Search for the Blazhko effect in field RR Lyrae stars using LINEAR and ZTF light curves

Introduction

RR Lyrae stars are pulsating variable stars with periods in the range of 3–30 hours and large amplitudes that increase towards blue optical bands (e.g., in the SDSS $g$ band from 0.2 mag to 1.5 mag; Sesar et al. 2010). For comprehensive reviews of RR Lyrae stars, we refer the reader to Smith (1995) and Catelan [9].

RR Lyrae stars often exhibit amplitude and phase modulation, or the so-called Blazhko effect¹ (hereafter, “Blazhko stars”). For examples of well-sampled observed light curves showing the Blazhko effect, see, e.g., Kepler data shown in Figure 1 and Figure 2 from Benkő et al. [7]. The Blazhko effect has been known for a long time (Blažko [8]), but its detailed observational properties and theoretical explanation of its causes remain elusive (Kolenberg 2008; Kovács 2009; Szabó 2014). Various proposed models for the Blazhko effect, and principal reasons why they fail to explain observations, are summarized in Kovacs (2016).

A part of the reason for the incomplete observational description of the Blazhko effect is difficulties in discovering a large number of Blazhko stars due to temporal baselines that are too short and insufficient number of observations per object (Kovacs 2016; Hernitschek & Stassun 2022). With the advent of modern sky surveys, several studies reported large increases in the number of known Blazhko stars, starting with a sample of about 700 Blazhko stars discovered by the MACHO survey towards the LMC (Alcock et al. [1]) and about 500 Blazhko stars discovered by the OGLE-II survey towards the Galactic bulge (Mizerski 2003). Most recently, about 4,000 Blazhko stars were discovered in the LMC and SMC (Soszyński et al. 2009, 2010), and an additional $\sim3{,}500$ stars were discovered in the Galactic bulge (Soszyński et al. 2011; Prudil & Skarka 2017), both by the OGLE-III survey. Nevertheless, discovering the Blazhko effect in field RR Lyrae stars that are spread over the entire sky remains a much harder problem: only about 400 Blazhko stars in total (Skarka 2013) from all the studies of field RR Lyrae stars have been reported so far (see also Table 1 in Kovacs 2016).

Corresponding author: Željko Ivezić

[email protected] Here, we report the results of a search for the Blazhko effect in a sample of $\sim3{,}000$ field RR Lyrae stars with LINEAR and ZTF light curve data. A preliminary subsample of about $\sim500$ stars was selected using various light curve statistics, and then 228 stars were confirmed visually as displaying the Blazhko effect. This new sample doubles the number of field RR Lyrae stars that exhibit the Blazhko effect. In § 2 and § 3 we describe our datasets and analysis methodology, and in § 4 we present our analysis results. Our main results are summarized and discussed in § 5.

Data Description and Period Estimation

Analysis of field RR Lyrae stars requires a sensitive time-domain photometric survey over a large sky area. For our starting sample, we used $\sim3{,}000$ field RR Lyrae stars with light curves obtained by the LINEAR asteroid survey. In order to study long-term changes in light curves, we also utilized light curves obtained by the ZTF survey which monitored the sky $\sim15$ years after LINEAR. The combination of LINEAR and ZTF provided a unique opportunity to systematically search for the Blazhko effect in a large number of field RR Lyrae stars over a large time span of two decades.

We first describe each dataset in more detail, and then introduce our analysis methods. All our analysis code, written in Python, is available on GitHub².

LINEAR Dataset

The properties of the LINEAR asteroid survey and its photometric re-calibration based on SDSS data are discussed in Sesar et al. (2011). Briefly, the LINEAR survey covered about 10,000 $\mathrm{deg}^{2}$ of the northern sky in white light (no filters were used, see Fig. 1 in Sesar et al. (2011)), with photometric errors ranging from $\sim0.03$ mag at an equivalent SDSS magnitude of $r = 15$ to 0.20 mag at $r \sim18$. Light curves used in this work include, on average, 270 data points collected between December 2002 and September 2008.

A sample of 7,010 periodic variable stars with $r < 17$ discovered in LINEAR data were robustly classified by Palaversa et al. (2013), including about $\sim3{,}000$ field RR Lyrae stars of both ab and c type, detected to distances of about 30 kpc (Sesar et al. 2013). The sample used in this work contains 2196 ab-type and 745 c-type RR Lyrae, selected using classification labels and the $g i$ color index from Palaversa et al. (2013). The LINEAR light curves, augmented with IDs, equatorial coordinates, and other data, were accessed using the astroML Python module³ (VanderPlas et al. 2012).

