Histogram of the reduced chi-squared for the observed photometric deviations about the mean for the J (top), H (middle), and K-band photometry from the 16 nights in which all 7 tiles were observed. The solid curve in each panel is the expected reduced chi-squared distribution for 15 degrees of freedom, where the curves have been normalized by the total number of stars. This figure demonstrates that the observed photometric scatter for the majority of the stars is consistent with random noise. Stars with large values of reduced chi-squared are candidate variable stars.
The observed photometric RMS in the time series data as a function of magnitude for stars brighter than the defined completeness limits. The observed RMS ranges from about 0.015 mag for the brightest stars to < 0.15 magnitudes (i.e. signal to noise ratio >= 7) for stars at the completeness limit.
The Stetson variable index (S) plotted as a function of the H magnitude for stars brighter than H=15.4. Only data for stars in the 16 nights common to all seven tiles were used to compute the variability index. The dashed line at S=0 shows the expected value of the variability index for non-variable stars, and the dotted line at $S=0.55$ represents the minimum adopted value used to identify variable stars in this study. The origin of the positive bias in the computed index values is unknown, and suggests that a weak correlation exists between the J, H, and K-band photometry, possibly from the fact that the three bands were observed at the same time. Note that 53 stars with S > 5.0 are not shown for clarity.
Photometric data for star 5123, which has a Stetson variability index (S=0.58) that is just above the limit (0.55) to be classified as a variable. The left panels show the J, H, and K light curves. The first data point in each light curve is from March 1998, the second data point from February 2000, and the remaining photometry from March/April 2000. The vertical bars through the data points represent the ± 1 sigma photometric uncertainties. The right panels show the K vs. H-K color-magnitude diagram and the J-H vs. H-K color-color diagrams for each data point in the time series, where the dotted line represents the interstellar reddening vector from Cohen etal. (1981) transformed into the 2MASS photometric system (Carpenter 2001). The solid line in the color-magnitude diagrams is the 1 Myr pre-main-sequence isochrone from D'Antona & Mazzitelli (1997) for stellar masses between 0.08 Mo and 3 Mo. The solid curves in the color-color diagram are the loci of red giant and main-sequence stars from Bessell & Brett (1988) in the 2MASS color system. Postage stamps of the 2MASS images from March 15, 2000 are shown on the far right. The summary information shown for the star was computed using all available photometry from March 1998-April 2000.
Photometric data for star 3730, classified as a periodic star based upon the Lomb-Scargle periodogram analysis. The derived period in each band is about 8 days.
Photometric data for star 8783 (also known as BM Ori), classified as an eclipsing system based on the simultaneous drops in the J, H, and K band magnitudes on 3 discrete days. This eclipsing system was identified previously from optical observations (see, e.g., Antokhina, Ismailov, & Cherepashchuk 1989 and references therein).
Photometric data for star 4067, an example of a star that steadily increased in brightness in the March/April~2000 time period. The March~1998 photometry though indicates that this has not been a long term trend.
Photometric data for star 5707 (also known as YY Ori), an example of a star in which the stellar colors get bluer as the star gets fainter. The sense of the color-magnitude changes are opposite of that expected from either rotational modulation by hot spots or extinction variations, but are consistent with a model in which the geometry or mass accretion rate in a circumstellar disk changes in time.
Photometric data for star 11926 (also known as AO Ori), an example of a star where the stellar colors get redder as the star gets fainter. Qualitatively the photometric fluctuations are consistent with either the presence of time variable hot spots or variations in the amount of extinction.
Photometric data for star 1048, another example of a star in which the stellar colors get redder as the star gets fainter (see also Fig. 9). In this instance, the stellar magnitudes are relatively constant for the first two weeks of the time series before the star becomes fainter by over 1 mag in each band with progressively redder colors over a period of a couple of days. As the star faded in brightness, a near-infrared excess become apparent for 2-3 days.
