Stable isotope variations of daily precipitation from 2014–2018 in the central United States

Stable isotopes of hydrogen and oxygen (δ2H, δ18O and δ17O) serve as powerful tracers in hydrological investigations. To our knowledge, daily precipitation isotope record especially 17O-excess is rare in the mid-latitudes. To fill such knowledge gap, daily precipitation samples (n=446) were collected from June 2014 to May 2018 in Indianapolis, Indiana, U.S. A Triple Water Vapor Isotope Analyzer (T-WVIA) based on Off-Axis Integrated Cavity Output Spectroscopy (OA-ICOS) technique was used to concurrently measure precipitation isotopic variations (δ2H, δ18O and δ17O). Meanwhile, 17O-excess and d-excess as second-order isotopic variables were calculated to provide additional information on precipitation formation and transport mechanisms. This study presents a four-year daily precipitation isotope dataset for mid-latitudes, and makes it available to researchers around the world who may use it as a reference for site comparisons and for assessing global hydrological models.

Stable isotopic compositions of precipitation are affected by complex meteorological and geographical factors, such as atmospheric conditions at the moisture source and precipitation site, moisture transport trajectories, altitude of condensation and latitude 5,[9][10][11] . There are two types of mass-dependent fractionation process (i.e., equilibrium fractionation and kinetic fractionation) during the precipitation formation. The individual stable isotopes (δ 2 H, δ 18 O and δ 17 O) demonstrate different sensitivities to equilibrium and kinetic fractionation processes 12 . Two second-order isotopic variables, deuterium excess (d-excess = δ 2 H − 8 × δ 18 O) 13 and 17 O-excess ( 17 O-excess = ln (δ 17 O + 1) − 0.528 × ln (δ 18 O + 1)) 14 , can be utilized to provide additional constraints. The d-excess is sensitive to the kinetic fractionation processes due to the elimination of the 2 H and 18 O co-variation during the equilibrium fractionation 9,15 . The d-excess of precipitation is influenced by both moisture source temperature and relative humidity (hereafter RH). Similar to d-excess, 17 O-excess is also sensitive to the kinetic fractionation (e.g., evaporation and condensation in supersaturation condition) 16,17 . However, theoretically 17 O-excess is mainly sensitive to the RH due to the canceled temperature effect on 18 O and 17 O 14,18,19 . 17 O-excess therefore could serve as a new tracer to better understand hydrological and meteorological processes. 17 O-excess in polar ice cores has been used to reconstruct past climate over glacial-interglacial cycles 12,[20][21][22] . The evolution of 17 O-excess reflects the different microphysical processes along the squall line and is sensitive to convective processes in African precipitation 23 . Recent studies show that the relationship between 18 O and 17 O can be used to differentiate drought type (e.g., synoptic drought vs. local drought) 24 and differentiate fog and dew formations at the Namib Desert 25 . Thus far, there are few studies on precipitation 17 O-excess in the middle latitude regions 11,26 . δ 17 O measurements with acceptable precision has been challenging because of its low natural abundance. The traditional Isotope Ratio Mass Spectrometry (IRMS) technique is one of the most widely used approaches to measure δ 17 O. However, it is complicated, expensive and time-consuming, and can only be carried out in a small number of laboratories worldwide 6,19,27 . In recent years, laser absorption spectroscopy (LAS) techniques including Cavity Ring Down Spectroscopy (CRDS) and Off-Axis Integrated Cavity Output Spectroscopy (OA-ICOS) technique have been developed for δ 17 O analysis. Based on the recent assessments, the precision of CRDS and OA-ICOS δ 17 O and 17 O-excess measurements are lower than traditional IRMS technique, but almost comparable 6,7,11,26,28 .
The  17 Oexcess data filter method was described which was found to be useful to quality control the dataset as demonstrated in our recent work 26 . It is the first publicly available daily precipitation isotope dataset (δ 2 H, δ 18 O, δ 17 O, d-excess and 17 O-excess) from the central United States, which would provide valuable information for scientists for site comparisons and assessing global hydrological models.

