1: %\documentstyle[12pt,aaspp4]{article}

2: \documentstyle[12pt,preprint]{aastex}

3: 

4: %\input epsf

5: 

6: %\tighten

7: %\received{2003 December 12}

8: 

9: \begin{document}

10: 

11: %\def\plotone#1{\centering \leavevmode

12: %\epsfxsize=\columnwidth \epsfbox{#1}}

13: 

14: \def\wisk#1{\ifmmode{#1}\else{$#1$}\fi}

15: 

16: \def\lt     {\wisk{<}}

17: \def\gt     {\wisk{>}}

18: \def\le     {\wisk{_<\atop^=}}

19: \def\ge     {\wisk{_>\atop^=}}

20: \def\lsim   {\wisk{_<\atop^{\sim}}}

21: \def\gsim   {\wisk{_>\atop^{\sim}}}

22: \def\kms    {\wisk{{\rm ~km~s^{-1}}}}

23: \def\Lsun   {\wisk{{\rm L_\odot}}}

24: \def\Zsun   {\wisk{{\rm Z_\odot}}}

25: \def\Msun   {\wisk{{\rm M_\odot}}}

26: \def\um     {$\mu$m}

27: \def\mic     {\mu{\rm m}}

28: \def\sig    {\wisk{\sigma}}

29: \def\etal   {{\sl et~al.\ }}

30: \def\eg     {{\it e.g.\ }}

31:  \def\ie     {{\it i.e.\ }}

32: \def\bsl    {\wisk{\backslash}}

33: \def\by     {\wisk{\times}}

34: \def\half {\wisk{\frac{1}{2}}}

35: \def\third {\wisk{\frac{1}{3}}}

36: \def\nwm2sr {\wisk{\rm nW/m^2/sr\ }}

37: \def\nw2m4sr {\wisk{\rm nW^2/m^4/sr\ }}

38: 

39: \title{New measurements of cosmic infrared background fluctuations

40: from early epochs.}

41: 

42: \author{

43: A. Kashlinsky\altaffilmark{1,2,4}, R. G. Arendt\altaffilmark{1,2},

44: J. Mather \altaffilmark{1,3}, S. H. Moseley \altaffilmark{1,3} }

45: \altaffiltext{1}{Observational Cosmology Laboratory, Code 665,

46: Goddard Space Flight Center, Greenbelt MD 20771}

47: \altaffiltext{2}{SSAI} \altaffiltext{3}{NASA}

48: \altaffiltext{4}{e--mail: kashlinsky@stars.gsfc.nasa.gov}

49: 

50: \begin{abstract}

51: Cosmic infrared background fluctuations may contain measurable

52: contribution from objects inaccessible to current telescopic

53: studies, such as the first stars and other luminous objects in the

54: first Gyr of the Universe's evolution. In an attempt to uncover

55: this contribution we have analyzed the GOODS data obtained with

56: the Spitzer IRAC instrument, which are deeper and cover larger

57: scales than the Spitzer data we have previously analyzed. Here we

58: report these new measurements of the cosmic infrared background

59: (CIB) fluctuations remaining after removing cosmic sources to

60: fainter levels than before. The remaining anisotropies on scales

61: $\sim\!0.5^\prime\!-\!10^\prime$ have a significant clustering

62: component with a low shot-noise contribution. We show that these

63: fluctuations cannot be accounted for by instrumental effects, nor

64: by the Solar system and Galactic foreground emissions and must

65: arise from extragalactic sources.

66: \end{abstract}

67: \keywords{cosmology: observations - diffuse radiation - early

68: Universe}

69: 

70: 

71: \section{Introduction}

72: 

73: Cosmic infrared background (CIB) contains emissions from all

74: luminous including those inaccessible to current telescopic

75: studies (see Kashlinsky 2005 for recent review). If the first

76: stars (commonly called Population III) were massive

77: \cite{abell,bromm,brommlarson}, they could produce significant

78: levels of the CIB with measurable fluctuations

79: \cite{santos,sf2003,kagmm,cooray,madausilk,komatsu}. Recently, in

80: an attempt to uncover these signatures of early stars, we have

81: detected significant CIB fluctuations after removing galaxies to

82: faint limits in deep Spitzer data (Kashlinsky et al 2005;

83: hereafter KAMM). The measured fluctuations cannot arise from the

84: Solar System and Galactic foreground emissions, or from the

85: remaining galaxies and must come from early extragalactic sources,

86: particularly the first stars. To solidify these findings we have

87: now analyzed deeper datasets from the GOODS survey \cite{goods}

88: covering different and larger areas of the sky and with observing

89: procedures which allow us to test for a wide range of systematic

90: errors in the data.

