs on the subseasonal isotope signal, we calculated the dates of CAIs using the method outlined in Garreaud (2000) and plotted them against the δ18O anomaly time series in Fig. 9. Over the four years in this study, there is no consistent response to CAIs observable in the subseasonal isotope signal, with some CAI dates preceding both high δ18O and low δ18O episodes. However, strong relationships are apparent between the strength of the SALLJ at Santa Cruz, domain averaged 500-hPa relative humidity, and δ18O (Fig. 9).
Standardized subseasonal δ18O anomaly signal (thick gray line). Standardized subseasonal variations in 500-hPa relative humidity averaged over the region 18.5°–12°S, 74°–65°W. Standardized subseasonal variations in 850-hPa meridional wind at Santa Cruz (dashed red line). Dates of CAIs are shown by blue circles plotted along y = 0. Pearson’s correlation coefficients in the upper-right corner are between each variable and the subseasonal δ18O anomaly signal. Asterisks identify correlations that are significant at the 99% confidence level.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
At the continental scale, there is a significant positive correlation between the δ18O anomaly signal and detrended 500-hPa relative humidity over the northeastern Brazil (Fig. 10) for each year sampled in the present study. Additionally, there are negative correlations between these same variables over northeast Bolivia. These results are consistent with the findings of Vimeux et al. (2011).
Pearson’s product-moment correlation coefficients between the time series of δ18O anomalies and time series of 500-hPa relative humidity anomalies at every grid point for each year. Black dots highlight areas where the correlation is significant at the 99% confidence level. On each plot the black star shows the location of Quelccaya.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
f. Subseasonal isotopic signal representation in annual-layer snowpack
The potential for subseasonal isotopic signals to become useful diagnostics in ice core paleoclimatic investigations requires that the signal be retained in annual-layer snowpacks. Based on snow height measurements from an automatic weather station on Quelccaya (Appalachian State University, unpublished data), the period of continuous wet season accumulation on Quelccaya approximately begins on 11 November 2016. The subseasonal δ18O signal in precipitation from this date exhibits excellent pattern matching to the δ18O profile from the annual-layer snowpit on Quelccaya (Fig. 11). The average snowpit δ18O is 3‰ lower than the average δ18O from precipitation; although this could be attributed to the observed “altitude effect” whereby precipitation falling at higher elevations is typically more depleted (Gonfiantini et al. 2001), it is odd that such a relationship is not observed in the precipitation data. Another possible explanation for this is that the majority of the observer sites receive liquid precipitation, whereas solid precipitation (accumulating on Quelccaya) is typically more depleted in 18O because isotopic exchange with boundary layer moisture is limited, and because the equilibrium fractionation factor is larger between vapor and ice than it is between ice and liquid (Jouzel and Merlivat 1984).
The δ18O profiles of the annual-layer snowpit on Quelccaya sampled in July 2017 (black line) and the 15-day moving average δ18O signal observed in regionwide precipitation (red dashed line). The start date of the snowpit is 11 Nov 2017 (the approximate first day of continuous wet season accumulation determined from snow height measurements on Quelccaya). Blue triangles are plotted every 5 days showing the snowpit age model, estimated using daily precipitation measurements on Quelccaya. Vertical blue lines are plotted on the first day of each month.
Citation: Journal of Hydrometeorology 20, 5; 10.1175/JHM-D-18-0163.1
5. Discussion
a. δ18O characteristics in the Cordilleras Vilcanota and Real
The range of δ18O variations each year in this study is similar to the range of δ18O identified by previous studies in this region (e.g., Gonfiantini et al. 2001; Fiorella et al. 2015) with the exception of some strongly 18O-depleted events (δ18O < −35‰) that occurred during 2017. It is also very close to the range of values predicted by Grootes et al. (1989). Higher δ18O in precipitation than predicted by Grootes et al. (1989) (δ18O > −4‰) may result from below-cloud evaporation on days when the boundary layer was particularly dry (Risi et al. 2008).
The 9–27-day subseasonal periods of higher and lower δ18O in precipitation superimposed upon the seasonal cycle are consistent with the Zongo Valley study (Vimeux et al. 2011), and with Hurley et al. (2016) who identify similar subseasonal variability in δ18O profiles from eight annual-layer snow pits on Quelccaya. This result suggests that subseasonal δ18O variations are a robust feature of precipitation in the eastern tropical Andes and are retained in annual-layer snowpacks.