ZTF Dataset

The Zwicky Transient Factory (ZTF) is an optical time-domain survey that uses the Palomar 48-inch Schmidt telescope and a camera with 47 $\mathrm{deg}^{2}$ field of view (Bellm et al. [4]). The dataset analyzed here was obtained with SDSS-like $g$, $r$, and $i$ band filters. Light curves for objects in common with the LINEAR RR Lyrae sample typically have smaller random photometric errors than LINEAR light curves because ZTF data are deeper (compared to LINEAR, ZTF data have about 2-3 magnitudes fainter $5\sigma$ depth). ZTF data used in this work were collected between February 2018 and December 2023, on average about 15 years after obtaining LINEAR data. The median number of observations per star for ZTF light curves is $\sim500$.

The ZTF dataset for this project was created by selecting ZTF IDs with matching equatorial coordinates to a corresponding LINEAR ID of an RR Lyrae star. This process used the ztfquery function, which searched the coordinates in the ZTF database within 3 arcsec from the LINEAR position. The resulting sample consisted of 2857 RR Lyrae stars with both LINEAR and ZTF data. The fractions of RRab and RRc type RR Lyrae in this sample, 71% RRab and 29% RRc type, are consistent with results from other surveys (e.g., Sesar et al. 2010; Clementini et al. [10]).

Period Estimation

The first step of our analysis is estimating best-fit periods, separately for LINEAR and ZTF datasets. We used the Lomb-Scargle method (Vanderplas 2015) as implemented in astropy (Astropy Collaboration et al. [2]). The period estimation used 3 Fourier components and a two-step process: an initial best-fit frequency was determined using the autopower frequency grid option and then the power spectrum was recomputed around the initial frequency using an order of magnitude smaller frequency step. In case of ZTF, we estimated period separately for each available passband and adopted their median value. Once the best-fit period was determined, a best-fit model of the phased light curve was computed using 6 Fourier components. Figure 1 shows an example of a star with LINEAR and ZTF light curves, phased light curves, and their best-fit models.

We found excellent agreement between the best-fit periods estimated separately from LINEAR and ZTF light curves. The median of their ratio is unity within $2 \times10^{-6}$ and the robust standard deviation of their ratio is $2 \times10^{-5}$. With a median sample period of 0.56 days, the implied scatter of period difference is about 1.0 sec.

Given on average about 15 years between LINEAR and ZTF data sets, and a typical period of 0.56 days, this time difference corresponds to about 10,000 oscillations. With a fractional period uncertainty of $2 \times10^{-5}$, LINEAR data can predict the phase of ZTF light curve with an uncertainty of 0.2. RR Lyrae light curves may experience phase changes of this magnitude (see e.g., [11] (Szeidl et al. 2011); therefore, each data set must be analyzed separately. On the other hand, amplitude modulation can be detected on time scales as long as 15 years, as discussed in the following section.

We did not try to identify double-mode (RRd) stars because their expected sample fraction is below 1% [10].

Analysis Methodology: Searching for the Blazhko Effect

Given the two sets of light curves from LINEAR and ZTF, we searched for amplitude and phase modulation, either during the 5-6 years of data taking by each survey, or during the average span of 15 years between the two surveys. Starting with a sample of 2857 RR Lyrae stars, we pre-selected a smaller sample that was inspected visually (see below for details). We also required at least 150 LINEAR data points and 150 ZTF data points (for the selected band from which we calculated the period) in analyzed light curves. We used two pre-selection methods that are sensitive to different types of light curve modulation: direct light curve analysis and periodogram analysis, as follows.

Direct Light Curve Analysis

Given statistically correct period, amplitude and light curve shape estimates, as well as data being consistent with reported (presumably Gaussian) uncertainty estimates, the $\chi^2$ per degree of freedom gives a quantitative assessment of the "goodness of fit",

Here, $d_i$ are measured light curve data values at times $t_i$, and with associated uncertainties $\sigma_i$, $m_i$ are best-fit models at times $t_i$, and $N_{\mathrm{dof}}$ is the number of degrees of freedom, essentially the number of data points. In the absence of any light curve modulation, the expected value of $\chi^2_{\mathrm{dof}}$ is unity, with a standard deviation of $\sqrt{2/N_{\mathrm{dof}}}$. If $\chi^2_{\mathrm{dof}}-1$ is many times larger than $\sqrt{2/N_{\mathrm{dof}}}$, it is unlikely that data $d_i$ were generated by the assumed (unchanging) model $m_i$. Of course, $\chi^2_{\mathrm{dof}}$ can also be large due to underestimated measurement

uncertainties $\sigma_i$, or to occasional non-Gaussian measurement error (the so-called outliers).