Sample collections
The sampling location is Zionsville (Indianapolis), Indiana of the central United States (39.88°N, 86.27°W). The sampling device is placed on the ground with a diameter of~35 cm and volume of 6000 ml. We collected 446 daily precipitation samples from June 2014 to May 2018. To reduce evaporation effects on isotopes, samples were immediately transferred from the precipitation collector to sealed glass vials (Qorpak Bottles, Fisher Scientific Co. Germany) except for those occurring after midnight. In those cases, they were collected at the earliest possible time in the morning. Snowfall samples were first melted in sealed plastic bags and then poured into the vials. All of the samples were stored at 4°C until isotope analysis. Notably, samples containing impurities were filtered with 0.45 μm syringe filters (Cellulose Nitrate Membrane Filters, GE Healthcare Co. UK) or centrifuged (Iec Centra CL2 Centrifuge, Thermo Electron Co. USA) depending on the size of the impurities before being measured. The meteorological data during the study period were obtained from the Zionsville meteorological station (https://www.wunderground.com).  16 O ratios of all the precipitation samples were continually and simultaneously measured at IUPUI (Indiana University-Purdue University Indianapolis) Ecohydrology Lab, as described in our previous studies 28,32 . Typically a minimum of 0.5 ml sample is needed to ensure the data quality. The water isotopic ratios were expressed in δ-notation as a deviation from a reference ratio:

Isotope measurements
where R is the atomic ratio (e.g., 2 H/H, 18  To achieve high precision, the following procedure was followed as described in our earlier work 28,32 . The internal temperature of WVISS was preheated to 80 o C to ensure complete vaporization of the liquid sample. The process usually takes about 2 h when the ambient temperature is about 25 o C. The T-WVIA was also turned on about 2 h before the measurements to ensure ideal measuring conditions with chamber temperature and gas pressure being around 50 o C and 40 Torr during measurements. Pipeheating cable was used to heat the Teflon tubing connecting the WVISS and T-WVIA to avoid condensation of water vapor.
To avoid memory effects from residual water, the WVISS nebulizer was first purged for at least 2 min, and then the "stabilize" option of the device was turned on for 2 min to expel residual air inside the vaporizing chamber. The vapor concentration was adjusted by the "dilution control" knob through controlling the flow rates of dry air and the liquid water sample. All the samples were measured under 13000 ppm with higher precision based on our previous work 26, 28 . Each sample was measured for 2 min, and the data output frequency was 1 Hz, which means 120 data points were generated for each sample.

Isotope calibration and normalization
To routine checking the instrument performance, five commercially available working standards from LGR with known isotopic composition (Table 1) were analyzed as reference waters after every five precipitation samples.
Additionally, in order to reduce inter-laboratory difference using different technique and calibration methods, all of the isotope ratios were normalized using two International Atomic Energy Agency (IAEA) standards VSMOW and Standard Light Antarctic Precipitation (SLAP) as calibration materials. "Measured" δ value with respect to VSMOW was first calculated using the formula below described by Steig et al. 7 : where δ is the δ 2 H, δ 18 O or δ 17 O, and "raw" value is directly derived from the ratio of measured isotopologue abundance. Then, normalization to the VSMOW-SLAP scale was following the procedure described in Schoenemann et al. 27 :  20 . To minimize sources of error, two types of quality control filters were used to check each individual data point. One is regression coefficient (λ = ln (δ 17 O + 1)/ln (δ 18 O + 1)), which will be the same as mass-dependent fractionation coefficient (θ) during the isotopic fractionation processes of liquid-vapor equilibrium and in water vapor diffusion in air 2,19 . The fractionation coefficient of oxygen isotope was found to be 0.511 ± 0.005 for kinetic transport effects 2 and 0.529 ± 0.001 for equilibrium effects 19 . The other restriction is 17    +100 per meg 11,17,23,[34][35][36] . Therefore, to attain better precision of 17 O-excess, any measurements outside the 0.506 and 0.530 range, as well as outside the observed range (−100 to +100 per meg), were removed from the analysis. The final 17 O-excess value for every precipitation sample was given as the mean value of quality-controlled data. To check the precision of our measurements, SLAP and the five working standards from LGR as mentioned above were used to calculate the precision. Additionally, Greenland  Ice Sheet Precipitation (GISP), another international standard, was also measured to check the stability of our instrument precision.

Code Availability
No custom code was used in this work.

Data Records
Daily precipitation isotope database is archived in PANGAEA in a single   36 (Fig. 2).

Technical Validation
Multiple standards were used to validate our measurements and our measurement precision was compared with reported values in the literature (Tables 4 and 5). The precision of SLAP in our measurements was 0.79‰, 0.04‰, 0.02‰ and 3 per meg for δ 2 H, δ 18 17,21,27,[34][35][36] , as well as for CRDS method (7 to 10 per meg) 7,11 and another type of OA-ICOS water analyzer (10 to 18 per meg) 6 (Table 5). Meanwhile, the precisions of the three individual isotopes (δ 2 H, δ 18 O and δ 17 O) were also acceptable compared with the previous studies (Table 5).