Subseasonal variability is coherent across a large domain encompassing, at a minimum, all of our observation sites in both the Cordilleras Real and Vilcanota. This has implications for paleoclimate studies because it demonstrates that the isotopic signals in precipitation that are ultimately preserved in glacial ice have common controls. Therefore, under the assumption that circulation patterns are similar to the present day, the isotopic profiles from ice cores extracted from Quelccaya and from Illimani, for example, can be meaningfully compared. Such agreement between ice cores extracted from Quelccaya and Illimani is already observed on interannual time scales (Hoffmann et al. 2003).
b. Synoptic controls on the subseasonal isotope signal
The excellent spatial matching between the subseasonal δ18O signals at all locations implies that this signal is reflecting synoptic scale rather than local conditions. This finding is in agreement with previous studies that have demonstrated that the local amount effect in this region is weak for precipitation sampled on daily or longer time scales (Vimeux et al. 2011; Fiorella et al. 2015).
However, strong negative correlations between δ18O and regionally averaged precipitation and 500-hPa relative humidity suggest that this variability does reflect regionwide variations in precipitation amount. Additionally, the duration of positive and negative δ18O anomalies closely aligns with the known duration of dry and wet periods in the northeast Altiplano (6–30 days). This result strongly supports that isotopes in tropical Andean precipitation are reflecting the regional “amount effect” at subseasonal time scales. Precipitation amount cannot explain all of the observed variability, however, and the question of what physical processes are controlling the amount effect here remains. The synoptic setup that results in precipitation anomalies over the study area is not the same for each positive and negative δ18O anomaly; therefore, interpreting them as an amount effect alone is likely to have large uncertainties. Of more use to improving paleoclimate reconstructions from ice cores would be to relate the positive and negative δ18O anomalies to the specific synoptic weather systems associated with these precipitation anomalies, and the attendant continental-scale modes of subseasonal variability. Lenters and Cook (1999) describe three synoptic patterns that can result in positive precipitation anomalies on the Altiplano: 1) propagating extratropical cyclones that result in a strengthening of the SALLJ, 2) a westerly shift of the SACZ and anomalous high pressure over the central Amazon basin forcing a strengthening and southward shift of the Bolivian High, and 3) deep narrow bands of low pressure from the subtropics extending into tropical regions along the eastern slopes of the Andes associated with CAIs. When the synoptic patterns associated with individual negative δ18O anomalies are examined (not shown) each anomaly appears to be associated with either scenario 1 or scenario 2. A strengthening and southward shift of the Bolivian high is only observed in scenario 2 and is insufficient to explain all the δ18O anomalies. However, a strengthening of the SALLJ is a robust feature in each negative δ18O anomaly; this is discussed further in section 5c.
Another way of distinguishing the synoptic setup associated with each δ18O anomaly is to consider d-excess in more detail. Several samples from 2017 had very high (>30‰) d-excess that did not occur at all stations simultaneously and that do not appear to be related to station elevation or moisture source trajectories. Most of these high d-excess values occur during the third negative δ18O anomaly that appears to be associated with scenario 1. This δ18O anomaly had weaker 500-hPa relative humidity anomalies compared to the other negative anomalies (both scenario 2) despite being of a similar magnitude. A possible explanation is that this precipitation event was limited in spatial extent; a strong positive precipitation anomaly with limited spatial extent is suggestive of deep convection. Localized high d-excess can occur in the presence of deep convective storms as a result of very high local rain rates (Bony et al. 2008; Galewsky et al. 2016) or kinetic processes during ice crystal formation (Jouzel and Merlivat 1984). Verifying this is beyond the scope of the present study; it is worth noting that Vimeux et al. (2011) observed a similar event in the 2000 season (very low δD with high d-excess).
c. The role of the South American low-level jet
The strengthening of the SALLJ directly to the east of the study site is a robust feature for all of the subseasonal negative δ18O anomalies in all years 2014–17 (Fig. 9) despite other differences in the synoptic setups (discussed in section 5b). Conversely, the majority of positive anomalies are associated with a SALLJ that is anomalously weak, absent or reversed (by the passage of a CAI). For the few positive anomalies during which the SALLJ is present, the 500-hPa relative humidity directly to the east of the Altiplano is anomalously low. This result implies that SALLJ plays a key role in enhancing moisture transport from the western Amazon basin to the study region, by strengthening the upslope flow that advects moisture to mountains along northwest-oriented valleys (Marengo et al. 2002; Junquas et al. 2018). When there is sufficient moisture available, this results in the positive relative humidity anomalies that are associated with negative δ18O anomalies.