Photometric data for star 13688 (also known as AW Ori), another example of a star in which the stellar colors get redder as the star gets fainter (see also Figs. 10 and 11). In this instance, the star exhibits quasi-periodic fluctuations, and the colors and magnitudes vary along a vector that is shallower in the color-color diagram than prior examples and steeper in the color-magnitude diagrams.
Photometric data for star 10527, an example of a star that is not variable in the March/April 2000 time frame, but exhibits longer term photometric fluctuations in both the March 1998 and February 2000 observations.
Photometric data for star 5841 (also known has JW 101 and V1314 Ori), a second example of a star that exhibits long term photometric variability relative to the March/April 2000 time series data (see also Fig. 12). In this case, the star is identified as a variable in the March/April 2000 data, but exhibits even larger fluctuations in the March 1998 data. The long term variability is such that the star got fainter at J-band while simultaneously getting brighter at K-band.
Spatial distribution of stars and molecular gas toward the Orion Nebula Cluster. Starting with the left-most panel, these figures show the (a) spatial distribution of stars with K <= 14.8; (b) surface density of stars with K <= 14.8, where the surface density map was created by convolving the stellar spatial distribution shown in (a) with a gaussian kernel of size sigma=60"; (c) spatial distribution of variable stars from Sample~1 (see Table 3); (d) distribution of variable stars shown in (c) that have a near-infrared excess in the J-H vs. H-K color-color diagram; (e) distribution of H alpha emitting stars from the Kiso H alpha survey with a Kiso class of 3, 4, or 5 (Wiramihardja etal 1991); (f) contour map of the integrated \thcoj\ emission (T_R dv) from Bally etal. (1987). The contours levels are 1, 5, 10, 20, 30, 40, and 50 K km/s. These panels indicate that the variable stellar population follow the large scale spatial distribution of the ONC as traced by the total K-band star counts and H alpha emitting objects.
Histogram of the \KB-band magnitudes for the ONC and the variable star population. The field star contribution to the ONC has been subtracted from the observed star counts using the procedure described in the Appendix applied to differential magnitude intervals. The completeness limit of the observations is K=14.8. This figure indicates that the variable star population identified with these observations tend to be the brighter stars in the cluster. The lack of faint variable stars is likely a result of increased photometric noise at these magnitudes that masks any low amplitude photometric fluctuations.
K vs. H-K color-magnitude diagram for all stars (color scale) and the variable stars (black circles) identified from the 16 nights in which all tiles were observed. For reference, the solid blue curve shows the 1 Myr pre-main-sequence isochrone from D'Antona & Mazzitelli(1997) for stellar masses between 0.08 Mo and 3.0 Mo. The dashed lines indicate the reddening vector for 10 magnitudes of visual extinction from Cohen etal. (1981) transformed into the 2MASS photometric system (Carpenter 2001). The lowest halftone is 1\% of the peak density. This figure shows that the observed magnitudes and colors for the majority of the variable population is consistent with reddened pre-main-sequence stars with masses <= 3 Mo.
J-H vs. H-K color-color diagram for all stars (left panel) and the variable stars (right panel) identified from the 16 nights in which all tiles were observed. The lowest halftone in each panel begin at 1% of the peak density. The stars represented in the left panel are dominated by field stars unrelated to the ONC. The variable stars are on average redder than the field star population, and about 33% have colors indicating the presence of a near-infrared excess.
Histograms of the variable star peak-to-peak (solid histogram) and RMS (dotted histogram) amplitudes in the J, H, and K band data after correcting the observed amplitudes for photometric noise (see text). The top panels show the histograms over the full dynamic range, and the bottom panels show in more detail the distribution at low amplitudes which contain most of the variables. The amplitudes were computed using all measurements in the March/April 2000 time series.
Similar to Figure 18, except for the J-K, J-H, and H-K colors.
Histograms of the derived slopes in the J-H vs. H-K, J vs. J-H, and K vs. H-K diagrams. Only variable stars in which the observed RMS of the appropriate colors/magnitudes exceeded the expected RMS by a factor of 1.5 are shown. The open histogram is for all stars that meet these criteria, and the hatched histogram is those stars in which the slope have been determined to an accuracy of better than 20%. The predicted slopes based on hot spot models, extinction variations, and circumstellar disks models (see text) are indicated. This figure shows that while each of these models can account for some aspect of the variability amongst these stars, none alone can account for all of the observed trends.