Therefore, to search for signatures of the Blazhko effect, manifested through statistically unlikely large values of $\chi^2_{\mathrm{dof}}$, we computed $\chi^2_{\mathrm{dof}}$ separately for LINEAR and ZTF data (see Figure 2). Using the two sets of $\chi^2_{\mathrm{dof}}$ values, we algorithmically pre-selected a sample of candidate Blazhko stars for further visual analysis of their light curves. The visual analysis is needed to confirm the expected Blazhko behavior in observed light curves, as well as to identify cases of data problems, such as photometric outliers.

We used a simple scoring algorithm, optimized through trial and error, that utilized the two values of $\chi^2_{\mathrm{dof}}$, augmented by period and amplitude differences, as follows. A star could score a maximum of 9 points, and a minimum of 5 points was required for further visual analysis. The $\chi^2_{\mathrm{dof}}$ selection boundaries are illustrated in Figure 2. If either value of $\chi^2_{\mathrm{dof}}$ exceeded 5, or both exceeded 3, a star was awarded 5 points and immediately selected for further analysis. If these $\chi^2_{\mathrm{dof}}$ criteria were not met, a star could still be selected by meeting less stringent $\chi^2_{\mathrm{dof}}$ selection if it also had large period or amplitude difference between LINEAR and ZTF datasets. Stars with at least one value of $\chi^2_{\mathrm{dof}}$ above 2 would receive 3 points and those with at least one $\chi^2_{\mathrm{dof}}$ above 3 would receive 4 points. A period difference exceeding $2 \times10^{-4}$ day would be awarded 1 point and two points for exceeding $5 \times10^{-4}$ day. Analogous limits for amplitude difference were 0.05 mag and 0.15 mag, respectively.

The candidate Blazhko sample pre-selected using this method includes 531 stars. For most selected stars, the $\chi^2_{\mathrm{dof}}$ values were larger for the ZTF data because the ZTF photometric uncertainties are smaller than for the LINEAR data set. Figure 3 summarizes the selection criteria and the resulting numbers of selected stars for each criterion and their combinations.

Periodogram Analysis

When light curve modulation is due to double-mode oscillation with two similar oscillation frequencies (periods), it is possible to recognize its signature in the periodogram computed as part of the Lomb-Scargle analysis. Depending on various details, such as data sampling and the exact values of periods, amplitudes, this method may be more efficient than direct light curve analysis (Skarka et al. 2020). We also employed this method to select additional candidates, as follows.

A sum of two sine functions with same amplitudes and with frequencies $f_1$ and $f_2$ can be rewritten using trigonometric equalities as

We can define

and

with $\Delta f \ll f_0$ when $f_1$ and $f_2$ are similar. The fact that $\Delta f$ is much smaller than $f_0$ means that the period of the $\cos$ term is much larger than the period of the basic oscillation ($f_0$). In other words, the $\cos$ term acts as a slow amplitude modulation of the basic oscillation. When the amplitudes of two $\sin$e functions are not equal, the results are more complicated but the basic conclusion about amplitude modulation remains. When the power spectrum of $y(t)$ is constructed, it will show 3 peaks: the main peak at $f_0$ and two more peaks at frequencies $f_0 \pm\Delta f$. We used this fact to construct an algorithm for automated searching for the evidence of amplitude modulation. Figure 4 compares the theoretical periodogram produced by interference beats with our algorithm’s periodogram, signifying that local Blazhko peaks are present in real data.