In addition, precipitation along the eastern edge of the Altiplano occurs under two key modes: afternoon convection and widespread nocturnal stratiform events (Perry et al. 2014, 2017; Chavez and Takahashi 2017; Endries et al. 2018). More organized precipitating systems with large stratiform components are associated with decreased δ18O in precipitation compared to localized convective storms (Kurita et al. 2011; Aggarwal et al. 2016). The marked difference between the isotopic composition in stratiform and convective precipitation types results from differences in cloud dynamics and hydrometeor formation. In convective storms, boundary layer moisture (more enriched in heavy isotopes) is entrained into the cloud by strong updrafts, whereas in stratiform precipitation, hydrometeors grow by vapor diffusion above the melting layer where moisture is relatively depleted in heavy isotopes, and below cloud subsidence may result in low δ18O vapor transport to the subcloud layer. See Aggarwal et al. (2016) for a detailed description of this process. The SALLJ is strongest overnight (Garreaud and Wallace 1997; Marengo et al. 2004; Junquas et al. 2018), and Junquas et al. (2018) present evidence from numerical simulations that the upslope flow influenced by the presence of a strong SALLJ persists overnight. It is possible that this enhanced moisture transport combined with the stable nighttime atmosphere may mechanically produce precipitation more stratiform in structure, and hence with lower δ18O.
Relationships between positive precipitation anomalies and the strength of the SALLJ (e.g., Garreaud 1999; Lenters and Cook 1999; Junquas et al. 2018), and negative δ18O anomalies and the strength of the SALLJ (e.g., Vimeux et al. 2011; Insel et al. 2013) have been previously suggested. Our results indicate that the presence and strength of the SALLJ is a key—and probably dominant—control on moisture availability and the type and extent of precipitation during the wet season that ultimately results in subseasonal δ18O anomalies that are synchronous and coherent across a wide region. On a broader scale, the strength of the SALLJ relates to continental scale precipitation variability by association with the South American seesaw dipole (González and Vera 2009) migration of the SACZ, this is discussed further in section 5e.
d. The role of cold air incursions
Lenters and Cook (1997) propose CAIs as the third mechanism to explain positive precipitation anomalies. More recent studies have found evidence that CAIs may modulate subseasonal variations of δ18O in the snowpack on Quelccaya (Hurley et al. 2015, 2016). However, on the subseasonal time scale that is the focus of the present study, none of the negative δ18O anomalies identified appear to result from CAIs, and we observe no response to CAIs in the 15-day moving average signal of isotope anomalies in precipitation. It is possible that CAIs do influence isotopes in precipitation on shorter time scales; however, signals over shorter time scales are not observed in the annual-layer snowpit on Quelccaya (Fig. 11).
e. Relationship with continental-scale circulation patterns
The controls on stable isotopes in precipitation using daily resolution sampling over an entire season (1999–2000 hydrological year) were examined by Vimeux et al. (2011), who used observations from several sites in the Zongo valley near Huayna Potosí in the Cordillera Real (Fig. 1). They found that intraseasonal variations in δD in the Zongo valley are associated with the South American seesaw, a continental precipitation dipole related to the position of the SACZ, whereby higher (lower) δD in precipitation in the Zongo valley coincides with enhanced (reduced) convection over northeastern Brazil and reduced (enhanced) convection over the subtropical plains. The results of the present study are consistent with these findings and demonstrate that this correspondence is observed over a number of years and applies to a broader region, encompassing both the Cordillera Vilcanota and Cordillera Real. The South American seesaw has a period of 15–20 days (Nogués-Paegle and Mo 1997), consistent with the leading mode of subseasonal variability in δD identified by Vimeux et al. (2011), and of the same order of the as the subseasonal oscillations depicted in the present study. In one phase, the SACZ strengthens, precipitation over the subtropics is reduced and the SALLJ weakens (González and Vera 2009)–associated with higher δ18O/δD along the northeastern margin of the Altiplano (Vimeux et al. 2011 and the present study). In the opposite phase, precipitation in the subtropics increases associated with a strengthening of the SALLJ, increased moisture influx from the Amazon basin, and lower δ18O/δD. The phase of the South American seesaw dipole is potentially related to the Madden–Julian oscillation (Nogués-Paegle and Mo 1997; Paegle et al. 2000; Alvarez et al. 2016), implying that subseasonal variations in isotopes preserved in annual-layer snowpits in the Cordilleras Real and Vilcanota might record information about hemispheric tropical climate variability.