91: 

92: In this {\it Letter} we present the new measurements using the

93: GOODS Spitzer/IRAC data at 3.6, 4.5, 5.8 and 8 \um.  At the

94: magnitude limit of the QSO1700 data we measure a similar CIB

95: signal consistent with its cosmological origin. After removing

96: still fainter sources, we find that the bulk of the CIB

97: fluctuations remain out to $\sim$10$^\prime$, the limit of our

98: fields. This shows that the signal originates in populations

99: significantly fainter than our cutoff. We also test for possible

100: systematic errors in the fluctuations by cross-correlating the

101: data for the same regions taken at different epochs and in

102: different detector orientations, showing only a small contribution

103: from the instrument systematics. The Galactic and zodiacal

104: foregrounds are generally also small, although as in the QSO1700

105: data, we find a possibility of a significant cirrus pollution of

106: the Channel 4 fluctuations. Cosmological interpretation of the

107: results is given in a separate paper (Kashlinsky, Arendt, Mather

108: \& Moseley 2006).

109: 

110: \section{Data processing}

111: 

112: Table 1 shows parameters of the data used in this and earlier

113: (KAMM) analyses. The new data from the GOODS Spitzer Legacy

114: Program%\footnote{

115: (PID=194, pipeline version=S11.4.0, except S11.0.2 for CDFS Epoch

116: 1, Dickinson et al 2003) come from measurements in two parts of

117: the sky, HDFN in the North and CDFS in the South, observed at two

118: different Epochs, E1 and E2, $\sim$6 months apart. The two epochs

119: had different detector orientations rotated by $\sim$180$^\circ$

120: with respect to the sky, allowing a test for certain instrumental

121: effects and zodiacal light. The final mosaics were assembled from

122: the individual Basic Calibrated Data (BCD) frames which were

123: self-calibrated using the method of \cite{calibration} similarly

124: to KAMM; no extended source calibration factors were applied to

125: the data. To evaluate the random noise level of the maps,

126: alternate calibrated frames for each dataset were mapped into

127: separate ``A'' and ``B'' mosaics.

128: 

129: The assembled maps were cleaned of resolved sources in two steps:

130: First, an iterative procedure was applied which computes the

131: standard deviation, $\sigma$, of the image, and masks pixels

132: exceeding $N_{\rm cut} \sigma$ along with $N_{\rm

133: mask}\!\times\!N_{\rm mask}$ surrounding pixels until no pixels

134: exceed  $N_{\rm cut} \sigma$. We adopted $N_{\rm mask}$=3 and

135: $N_{\rm cut}$=4, when enough pixels ($\geq$65\%) remain for robust

136: Fourier analysis. Second, similar to KAMM, we subtracted the Model

137: of the individual sources using a variant of the CLEAN algorithm

138: \cite{clean}. In a normal CLEAN procedure, the clean components

139: are convolved with an idealized clean beam to produce a map

140: without sidelobe artifacts; we stop short of this, working with

141: the residual map from which the dirty beam and its artifacts have

142: been removed. We construct the Model from the original unmasked

143: mosaics for each channel as follows: (1) the maximum pixel

144: intensity is located, (2) the PSF is then scaled to half of this

145: intensity and subtracted from the image, (3) the process is

146: iterated, saving intermediate results.

147: 

148: Because spectroscopic redshifts are unavailable and resolving

149: individual sources is difficult in the confusion limit of these

150: observations, we removed the sources via the Model to a fixed

151: level of the shot noise. Its amplitude due to remaining sources is

152: $P_{\rm SN} = \int_{m_{\rm lim}}^\infty [S(m)]^2 dN/dm \; dm$.