After the strongest peak in the Lomb-Scargle periodogram is found at frequency $f_0$, we search for two equally distant local peaks at frequencies $f_-$ and $f_+$, with $f_- < f_0 < f_+$. The sideband peaks can be highly asymmetric ([1]) and observed periodograms can sometimes be much more complex ((Szczygieł & Fabrycky 2007)). We fold the periodogram through the main peak at $f_0$, multiply the two branches and then search for the strongest peaks in the resulting folded periodogram that is statistically more significant than the background noise. The multiplication of the two branches allows for efficient detection of moderately asymmetric peaks. However, this method is not sensitive to the subset of modulated stars where only one sidepeak is detectable ((Netzel et al. 2018; Molnár et al. 2023)). The background noise is computed as the scatter of the folded periodogram estimated from the interquartile range. We require a “signal-to-noise” ratio of at least 5, as well as the peak strength of at least 0.05 for ZTF, while 0.10 for LINEAR data. If such a peak is found, and it doesn’t correspond to yearly alias, we select the star as a candidate Blazhko star and compute its Blazhko period as

where $f_{-,+}$ means the Blazhko sideband frequency with a higher amplitude is chosen. The observed Blazhko periods range from 3 to 3,000 days, and Blazhko amplitudes range from 0.01 mag to about 0.3 mag ((Szczygieł & Fabrycky 2007)). In this work, we selected a smaller Blazhko range due to the limitations of our data: 30–325 days. With this additional constraint, we selected 52 candidate Blazhko stars. Figure 4 shows an example where two very prominent peaks were identified in the LINEAR periodogram.

Visual Confirmation

The sample pre-selected for visual analysis includes 531 RR Lyrae stars (479 + 52), or 18.1% of the starting LINEAR-ZTF sample. Visual analysis included the following standard steps (e.g., (Jurcsik et al. 2009; Prudil & Skarka 2017)):

1. The shape of the phased light curves and scatter of data points around the best-fit model were examined for signatures of anomalous behavior indicative of the Blazhko effect. Figure 6 shows an example of such behavior where the ZTF data and fit show multiple coherent data point sequences offset from the best-fit mean model.

2. Full light curves were inspected for their repeatability between observing seasons (Figure 7). This step was sensitive to amplitude modulations with periods of the order a year or longer.

3. The phased light curves normalized to unit amplitude were inspected for their repeatability between observing seasons. This step was sensitive to phase modulations of a few percent or larger on time scales of the order a year or longer. Figure 8 shows an example of a Blazhko star where season-to-season phase and amplitude modulations are seen in both the LINEAR data and (especially) the ZTF data. Another example is shown in Figure 9 where only phase modulation is visible, without any discernible amplitude modulation⁴.

After visually analyzing the starting sample of 531 Blazhko candidates, we visually confirmed expected Blazhko behavior for 228 stars (214 out of 479 and 14 out of 52). LINEAR IDs and other characteristics for confirmed Blazhko stars are listed in Table 1 (Appendix A). Statistical properties of the selected sample of Blazhko stars are discussed in detail in the next section.

Results

Starting with a sample of 2857 field RR Lyrae stars with both LINEAR and ZTF data, we constructed a subsample of 1996 with light curves of sufficient quality and selected and verified 228 stars that exhibit convincing Blazhko effect. In this section we compare various statistical properties of selected Blazhko stars to those of the starting sample.

The Blazhko Incidence Rate

The implied incidence rate for the Blazhko effect is 11.4±0.8%. Due to selection effects and unknown completeness, this rate should be considered as a lower limit. When ab and c types are considered separately, the rate is slightly higher for the former than for the latter: 12.1±0.9% vs. 9.2±1.3%. The difference of 2.9% has low statistical significance (< 2σ).

Period, Amplitude and Magnitude Distributions

Marginal distributions of period, amplitude and apparent magnitude for the starting sample and Blazhko stars are compared in Figure 10. Encouragingly, their magnitude distributions are statistically indistinguishable which indicates that the completeness is not a strong function of the photometric signal-to-noise ratio. This result is probably due to the fact that the sample is defined by the depth of LINEAR survey, while ZTF survey is deeper than this limit and its photometric quality is approximately constant across the probed magnitude range.

The suspected differences in amplitude and period distributions are further explored in Figure 11. It is already discernible by eye that the period distribution for Blazhko stars of ab type is shifted to smaller values than for the starting sample. We have found that the median period for ab type Blazhko stars is about 5% shorter than for the starting RR Lyrae sample. This difference is significant at the 7.1σ level. At the same time, the difference in median amplitudes for ab type stars corresponds to only 0.6σ deviation. No statistically significant differences are found in period and amplitude distributions for c type stars.