f. The 2015/16 El Niño recorded in isotopes in precipitation
Although not a primary focus of this study, our isotope sampling period includes the strong El Niño year of 2015/16, providing the opportunity to compare isotopic variations between a strong El Niño and non–El Niño years. During 2015/16, there was a severe reduction in snow accumulation on Quelccaya and annually averaged δ18O in the annual-layer snowpack was higher by 5.2‰ (Thompson et al. 2017). The results of this study are consistent with this finding: the average δ18O in precipitation during 2016 is 5.6‰ higher than the average of the other three years. This establishes that the change in δ18O in the snowpack on Quelccaya resulted from more precipitation that was more enriched in 18O at the time of deposition rather than postdepositional enrichment. Despite this, we still observe subseasonal variations in δ18O that occur on similar time scale to the other years, suggesting that while the El Niño event affected the baseline isotopic composition of precipitation, as described by previous studies (e.g., Vuille et al. 2003), the processes driving the subseasonal variability are the same.
g. Retention of the subseasonal signal in annual-layer snowpack
The annual-layer snowpack on Quelccaya deposited in 2017 and measured in the subsequent dry season clearly retains the regionwide 15-day averaged δ18O signal observed in precipitation (Fig. 11). The isotopic profile therefore effectively records subseasonal changes in synoptic conditions and, in particular, changes in the strength of the SALLJ and in moisture availability in the Amazon lowlands directly east of the Altiplano. To apply the results of this study to interpret paleoclimate signals, it is necessary to consider how the processes that act together to generate the signal we observe today might have changed in the past, and how the subseasonal isotopic signal is retained and possibly modified in deeper firn and ice. Molecular diffusion during firnification smooths the isotopic signal over time and, in deep ice, even the seasonal isotopic signal cannot be identified (e.g., Ramirez et al. 2003). However, the isotopic profile from the ice core extracted from Quelccaya in 2003 exhibits annually resolvable resolution for 1500 years (Thompson et al. 2006). Although subseasonal isotopic signals are diminished before they are interred in glacial ice (e.g., Thompson et al. 2017), some insoluble chemical tracers that fluctuate in response to the same climatic variability may remain in place within annual layering. How much subseasonal variability can be detected in both isotopes and chemical tracers from new ice cores extracted from both cordilleras, which will be sampled using new techniques, remains to be seen.
6. Conclusions
This study demonstrates that subseasonal variations in δ18O in precipitation in the Cordillera Vilcanota and the Cordillera Real are spatially coherent and reflect variability in synoptic and continental-scale meteorology. Periods of relatively high and low δ18O in precipitation during the wet season superimposed on the seasonal cycle last between 6 and 30 days and are observed over a multiyear period including during a strong El Niño year. These results expand upon a previous study of subseasonal isotopic variations in the same region (Vimeux et al. 2011), and demonstrate that the findings in Vimeux et al. (2011) hold for a wider region along the northeastern margin of the Altiplano and over a multiyear period. Events with low (high) δ18O are associated with anomalously high (low) midtropospheric relative humidity averaged over the study area that is in turn influenced by propagating extratropical Rossby waves and correlate with variability in the strength and position of the SACZ. Anomalously low δ18O in precipitation occurs when the SALLJ is strengthened, and anomalously high δ18O occurs when the SALLJ is weakened or reversed, or when there is reduced relative humidity in the Amazon basin along the eastern tropical Andes. The relationship between subseasonal δ18O variability and the continental-scale precipitation dipole associated with the changes in the position of the SACZ implies that subseasonal δ18O variability encodes information about continental climatic anomalies and may be related to the Madden Julian oscillation; such a relationship is an important area for future research. This subseasonal variability is retained in the isotopic profile of the annual-layer snowpit on Quelccaya, enabling the development of an accurate snowpit age model, and demonstrating that the snowpit isotopic profile reflects synoptic and continental-scale meteorological variability. These results suggest that annual layering in Andean ice cores may contain information about subseasonal variability in precipitation delivery and associated circulation anomalies, which will improve future efforts to better understand Andean hydroclimate through ice cores.
Acknowledgments
We thank our citizen scientist observers in Peru and Bolivia; all participants in fieldwork and snow pit efforts 2009–17; and university partners at UNSAAC-Cusco and UMSA-La Paz. In particular, we thank Courtney Cooper, Jason Endries, Eric Burton, Evan Montpellier, Adrián Ccahuana, Nilton Montoya, Sandro Arias, Ronald Winkelmann, Marcelo Peñaloza, and Laura Ticona for their contributions. We thank the University of Arkansas Stable Isotope Lab for analyzing samples. Finally, we thank Françoise Vimeux and two anonymous reviewers for their valuable and constructive feedback. This work is funded by U.S. National Science Foundation Grants AGS 1566450 and AGS1347179.