J-H vs. H-K diagrams for three groups of variable stars in the ONC. The left panel shows variable stars that have significant fluctuations in both the observed K magnitudes and H-K colors, and the photometric fluctuations are such that the colors become redder as the star gets fainter (i.e. slope in the K vs H-K diagram is positive). The middle panel shows stars that have negative slopes such that the colors become bluer as the stars gets fainter. The right panel is the color-color diagram for stars that are identified as periodic variables in the near-infrared. Stars with significant color and magnitude variations are generally redder than the periodic stars, and in particular, stars with positive slope variations tend to have near-infrared excesses more so than periodic variables.
Distribution of time lags inferred from the autocorrelation function for the variables stars in the March/April 2000 time series data. The three solid lines represent the time lags for the J, H, and K band data. The dotted line shows the time lags for a simulated data set with the same random noise characteristics and time sampling as the observations. The maximum time lag possible in this ACF analysis is approximately half the time period of the observations. For 77% of the variable stars, the maximum possible time lag is about 14 days. This figure shows that in the about 1 month time series observations, most of the variability occurs of time scales of a few days, but can be as long as 10-15 days in the more extreme cases.
Comparison of the optical, I-band periods (Stassun etal 1998; Herbst et al. 2000; Rebull 2001) with the H-band periods derived in this study. For about 80% of the stars, the optical and near-infrared periods agree to better than 10%. Stars with periods less than about 2 days as indicated by the optical data are aliased to longer periods in this study due to the 1 day time sampling of the near-infrared observations. Stars that have optical periods roughly twice that of the near-infrared period may be examples of ``period doubling'' which results when multiple star spots are present \citep{Herbst01}. One star with an optical period of about 60 days and a near-infrared period of about 14 days is not shown in this figure since the optical period is uncertain (Rebull 2001).
Frequency distribution of periods for stars in this study that have false-alarm-probabilities less than 10^-4. The open histogram is for all the stars, and the hatched histogram are stars which are suspected to be aliased with a sub-2 day period based on comparison to the optical derived periods (see Fig. 23). Only about half the stars represented by the open histogram have the necessary information to establish if the inferred period is aliased in this manner.
Model J vs. J-H, K vs. H-K, and J-H vs. H-K diagrams for cool and hot spots, extinction, and accretion disk variations. These models can be compared with the observed diagrams in Figures 4-13 to investigate the origin of the near-infrared variability. The models assume an photospheric temperature of 4000 K, appropriate for a 1 Myr, about 0.5 Mo star (D'Antona & Mazzitelli 1997), spot temperatures of 2000 K for cool spots (asterisk symbols) or 8000 K for hot spots (open circles), and spot coverages of 1, 2, 5, 10, 20, and 30% (see Eq. 7). The extinction vectors were calculated from the interstellar reddening law from Cohen etal. (1981) transformed into the 2MASS color system (Carpenter 2001). The length of the vectors correspond to Delta(Av) = ± 2 mag. The disk models have been provided courtesy of N. Calvet, and represent the effects of varying the mass accretion rate and inner hole size of the accretion disks. Results are presented for mass accretion rates of 10^-8.5 Mo/year (open triangles) and 10^-7.0 Mo/year (filled triangles), and for each accretion rate, inner hole sizes of 1, 2, and 4 Ro. The larger hole sizes correspond to small infrared excesses.
J-H vs. H-K diagrams for variable stars as a function of the J-band peak-to-peak amplitude. This figure shows that stars with larger J-band amplitudes tend to have redder colors and larger near-infrared excesses than stars with smaller fluctuations.
Spatial distribution of stars in differential K magnitude intervals. This figure demonstrates that the stellar population associated with the Orion~A molecular cloud is prominent only for K < 14. At fainter magnitudes, the star counts are dominated by the field star population.