If modulation amplitudes are correlated with periods such that larger modulation amplitudes occur in shorter period RRab stars, and if our selection efficiency is lower for smaller modulation amplitudes, then the detected period shift for ab type Blazhko stars might be at least partially due to combination of these two effects. This possibility does not appear likely. First, as we discussed in preceding section, our sample is defined by the depth of LINEAR survey, while ZTF survey is significantly deeper than this limit and its photometric quality is approximately constant across the probed magnitude range. Since it is sufficient for a star to display the Blazhko effect only in ZTF to be included in the sample, we do not expect strong selection effects (except for the LINEAR magnitude cutoff of course). Furthermore, Skarka et al. (2020) searched for period - modulation amplitude correlation using a large sample of stars with OGLE measurements and did not find any.

Long-term behavior of Blazhko Stars

During visual analysis, we noticed that some Blazhko stars exhibit convincing Blazhko effect either in LINEAR or in ZTF data, but not in both surveys. Figure 12 shows an example where amplitude modulation is clearly seen in LINEAR light curves, while not discernible in ZTF light curves. There are also examples of stars where Blazhko effect is evident in ZTF but not in LINEAR data (e.g., LINEARid = 19466437, 14155360). This finding strongly suggests that Blazhko effect can appear and disappear on time scales shorter than about a decade.

Discussion and Conclusions

We found excellent agreement between the best-fit periods for RR Lyrae stars estimated separately from LINEAR and ZTF light curves. Only one star in our sample (CT CrB, LINEARid=17919686), was previously reported as a Blazhko star (Skarka 2013). The sample of 228 stars presented here increases the number of field RR Lyrae stars displaying the Blazhko effect by more than 50% and places a lower limit of ($11.4 \pm0.8$)% for their incidence rate. The reported incidence rates for the Blazhko effect range from 5% (Szczygieł & Fabrycky 2007) to 60% (Szabó et al. 2014). Differences in reported incidence rates can occur due to varying data precision, the temporal baseline length, and differences in visual or algorithmic analysis. For a relatively small sample of 151 stars with Kepler data, a claim has been made that essentially every RR Lyrae star exhibits modulated light curve (Kovacs 2018). The difference in Blazhko incidence rates for the two most extensive samples, obtained by the OGLE-III survey for the Large Magellanic Cloud (LMC, 20% out of 17,693 stars; Soszyński et al. 2009) and the Galactic bulge (30% out of 11,756 stars; Soszyński et al. 2011) indicates a possible variation of

the Blazhko incidence rate with underlying stellar population properties. We find that ab type RR Lyrae which show the Blazhko effect have about 5% (0.030 day) shorter periods than starting sample. While not large, the statistical significance of this difference is 7.1$\sigma$. At a similar uncertainty level ($\sim1\%$), we don’t detect period difference for c type stars, and don’t detect any difference in amplitude distributions. We also find that for some stars the Blazhko effect is discernible in only one dataset. This finding strongly suggests that Blazhko effect can appear and disappear on time scales shorter than about a decade, in agreement with literature ([5] (Jurcsik et al. 2009; Poretti et al. 2010)). The LINEAR and ZTF datasets analyzed in this work were sufficiently large that we had to rely on algorithmic pruning of the initial sample. The sample size problem will be even larger for surveys such as the Legacy Survey of Space and Time (LSST; (Ivezić et al. 2019)). LSST will be an excellent survey for studying Blazhko effect ((Hernitschek & Stassun 2022)) because it will have both a long temporal baseline (10 years) and a large number of observations per object (nominally 825; LSST Science Requirements Document⁵). We anticipate a higher fraction of discovered Blazhko stars with LSST than reported here due to better sampling and superior photometric quality, since the incidence rate of the Blazhko effect increases with sensitivity to small-amplitude modulation, and thus with photometric data quality ((Jurcsik et al. 2009)).

The size and quality of LSST sample will motivate further developments of the selection algorithms. One obvious improvement will be inspection of neighboring objects to confirm photometric quality, as well as inspection of images to test implication of an isolated point source (e.g., blended object photometry can be affected by variable seeing beyond aperture correction valid for isolated point sources). Another improvement is forward modeling of the Blazhko modulation, rather than searching for $\chi^2$ outliers ([6] (Guggenberger et al. 2012)). For example, Skarka et al. (2020) classified Blazhko stars in 6 classes using the morphology of their amplitude modulation (the most dominant class includes 90% of the sample). They also found bimodal distribution of Blazko periods, with two components centered on 48 d and 186 d. These results give hope that forward modeling of the Blazhko effect will improve the selection of such stars.