REFERENCES
Aggarwal, P. K., U. Romatschke, L. Araguas-Araguas, D. Belachew, F. J. Longstaffe, P. Berg, C. Schumacher, and A. Funk, 2016: Proportions of convective and stratiform precipitation revealed in water isotope ratios. Nat. Geosci., 9, 624–629, https://doi.org/10.1038/ngeo2739.
Alvarez, M. S., C. Vera, G. Kiladis, and B. Liebmann, 2016: Influence of the Madden Julian Oscillation on precipitation and surface air temperature in South America. Climate Dyn., 46, 245–262, https://doi.org/10.1007/s00382-015-2581-6.
Appalachian State University, 2018: Andean citizen scientist network: Precipitation isotope data 2013–2017. http://climate.appstate.edu/Data/Isotopes/.
Bony, S., C. Risi, and F. Vimeux, 2008: Influence of convective processes on the isotopic composition (δ18O and δD) of precipitation and water vapor in the tropics: 1. Radiative-convective equilibrium and tropical ocean–global atmosphere–coupled ocean-atmosphere response experiment (TOGA-COARE) simulations. J. Geophys. Res., 113, D19305, https://doi.org/10.1029/2008JD009942.
Chavez, S. P., and K. Takahashi, 2017: Orographic rainfall hot spots in the Andes-Amazon transition according to the TRMM precipitation radar and in situ data. J. Geophys. Res. Atmos., 122, 5870–5882, https://doi.org/10.1002/2016JD026282.
Cifelli, R., N. Doesken, P. Kennedy, L. D. Carey, S. A. Rutledge, C. Gimmestad, and T. Depue, 2005: The community collaborative rain, hail, and snow network: Informal education for scientists and citizens. Bull. Amer. Meteor. Soc., 86, 1069–1077, https://doi.org/10.1175/BAMS-86-8-1069.
Craig, H., 1961: Isotopic variations in meteoric waters. Science, 133, 1702–1703, https://doi.org/10.1126/science.133.3465.1702.
Dansgaard, W., 1964: Stable isotopes in precipitation. Tellus, 16, 436–468, https://doi.org/10.3402/tellusa.v16i4.8993.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Endries, J. L., and Coauthors, 2018: Radar-observed characteristics of precipitation in the tropical high Andes of southern Peru and Bolivia. J. Appl. Meteor. Climatol., 57, 1441–1458, https://doi.org/10.1175/JAMC-D-17-0248.1.
Field, R. D., D. Jones, and D. P. Brown, 2010: Effects of postcondensation exchange on the isotopic composition of water in the atmosphere. J. Geophys. Res., 115, D24305, https://doi.org/10.1029/2010JD014334.
Fiorella, R. P., C. J. Poulsen, R. S. Pillco Zolá, J. B. Barnes, C. R. Tabor, and T. A. Ehlers, 2015: Spatiotemporal variability of modern precipitation δ18O in the central Andes and implications for paleoclimate and paleoaltimetry estimates. J. Geophys. Res. Atmos., 120, 4630–4656, https://doi.org/10.1002/2014JD022893.
Friedman, I., L. Machta, and R. Soller, 1962: Water-vapor exchange between a water droplet and its environment. J. Geophys. Res., 67, 2761–2766, https://doi.org/10.1029/JZ067i007p02761.
Galewsky, J., H. C. Steen-Larsen, R. D. Field, J. Worden, C. Risi, and M. Schneider, 2016: Stable isotopes in atmospheric water vapor and applications to the hydrologic cycle. Rev. Geophys., 54, 809–865, https://doi.org/10.1002/2015RG000512.
Garreaud, R., 1999: Multiscale analysis of the summertime precipitation over the central Andes. Mon. Wea. Rev., 127, 901–921, https://doi.org/10.1175/1520-0493(1999)127<0901:MAOTSP>2.0.CO;2.
Garreaud, R., 2000: Intraseasonal variability of moisture and rainfall over the South American Altiplano. Mon. Wea. Rev., 128, 3337–3346, https://doi.org/10.1175/1520-0493(2000)128<3337:IVOMAR>2.0.CO;2.
Garreaud, R., 2009: The Andes climate and weather. Adv. Geosci., 22, 3–11, https://doi.org/10.5194/adgeo-22-3-2009.
