NICMOS Count Rate Dependent Non-Linearity in ... - Semantic Scholar

9 downloads 0 Views 1MB Size Report
Observations of the star P041C on high background from an internal flat field ... The count rate dependent non-linearity discovered by Bohlin, Lindler, & Riess.
Instrument Science Report NICMOS 2006-002

NICMOS Count Rate Dependent Non-Linearity in G096 and G141 Ralph C. Bohlin, Adam Riess, & Roelof de Jong January 2006 ABSTRACT

Since the discovery of the NICMOS count rate dependent non-linearity documented in NICMOS ISR 2005-002, additional tests have been conducted to further understand and quantify the effect. Long integrations of up to 25 minutes exhibited the same level of non-linearity as the original 1-3 minute integrations. Observations of the star P041C on high background from an internal flat field lamp provides another measure of the wavelength dependence of the nonlinearity to compliment the measures based on pure hydrogen WD models.

1. Introduction

The count rate dependent non-linearity discovered by Bohlin, Lindler, & Riess (2005, BLR) is characterized in more detail with new NICMOS grism data. In Section 2, integration times of ~25 min that are typical of science observations are compared with the shorter times of 1-3 min used by BLR. Section 3 is an evaluation of the new data where the number of dither steps on the detector is much less than the 15 steps used by BLR. Section 4 details the revised flat fielding strategy, while Section 5 includes the new P041C data in a re-analysis 1

to update the BLR correction algorithm. Section 6 dispels worries that persistence effects from previous observations can seriously affect the grism data considered here. Section 7 presents typical results on the accuracy of the final NICMOS spectrophotometric fluxes. In a companion ISR, de Jong, et al. (2006, JBRB) present a similar study of the non-linearity effects for NICMOS photometry. 2. Dependence of Count Rate on Integration Time As summarized in Table 1, three stars were re-observed with integration times of several minutes in order to better represent typical science observations of faint targets. To fit within the total time available in each orbit, the number of dither positions were reduced from the BLR standard of 15 positions. However, the total range of the dithers in Y, perpendicular to the dispersion axis, was maintained at 15 pixels, i.e. 3arcsec for the 0.2arcsec/px plate scale of NIC3. Table 1. Observations with Long Integrations and Fewer Dither Positions Star GD153 WD1057+719 WD1657+343

Rootname n94a02 n9g501 n97u54

Obs-Date 2005Jul26 2005Jun13 2005Jul25

Dithers 3 2 4

Integ(min) 7.5 25.6 23.5

Normally, the grism spectroscopy from the NIC3 camera is derived from the “CAL” files from the STScI NICMOS pipeline (see BLR). These CAL files represent the best final values of the measurements in the 256x256 imaging format, as corrected for cosmic ray hits. Also provided by the STScI NICMOS pipeline are the intermediate readout IMA images that are the count rates for each non-destructive read. Analysis of the IMA files reveals the evolution of the differential count rates up to the full integration period. These average differential count rates in each time step in the exposure sequence can be derived from the difference in total counts in the adjacent images in the IMA file. For the 0.8-1.2µm range of G096 where the non-linearity of BLR is the 2

largest, Figures 1-2 show the differential net count rates as a function of cumulative integration time for the two brighter stars. WD1647+343 is so faint that the weak, un-removed cosmic ray hits cause too much noise for its analogous plot to be meaningful. The gross signal is the total response of all pixels centered within one pixel of the spectral trace center-line. Because the NICMOS detectors are subject to amplifier drift during an observation, the net signal is defined by subtracting from the gross a background that is the median signal within a strip lying 20 to 50 pixels below the spectrum.

Figure 1 - Count rate in each step of the sample sequence as a function of cumulative integration time of the intermediate non-destructive readouts for WD1057+719. Solid line and squares - 25.6 minute integration at two dither positions per Table 1. Dotted and triangles - Original BLR observation n8vj04 with 2.9min integrations at each of 15 dither positions. The rate at each point is normalized to the average count rate of the last four

3

integrations.

Figures 1-2 demonstrate a scatter in the count rates that is larger for the first few reads with small time steps. The fainter star WD1057+719 has more scatter than the brighter GD153. The typical count rates converge within 1-2% of their average after one minute of integration. The occasional high point can be caused by cosmic ray hits, which are not removed.

Figure 2 - As in Figure 1 for GD153. The original observation n8u402 (triangles) has one min integrations at each of 15 dither positions.

3. Spectrophotometric Repeatability Figure 3 shows the broad and narrow band spectrophotometric repeatability for the repeated observations of the five white dwarf (WD) stars, the two solar 4

analogs (P041C and P330E), and BD+17°4708. The journal of observations for these seven stars appears in Tables 1-2. The WD stars include the three primary standards G191B2B, GD71, and GD153 that determine the absolute sensitivity of the grism modes (BLR), while the fainter two WD stars WD1057+719 and WD1657+343 also have pure hydrogen NLTE model atmosphere calculations. Residuals for the WD stars are computed relative to these models. (See Section 5 for more discussion). Residuals for the other three stars are relative to the first NICMOS observation of each star, where these residuals are reduced by 1.414 for comparison with the WD residuals computed relative to the noise free models. Table 2. Repeat Observations with 15 Dither Positions Star

Rootname

BD+17°4708 n8vj07 BD+17°4708 n94701 BD+17°4708 n94702 G191B2B n8u405 G191B2B n94a03 GD71 n8u401 GD71 n94a01 GD153 n8u402 P041C n8u403 P041C n9jj02-bef P041C n9jj02-aft P330E n8br01 P330E n8u406 WD1057+719 n8vj04 WD1657+343 n8vj03

Obs-Date

Integ(s)

2004Jan10 2004Nov15 2004Nov25 2003Sep6 2005Mar1 2004Jan19 2005Feb28 2004Jun9 2004Jan19 2005Nov12 2005Nov12 2002May27 2004Jun19 2004Feb2 2004Jun6

32 10 10 104 104 56 64 56 72 12* 12* 56 96 176 152

* Before and after the lamp-on observations discussed in Section 5. Both have 5 dither positions.

Except for the faintest star WD1657+343, the excellent NICMOS photometric repeatability is demonstrated by average residuals of 1.5% for both G096 and G141. Figure 4 illustrates the G096 residuals for WD1057+719 along with GD153, which has a three dither observation. The noise in the two-dither WD1057+719 data is much worse than the corresponding 15-dither observation, while the scatter of the three- and 15-dither GD153 spectra are comparable. The region below 0.92µm generally shows excess scatter because of the strong, sharp features in the spectra, eg. GD153 shown in Figure 5. These features arise in the interference filter used to delimit the bandpass and are under sampled, i.e. there are large changes across the wavelength coverage of a single pixel. Any actual spectral features from 0.8 to 0.92 are

Suggest Documents