Acknowledgments

We thank Mathew Graham for providing ztfquery code example to us, and Robert Szabó for expert comments that improved presentation.

Ž.I. acknowledges funding by the Fulbright Foundation and thanks the Ruder Bošković Institute (Zagreb, Croatia) for hospitality.

This material is based on work supported in part by the National Science Foundation through Cooperative Agreement 1258333 managed by the Association of Universities for Research in Astronomy (AURA), and the Department of Energy under Contract No. DE-AC02-76SF00515 with the SLAC National Accelerator Laboratory. Additional LSST funding comes from private donations, grants to universities, and in-kind support from LSSTC Institutional Members. This research has made use of NASA’s Astrophysics Data System Bibliographic Services.

Based on observations obtained with the Samuel Oschin Telescope 48-inch and the 60-inch Telescope at the Palomar Observatory as part of the Zwicky Transient Facility project. ZTF is supported by the National Science Foundation under Grants No. AST-1440341 and AST-2034437 and a collaboration including current partners Caltech, IPAC, the Weizmann Institute of Science, the Oskar Klein Center at Stockholm University, the University of Maryland, Deutsches Elektronen-Synchrotron and Humboldt University, the TANGO Consortium of Taiwan, the University of Wisconsin at Milwaukee, Trinity College Dublin, Lawrence Livermore National Laboratories, IN2P3, University of Warwick, Ruhr University Bochum, Northwestern University and former partners the University of Washington, Los Alamos National Laboratories, and Lawrence Berkeley National Laboratories. Operations are conducted by COO, IPAC, and UW.

The LINEAR program is funded by the National Aeronautics and Space Administration at MIT Lincoln Laboratory under Air Force Contract FA8721-05-C-0002. Opinions, interpretations, conclusions and recommendations are those of the authors and are not necessarily endorsed by the United States Government.

Software: Astropy ([2, 3]), Matplotlib ((Hunter 2007)), SciPy ((Virtanen et al. 2020)), astroML ((VanderPlas et al. 2012))

Appendix

periodograms (Z, L or “-” for no detection; the strength and period of Blazhko peaks are not reliable when “-” ) and the selection score (see § 3.1 and § 3.2 for details). The full table is available in online edition.

LID	$P_L$	$P_Z$	$N_L$	$N_Z$	$\chi^2_{L,r}$	$\chi^2_{Z,r}$	$\chi^2_L$	$\chi^2_Z$	$A_L$	$A_Z$	$\delta A$	$B_{pL}$	$B_{pZ}$	$B_{pL}$	$B_{pZ}$	$t$	$f$	$B_s$	$B_f$
158779	0.609207	0.609189	293	616	1.6	3.9	3.7	34.2	0.47	0.68	0.21	1.6443	1.6444	352.7337	350.2	1	-	7	1
263541	0.558218	0.558221	270	503	2.9	6.6	15.8	110.4	0.64	0.82	0.18	1.8621	1.8025	14.1513	89.9	1	-	7	1
393084	0.530027	0.530033	493	372	1.1	3.2	1.6	19.2	0.96	1.31	0.35	1.9447	1.8896	17.2369	347.2	1	-	6	1
810169	0.465185	0.465212	289	743	2.1	2.8	6.0	15.1	0.77	0.75	0.02	2.2232	2.2230	13.6017	13.6	1	-	5	1
924301	0.507503	0.507440	418	189	1.9	9.3	13.8	162.9	0.87	0.79	0.08	2.0043	1.9763	29.5072	178.4	1	-	8	1
970326	0.592233	0.592231	275	552	1.1	2.1	1.9	7.7	0.51	0.75	0.24	1.7563	1.6992	14.7656	93.2	1	-	5	1
999528	0.658401	0.658407	564	213	1.2	2.7	1.8	21.7	0.57	0.92	0.35	1.5527	1.5510	29.5247	31.0	1	-	5	1
1005497	0.653607	0.653605	607	192	1.1	2.1	2.1	12.4	0.60	0.83	0.23	1.5639	1.5481	29.4638	55.1	1	-	5	1
1092244	0.649496	0.649558	590	326	1.2	3.6	2.3	32.1	0.72	0.58	0.14	1.5735	1.5640	29.5421	40.8	1	-	7	1
1240665	0.632528	0.632522	468	311	3.0	1.1	25.2	1.6	0.33	0.33	0.00	1.6149	1.5865	29.4942	182.3	1	Z	0	2