Garreaud, R., and J. M. Wallace, 1997: The diurnal march of convective cloudiness over the Americas. Mon. Wea. Rev., 125, 3157–3171, https://doi.org/10.1175/1520-0493(1997)125<3157:TDMOCC>2.0.CO;2.
Garreaud, R., M. Vuille, and A. C. Clement, 2003: The climate of the Altiplano: observed current conditions and mechanisms of past changes. Palaeogeogr. Palaeoclimatol. Palaeoecol., 194, 5–22, https://doi.org/10.1016/S0031-0182(03)00269-4.
Gat, J., and E. Matsui, 1991: Atmospheric water balance in the Amazon Basin: an isotopic evapotranspiration model. J. Geophys. Res., 96, 13 179–13 188, https://doi.org/10.1029/91JD00054.
Gehre, M., H. Geilmann, J. Richter, R. Werner, and W. Brand, 2004: Continuous flow 2H/1H and 18O/16O analysis of water samples with dual inlet precision. Rapid Commun. Mass Spectrom., 18, 2650–2660, https://doi.org/10.1002/rcm.1672.
Gonfiantini, R., 1978: Standards for stable isotope measurements in natural compounds. Nature, 271, 534–536, https://doi.org/10.1038/271534a0.
Gonfiantini, R., M.-A. Roche, J.-C. Olivry, J.-C. Fontes, and G. M. Zuppi, 2001: The altitude effect on the isotopic composition of tropical rains. Chem. Geol., 181, 147–167, https://doi.org/10.1016/S0009-2541(01)00279-0.
González, P. L., and C. S. Vera, 2009: Intraseasonal variability over South America in WCRP/CMIP3 simulations. Proc. 9th Int. Conf. on Southern Hemisphere Meteorology and Oceanography, Melbourne, Australia, Amer. Meteor. Soc.
Grootes, P., M. Stuiver, L. Thompson, and E. Mosley-Thompson, 1989: Oxygen isotope changes in tropical ice, Quelccaya, Peru. J. Geophys. Res., 94, 1187–1194, https://doi.org/10.1029/JD094iD01p01187.
Hoffmann, G., and Coauthors, 2003: Coherent isotope history of Andean ice cores over the last century. Geophys. Res. Lett., 30, 1179, https://doi.org/10.1029/2002GL014870.
Hurley, J. V., M. Vuille, D. R. Hardy, S. J. Burns, and L. G. Thompson, 2015: Cold air incursions, δ18O variability, and monsoon dynamics associated with snow days at Quelccaya Ice Cap, Peru. J. Geophys. Res. Atmos., 120, 7467–7487, https://doi.org/10.1002/2015JD023323.
Hurley, J. V., M. Vuille, and D. Hardy, 2016: Forward modeling of δ18O in Andean ice cores. Geophys. Res. Lett., 43, 8178–8188, https://doi.org/10.1002/2016GL070150.
Insel, N., C. J. Poulsen, and T. A. Ehlers, 2010: Influence of the Andes Mountains on South American moisture transport, convection, and precipitation. Climate Dyn., 35, 1477–1492, https://doi.org/10.1007/s00382-009-0637-1.
Insel, N., C. J. Poulsen, C. Sturm, and T. A. Ehlers, 2013: Climate controls on Andean precipitation δ18O interannual variability. J. Geophys. Res. Atmos., 118, 9721–9742, https://doi.org/10.1002/jgrd.50619.
Jouzel, J., and L. Merlivat, 1984: Deuterium and oxygen 18 in precipitation: Modeling of the isotopic effects during snow formation. J. Geophys. Res., 89, 11 749–11 757, https://doi.org/10.1029/JD089iD07p11749.
Junquas, C., K. Takahashi, T. Condom, J.-C. Espinoza, S. Chavez, J.-E. Sicart, and T. Lebel, 2018: Understanding the influence of orography on the precipitation diurnal cycle and the associated atmospheric processes in the central Andes. Climate Dyn., 50, 3995–4017, https://doi.org/10.1007/s00382-017-3858-8.
Kurita, N., 2013: Water isotopic variability in response to mesoscale convective system over the tropical ocean. J. Geophys. Res. Atmos., 118, 10 376–10 390, https://doi.org/10.1002/jgrd.50754.
Kurita, N., D. Noone, C. Risi, G. A. Schmidt, H. Yamada, and K. Yoneyama, 2011: Intraseasonal isotopic variation associated with the Madden-Julian Oscillation. J. Geophys. Res., 116, D24101, https://doi.org/10.1029/2010JD015209.