153: $S(m)$ is the flux of magnitude $m$ in nW/m$^2$; diffuse flux is

154: defined as $\nu I_\nu$, with $I_\nu$ being the surface brightness

155: in MJy/sr. With the Model iteration where the maps remain close to

156: a fixed $P_{\rm SN}$ we can probe the CIB fluctuations produced by

157: populations below a relatively well-defined flux threshold. In

158: principle, the remaining instrument noise, truncated of its high

159: {\it positive} peaks, may have an imprint of the beam and mimic

160: part (or even most) of the shot noise from cosmic sources. In that

161: case, the shot noise we present in Fig. 1 represents an {\it

162: upper} limit on the contribution from cosmic sources making our

163: conclusions below stronger.

164: 

165: The maps, clipped and with the Model subtracted, were

166: Fourier-transformed and subjected to all the same tests as in

167: KAMM; the additional tests possible with this data are described

168: below. The fluctuation field, $\delta F(${\mbox{\boldmath$x$}),

169: was not weighted for the results presented here (although we

170: checked that weighting by the observation time in each pixel does

171: not lead to any appreciable changes), and its Fourier transform,

172: $f($\mbox{\boldmath$q$}$)= \int \delta F($\mbox{\boldmath$x$}$)

173: \exp(-i$\mbox{\boldmath$x$}$\cdot$\mbox{\boldmath$q$}$) d^2x$ was

174: calculated using the fast Fourier transform. The 2-D power

175: spectrum is defined as $P_2(q)\equiv \langle |f_q|^2\rangle$ and

176: in this definition a typical flux fluctuation is $\simeq

177: \sqrt{q^2P_2(q)/2\pi}$ on the angular scale of wavelength $2

178: \pi/q$. The blanked pixels were assigned $\delta F$=0, thereby not

179: adding power to the eventual power spectrum. Muxbleed was removed

180: by zeroing the corresponding frequencies in the $(u,v)$ plane

181: before computing the power spectrum of the maps, $P_S$. We

182: subtracted remaining linear gradients from the maps before

183: clipping and again after clipping. The noise power spectrum,

184: $P_N$, was evaluated from the $\frac{1}{2}(A-B)$ data using the

185: mask from the main maps. The remaining power spectrum was

186: evaluated as $P=P_S - P_N$.

187: 

188: \section{Results}

189: 

190: Fig. \ref{fig:results} shows results from the new data compared to

191: the earlier QSO1700 data analysis. As discussed above, in the

192: present analysis we can regulate the faint source removal from the

193: maps by fixing the floor level of the shot noise, whose amplitude

194: $P_{\rm SN}$ was evaluated from the fits to the small scale power

195: spectrum using our model for the beam. Because we mapped the GOODS

196: data onto 1.2$^{\prime\prime}$ pixels (instead of

197: $0.6^{\prime\prime}$ for the QSO1700 data) the beam used for

198: generating the Model was a slightly smoothed version of that used

199: in the QSO1700 data.

200: 

201: The upper panels in the figure show the CIB fluctuations in the

202: GOODS fields evaluated at the Model iteration when $P_{\rm SN}$

203: roughly corresponds to that in the earlier QSO1700 analysis. The

204: figure shows CIB fluctuations consistent with those in the earlier

205: analyzed QSO1700 dataset. Our results are consistent that at the

206: same level of $P_{\rm SN}$ all data are probing the same

207: populations. (At the deeper Model iterations extra galaxy

208: populations were removed in the present analysis compared to the

209: QSO1700 field and at shallower Model iterations extra galaxy

210: populations were removed in the QSO1700 field). Because of longer

211: integration in the new data, we can remove sources to still lower

212: $P_{\rm SN}$. The lower panels of Fig. \ref{fig:results} show the

213: CIB fluctuations remaining in the maps at the lowest common value

214: of $P_{\rm SN}$; the signal shown here comes from still fainter

215: sources than the limits in the earlier KAMM analysis.

216: 

217: \section{Discussion}

218: 

219: The following can, in principle, contribute to the detected

220: fluctuations 1) instrumental (systematic and random), 2) source

221: artefacts, 3) Solar System and Galactic foregrounds, and 4)

222: extragactic sources. We briefly discuss the contributions of each

223: and conclude that (with the possible exception of the 8\um\ data)

224: the detected fluctuations are due to CIB from extragalactic

225: sources below our removal threshold set by the shot-noise

226: amplitude.