Lenters, J., and K. Cook, 1997: On the origin of the Bolivian High and related circulation features of the South American climate. J. Atmos. Sci., 54, 656–678, https://doi.org/10.1175/1520-0469(1997)054<0656:OTOOTB>2.0.CO;2.
Lenters, J., and K. Cook, 1999: Summertime precipitation variability over South America: Role of the large-scale circulation. Mon. Wea. Rev., 127, 409–431, https://doi.org/10.1175/1520-0493(1999)127<0409:SPVOSA>2.0.CO;2.
Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28, 702–708, https://doi.org/10.1175/1520-0469(1971)028<0702:DOADOI>2.0.CO;2.
Mann, M. E., and P. D. Jones, 2003: Global surface temperatures over the past two millennia. Geophys. Res. Lett., 30, 1820, https://doi.org/10.1029/2003GL017814.
Marengo, J. A., M. W. Douglas, and P. L. Silva Dias, 2002: The South American low-level jet east of the Andes during the 1999 LBA-TRMM and LBA-WET AMC campaign. J. Geophys. Res., 107, 8079, https://doi.org/10.1029/2001JD001188.
Marengo, J. A., W. R. Soares, C. Saulo, and M. Nicolini, 2004: Climatology of the low-level jet east of the Andes as derived from the NCEP NCAR reanalyses: Characteristics and temporal variability. J. Climate, 17, 2261–2280, https://doi.org/10.1175/1520-0442(2004)017<2261:COTLJE>2.0.CO;2.
Mayewski, P., S. Sneed, S. Birkel, A. Kurbatov, and K. Maasch, 2014: Holocene warming marked by abrupt onset of longer summers and reduced storm frequency around Greenland. J. Quat. Sci., 29, 99–104, https://doi.org/10.1002/jqs.2684.
Nelson, S. T., 2000: A simple, practical methodology for routine VSMOW/SLAP normalization of water samples analyzed by continuous flow methods. Rapid Commun. Mass Spectrom., 14, 1044–1046, https://doi.org/10.1002/1097-0231(20000630)14:12<1044::AID-RCM987>3.0.CO;2-3.
Nogués-Paegle, J., and K. C. Mo, 1997: Alternating wet and dry conditions over South America during summer. Mon. Wea. Rev., 125, 279–291, https://doi.org/10.1175/1520-0493(1997)125<0279:AWADCO>2.0.CO;2.
Paegle, J. N., L. A. Byerle, and K. C. Mo, 2000: Intraseasonal modulation of South American summer precipitation. Mon. Wea. Rev., 128, 837–850, https://doi.org/10.1175/1520-0493(2000)128<0837:IMOSAS>2.0.CO;2.
Perry, L. B., A. Seimon, and G. M. Kelly, 2014: Precipitation delivery in the tropical high Andes of southern Peru: new findings and paleoclimatic implications. Int. J. Climatol., 34, 197–215, https://doi.org/10.1002/joc.3679.
Perry, L. B., and Coauthors, 2017: Characteristics of precipitating storms in glacierized tropical Andean cordilleras of Peru and Bolivia. Ann. Assoc. Amer. Geogr., 107, 309–322, https://doi.org/10.1080/24694452.2016.1260439.
Pfahl, S., and H. Sodemann, 2014: What controls deuterium excess in global precipitation? Climate Past, 10, 771–781, https://doi.org/10.5194/cp-10-771-2014.
Rabatel, A., and Coauthors, 2013: Current state of glaciers in the tropical Andes: A multi-century perspective on glacier evolution and climate change. Cryosphere, 7, 81–102, https://doi.org/10.5194/tc-7-81-2013.
Ramirez, E., and Coauthors, 2003: A new Andean deep ice core from Nevado Illimani (6350 m), Bolivia. Earth Planet. Sci. Lett., 212, 337–350, https://doi.org/10.1016/S0012-821X(03)00240-1.
Risi, C., S. Bony, F. Vimeux, L. Descroix, B. Ibrahim, E. Lebreton, I. Mamadou, and B. Sultan, 2008: What controls the isotopic composition of the African monsoon precipitation? Insights from event-based precipitation collected during the 2006 AMMA field campaign. Geophys. Res. Lett., 35, L24808, https://doi.org/10.1029/2008GL035920.
Rozanski, K., L. Araguás-Araguás, and R. Gonfiantini, 1993: Isotopic patterns in modern global precipitation. Climate Change in Continental Isotopic Records, Geophys. Monogr., Vol. 78, Amer. Geophys. Union, 1–36.