227: 

228: {\it Checking for systematics: E1 vs E2}. In measuring the faint,

229: low spatial frequency backgrounds, control of systematic errors

230: remains a major challenge.  Of particular concern is scattered

231: light  in the Spitzer and IRAC optical systems, which can spread

232: the light from point sources over large spatial scales.  Some

233: scattered light retains a fixed relative position with respect to

234: the originating point source. Other scattered light gets to the

235: focal plane after multiple reflections, and its spacing with

236: respect to the originating source will vary as the source is moved

237: in the field of view. Finally, some scattered light may arise from

238: sources outside the field of view, and may change its illumination

239: of the detector in a complex way as the telescope is moved on the

240: sky. Given that scattered light and  detector artifacts can cause

241: structure in the image, it is important that we use observations

242: for these studies which allow us to  evaluate the size of such

243: possible effects in the images we analyze.

244: 

245: To search for  instrumental sources of large scale power, we used

246: the partially overlapping E1 and E2 data of the HDN and CDFS

247: regions in which the telescope and optical system are rotated by

248: $\sim$180 degrees with respect to the field of view. Using such

249: observations, we can compare the source-removed sky maps from the

250: two epochs to separate structure which is unchanged in inertial

251: coordinates, and thus presumably arising from the sky, and that

252: which changes, which could represent an instrumental contribution.

253: 

254: We selected overlapping subfields for the final mosaicked images

255: which had fairly homogeneous exposures. The latter limited the

256: selected areas to approximately the same size: $142\times504$

257: arcsec in the CDFS field and $149\times 504$ arcsec in the HDFN

258: area. We then computed the subfield and cross-subfield correlation

259: functions and coefficients to verify that the signal is the same

260: at each Epoch and is detector-orientation independent. Having

261: carried out the point source subtraction and gradient removal on

262: the two epochs, we  compared the correlation function of the

263: difference map to that of each of the individual maps, and also

264: did a cross correlation analysis. In both cases, we find that all

265: of the large scale power arises from the sky down to the presented

266: statistical error bars; this rules out any significant

267: contribution from instrumental scattered light and detector

268: artifacts.

269: 

270: We also estimated that there is at most only a small contribution

271: to the measured power spectrum from image artefacts, such as e.g.

272: associated with occasional strong muxbleed. This was done by

273: selecting sub-regions excluding the areas of obvious artefacts.

274: (This is responsible for a slight excess in power at 3.6 \um\

275: around 1-2 arcmin for one of the fields).

276: 

277: {\it Procedural sources of large-scale power}. Because IRAC's

278: calibration does not include zero-flux closed-shutter data, a

279: degeneracy remains in the absolute zero point of the solution. A

280: similar degeneracy is present with respect to first-order

281: gradients in the mosaics. However higher-order gradients are not

282: degenerate. Because of these degeneracies, linear gradients have

283: been fit and subtracted from the self-calibrated mosaicked images.

284: The zero-level is unimportant to this study, so all analyzed maps

285: are set to a mean intensity equal to zero.

286: 

287: We find - via simulations - that in small, but non-negligible

288: number of cases the gradient-subtraction can also remove the

289: genuine cosmic power at the largest scales and the power shown

290: there should thus be treated as the {\it lower} limit on any

291: cosmic fluctuations. However, we find in the QSO1700, the HDFN-E1

292: and HDFN-E2 analyses that the results do not change appreciably

293: even if the extra-subtraction is done at {\it each} step of the

294: Model iteration. The CDFS fields, which have a common overlap at

295: both Epochs, exhibit more sensitivity (at the level of $\sim 20\%$

296: in the largest bins) to this extra gradient subtraction. Taken

297: together this is consistent with the cosmic nature of the

298: fluctuations shown in Fig. \ref{fig:results} and the CDFS field

299: happening to be statistically more sensitive to the extra gradient

300: subtraction.

301: 

302: The  possible contributions from incompletely removed galaxies are

303: also small as discussed in KAMM. Briefly: 1) the measured

304: fluctuations are independent of the masking and clipping

305: parameters. E.g. when clipping is $N_{\rm cut}$=2 only 6\% of the

306: maps remained \cite{kamm}, but the correlation function, which

307: replaces the power spectrum as a measure of large-scale

308: correlations for such deeply cut maps, remained practically the

309: same. 2) The results are invariant when the masking size around

310: each clipped pixel is increased (from $N_{\rm mask}$=3 to 7). This

311: mask is larger than the typical galaxy size at $z\gsim$0.1 so

312: individual galaxies are expected to have been removed completely.