Sneed, S. B., and Coauthors, 2015: New LA-ICP-MS cryocell and calibration technique for sub-millimeter analysis of ice cores. J. Glaciol., 61, 233–242, https://doi.org/10.3189/2015JoG14J139.
Sturm, C., F. Vimeux, and G. Krinner, 2007: Intraseasonal variability in South America recorded in stable water isotopes. J. Geophys. Res., 112, D20118, https://doi.org/10.1029/2006JD008298.
Thompson, L. G., 2000: Ice core evidence for climate change in the Tropics: Implications for our future. Quat. Sci. Rev., 19, 19–35, https://doi.org/10.1016/S0277-3791(99)00052-9.
Thompson, L. G., and Coauthors, 2006: Abrupt tropical climate change: Past and present. Proc. Natl. Acad. Sci. USA, 103, 10 536–10 543, https://doi.org/10.1073/pnas.0603900103.
Thompson, L. G., and Coauthors, 2017: Impacts of recent warming and the 2015/2016 El Niño on tropical Peruvian ice fields. J. Geophys. Res. Atmos., 122, 12 688–12 701, https://doi.org/10.1002/2017JD026592.
Thompson, L. G., E. Mosley-Thompson, M. Davis, V. Zagorodnov, I. Howat, V. Mikhalenko, and P.-N. Lin, 2013: Annually resolved ice core records of tropical climate variability over the past ~1800 years. Science, 340, 945–950, https://doi.org/10.1126/science.1234210.
Vera, C., and Coauthors, 2006: The South American low-level jet experiment. Bull. Amer. Meteor. Soc., 87, 63–77, https://doi.org/10.1175/BAMS-87-1-63.
Vimeux, F., R. Gallaire, S. Bony, G. Hoffmann, and J. C. Chiang, 2005: What are the climate controls on δd in precipitation in the Zongo Valley (Bolivia)? Implications for the Illimani ice core interpretation. Earth Planet. Sci. Lett., 240, 205–220, https://doi.org/10.1016/j.epsl.2005.09.031.
Vimeux, F., P. Ginot, M. Schwikowski, M. Vuille, G. Hoffmann, L. G. Thompson, and U. Schotterer, 2009: Climate variability during the last 1000 years inferred from Andean ice cores: A review of methodology and recent results. Palaeogeogr. Palaeoclimatol. Palaeoecol., 281, 229–241, https://doi.org/10.1016/j.palaeo.2008.03.054.
Vimeux, F., G. Tremoy, C. Risi, and R. Gallaire, 2011: A strong control of the South American SeeSaw on the intra-seasonal variability of the isotopic composition of precipitation in the Bolivian Andes. Earth Planet. Sci. Lett., 307, 47–58, https://doi.org/10.1016/j.epsl.2011.04.031.
Vuille, M., 1999: Atmospheric circulation over the Bolivian Altiplano during dry and wet periods and extreme phases of the Southern Oscillation. Int. J. Climatol., 19, 1579–1600, https://doi.org/10.1002/(SICI)1097-0088(19991130)19:14<1579::AID-JOC441>3.0.CO;2-N.
Vuille, M., and M. Werner, 2005: Stable isotopes in precipitation recording South American summer monsoon and ENSO variability: Observations and model results. Climate Dyn., 25, 401–413, https://doi.org/10.1007/s00382-005-0049-9.
Vuille, M., R. Bradley, R. Healy, M. Werner, D. Hardy, L. Thompson, and F. Keimig, 2003: Modeling δ18O in precipitation over the tropical Americas: 2. Simulation of the stable isotope signal in Andean ice cores. J. Geophys. Res., 108, 4175, https://doi.org/10.1029/2001JD002039.
Vuille, M., and Coauthors, 2018: Rapid decline of snow and ice in the tropical Andes–impacts, uncertainties and challenges ahead. Earth Sci. Rev., 176, 195–213, https://doi.org/10.1016/j.earscirev.2017.09.019.
Xue, Y., and A. Kumar, 2017: Evolution of the 2015/16 El Niño and historical perspective since 1979. Sci. China Earth Sci., 60, 1572–1588, https://doi.org/10.1007/s11430-016-0106-9.
Since the writing of this article, it has come to our attention (by cross-checking with rainfall data from Cusco airport) that the observer at Perayoc began back-dating all observations from 18 November 2016 by one day. As such, the precipitation amount, δ18O and d-excess observations published here are shifted one day backwards relative to observations from the other sites. Since the results of this study are based on 15-day averages, this discrepancy does not change any of the conclusions reached in this paper. A note has also been added to the online data repository.