313: The Model further removes much of the remaining emissions at

314: larger angles. 3) We measure the clustering component from $\sim

315: 0.5^\prime$ to $\sim$5-10$^\prime$, which subtend 1 to 10 Mpc at

316: $z$=0.1,1. If the power comes from the incompletely removed local

317: galaxies, its angular spectrum should reflect the slope of the

318: observed galaxy two-point correlation function, contrary to Fig.

319: \ref{fig:results}. 4) The results are the same for all fields at

320: the fixed level of shot-noise. The level of the remaining $P_{\rm

321: SN}$ fixes the amount of the remaining flux from incompletely

322: removed sources of full magnitude $m$. The contribution from them

323: to the large scale clustering should then have been a proportional

324: fraction of the CIB clustering produced by the parental sources.

325: 5) By construction in their results, KAMM identified the

326: appropriate Model iteration number where the clean components

327: become largely uncorrelated with the sources in the original

328: image. (Here we use $P_{\rm SN}$ as the alternative criterion).

329: 

330: {\it Zodiacal and Galactic foregrounds}. The cirrus flux in Table

331: 1 are the SPOT estimates of ISM intensity, based on the Schlegel

332: et al (1998) 100 \um\ IRAS intensity and temperature maps and a

333: spectral scaling relation derived from DIRBE, ISO and AROME

334: observations of the ISM

335: (http://ssc.spitzer.caltech.edu/documents/background/bgdoc\_release.html;

336: Reach \& Boulanger 1997). Fluctuations in the cirrus emission are

337: assumed to be at the 1\% level as suggested by the intensity of

338: the fluctuation spectrum of a low galactic latitude field in which

339: cirrus emission was clearly evident at 8 \um\ (KAMM). As in the

340: QSO1700 data, the cirrus may contribute a non-negligible

341: contribution at 8 \um, and prevents us from isolating the

342: cosmological signal there. Even assuming that the entire signal at

343: 8 \um\ is produced by cirrus and taking the mean Galactic cirrus

344: energy distribution \cite{cirrus} gives an upper limit on the

345: cirrus emission at shorter wavelengths well below the signal we

346: detect.

347: 

348: The range in zodiacal light intensity in the time spanned by the

349: observations is shown in Table 1. Its fluctuations are a factor of

350: $\simeq$4-15 below those from cirrus. A limit of $\delta F <$0.1

351: \nwm2sr at 8 \um\ due to zodiacal light was derived by KAMM from

352: the observed dispersion of the data at two epochs separated by

353: about six months. This limit at 8 \um\ (with relative fluctuations

354: at a much lower level than the $<$0.2\% limit by Abraham et al.

355: (1997) at 25 \um) was then scaled to the shorter wavelengths using

356: the zodiacal light spectrum derived from COBE/DIRBE data (Kelsall

357: et al. 1998). In the present study too we find no evidence for

358: zodiacal emission above the instrument noise levels; the arcminute

359: limits on zodiacal fluctuations from subtracting the two Epochs

360: are $\delta F \lsim$0.05 \nwm2sr at 8 \um.

361: 

362: {\it Extragalactic sources}. The detected signal is thus due to

363: CIB fluctuations from extragalactic sources, such as ordinary

364: galaxies and the putative Population III. KAMM estimated that the

365: CIB flux from the remaining galaxies was only $\simeq$0.15 \nwm2sr

366: so that they were unlikely to account for the strong clustering

367: signal. The present data enable us to eliminate intervening

368: sources down to lower levels of the shot noise, or lower flux

369: limits. The detected signal has to originate in still fainter

370: sources. In a companion paper (Kashlinsky, Arendt, Mather \&

371: Moseley 2006) we discuss the constraints our results (both the

372: shot-noise and clustering components of the fluctuations) place on

373: the nature of the sources contributing them. We show that the

374: signal at 3.6, 4.5 \um\ must arise from very faint populations

375: with individual fluxes $\lsim$10-20 nJy and that the amplitude of

376: the fluctuations at arcminute scales requires these populations to

377: be significantly more luminous per unit mass than the present-day

378: ones.

379: 

380: This work is supported by NSF AST-0406587 and NASA Spitzer

381: NM0710076 grants.

382: 

383: %\texttt{\{thebibliography\}}%

384: \begin{thebibliography}{3}

385: \bibitem [Abell 2002]{abell}{Abell, T. 2002, Science, 295, 93}

386: \bibitem [\'Abrah\'am et al 1997]{abraham}{\'Abrah\'am, P., Leinert, Ch., \& Lemke, D. 1997, A\&A, 328, 702}

387: \bibitem [Arendt et al 1998]{cirrus}{Arendt, R. et al 1998, Ap.J., 508,74}

388: %\bibitem [Arendt \& Dwek 2003]{arendtdwek}{Arendt, R. \& Dwek, E. 2003, Ap.J., 585, 305}

389: %\bibitem [Bertin \& Arnouts 1996]{sextractor}{Bertin, E. \& Arnouts, S. 1996, Astron.Astrophys. Suppl., 117, 393}

390: \bibitem [Bromm et al 1999]{bromm}{Bromm, V. et al 1999, Ap.J., 527,

391: L5}

392: \bibitem [Bromm \& Larson 2004]{brommlarson}{Bromm, V. \& Larson,

393: R. 2004, Ann. Rev. A. A., 42, 79}

394: \bibitem [Cooray et al 2004]{cooray}{Cooray, A. et al 2004, Ap.J., 606, 611}

395: %\bibitem [Cooray \& Yoshida 2004]{pop3-qso}{Cooray, A. \& Yoshida,

396: %N. 2004, MNRAS, 351, L71}

397: \bibitem [Dickinson et al 2003]{goods}{Dickinson, M. et al 2003, ``The great observatories origins deeps survey",

398: in ``The mass of galaxies at low and high redshift", ed. R. Bender

399: \& A. Renzini, astro-ph/0204213}

400: %\bibitem [Dwek \& Arendt 1998]{dwekarendt}{Dwek, E. \& Arendt, R. 1998,

401: %Ap.J.,508,L9}

402: %\bibitem [Fazio et al 2004]{fazio}{Fazio, G. et al 2004,

403: %Ap.J.Suppl., 154, 39}

404: \bibitem [Fernandez \& Komatsu 2006]{komatsu}{Fernandez, E. R. \&

405: Komatsu, E. 2005, Ap.J., 646, 703}

406: \bibitem [Fixsen et al 2000]{calibration}{Fixsen, D. J.,

407: Moseley, S. H. \& Arendt, R. G. 2000, Ap. J. Suppl., 128, 651}

408: %\bibitem [Gardner et al 2006]{jwst}{Gardner, J.P. et al 2006,

409: %Space Sci Reviews, in press. astro-ph/0606175}

410: \bibitem [H$\ddot{\rm o}$gbom 1974]{clean}{H$\ddot{\rm o}$gbom, J. 1974, Ap.J.Suppl., 15,417}

411: \bibitem [Kashlinsky 2005]{review}{Kashlinsky, A. 2005, Phys. Rep., 409,

412: 361-438}

413: \bibitem [Kashlinsky et al 2004]{kagmm}{Kashlinsky, A., Arendt, R., Gardner, J.P., Mather, J.C., \& Moseley, S.H. 2004, Ap.J., 608, 1 }

414: \bibitem [KAMM]{kamm}{Kashlinsky, A., Arendt,

415: R., Mather, J.C. \& Moseley, S.H. 2005, Nature, 438, 45}

416: \bibitem [Kashlinsky et al 2006]{interpretation}{Kashlinsky, A., Arendt,

417: R., Mather, J.C. \& Moseley, S.H. 2006, Ap.J., submitted.}

418: %\bibitem [Kashlinsky \& Rees 1983]{kr83}{Kashlinsky, A. \& Rees,

419: %M.J. 1983, MNRAS, 205, 955}

420: \bibitem [Kelsall et al 1998]{kelsall}{Kelsall, T. et al 1998,

421: Ap.J., 508, 44}

422: %\bibitem [Leinert et al 1997]{leinert}{Leinert, Ch. et al 1997, A\&A Suppl,

423: %127,1}

424: \bibitem [Madau \& Silk 2005]{madausilk}{Madau, P. \& Silk, J.

425: 2005,MNRAS, 359, L37}

426: \bibitem [Reach \& Boulagger 1997]{reach}{Reach, W. T. \& Boulanger, F. 1997, in

427: ``Infrared emission from interstellar dust in the local

428: interstellar medium: The Local Bubble and Beyond", eds. D.

429: Breitschwerdt, M. J. Freyburg, \& J. Trumper;352-362,

430: Springer:Berlin}

431: \bibitem [Santos et al 2002]{santos}{Santos, M.R., Bromm, V., Kamionkowski, M. 2002,MNRAS,336,1082}

432: \bibitem [Salvaterra \& Ferrara 2003]{sf2003}{Salvaterra, R. \& Ferrara, A. 2003, MNRAS, 339, 973}

433: \bibitem[Schlegel et al 1998]{schlegel}{Schlegel, D., Finkbeiner,

434: D. \& Davis, M. 1998, Ap.J., 500, 525}

435: 

436: \end{thebibliography}

437: 

438: \clearpage

439: 

440: \begin{deluxetable}{lcccccc}

441: \tabletypesize{\scriptsize} \tablewidth{0pt}

442: \tablecaption{Analyzed Fields\label{table1}} \tablehead{

443: \colhead{Parameter} & \colhead{Channel} &\colhead{QSO1700} &

444: \colhead{HDFN-E} & \colhead{HDFN-E2} & \colhead{CDFS-E1} &

445: \colhead{CDFS-E2} } \startdata

446: ($l_{\rm Gal},b_{\rm Gal}$) (deg) & \nodata &  ($94.4,36.1$) & ($125.9,54.8$) & ($125.9,54.8$) & ($223.6,-54.4$) & ($223.6,-54.4$) \\

447: ($\lambda_{\rm Ecl},\beta_{\rm Ecl}$) (deg) & \nodata & ($194.3, 83.5$) & ($148.4, 57.3$) & ($148.4, 57.3$) & ($41.1, -45.2$) & ($41.1, -45.2$) \\

448: Size (arcmin) & \nodata & $5.1\!\times\!11.5$ & $10.2\!\times\!10.2$ & $10.2\!\times\!10.2$ & $8.8\!\times\!8.4$\tablenotemark{a} & $9.0\!\times\!8.4$\tablenotemark{b}\\

449: $\langle t_{\rm obs}\rangle$ (hr)\tablenotemark{c} & \nodata & 7.8 & 20.9 & 20.7 & 23.7 & 22.4 \\

450: %\cutinhead{Channel 1 (3.6 $\micron$)}

451: $P_{\rm SN}$ $({\rm nW}2{\rm m}^{-4}{\rm sr})$ & 1 (3.6 $\micron$) & $5.8\!\times\!10^{-11}$ & $1.9\!\times\!10^{-11}$ & $1.9\!\times\!10^{-11}$ & $2.2\!\times\!10^{-11}$ & $2.3\!\times\!10^{-11}$\\

452: Flux: zodi (nW/m$^2$/sr) & 1 (3.6 $\micron$) & 32 & 45--47 & 37--38 & 48--51 & 48--49\\

453: Flux: cirrus (nW/m$^2$/sr) & 1 (3.6 $\micron$) & 2.5 & 0.8 & 0.8 & 0.8 & 0.8\\

454: %\sidehead{Channel 2 (4.5 $\micron$)}

455: $P_{\rm SN}$ $({\rm nW}2{\rm m}^{-4}{\rm sr})$ & 2 (4.5 $\micron$) & $6.0\!\times\!10^{-11}$ & $1.1\!\times\!10^{-11}$ & $1.0\!\times\!10^{-11}$ & $9.5\!\times\!10^{-12}$ & $1.1\!\times\!10^{-11}$\\

456: Flux: zodi (nW/m$^2$/sr) & 2 (4.5 $\micron$) & 132 & 174--185 & 146--153 & 189--204 & 165--176\\

457: Flux: cirrus (nW/m$^2$/sr) & 2 (4.5 $\micron$) & 2.7 & 1.3 & 1.3 & 1 & 1 \\

458: %\sidehead{Channel 3 (5.8 $\micron$)}

459: $P_{\rm SN}$ $({\rm nW}2{\rm m}^{-4}{\rm sr})$ & 3 (5.8 $\micron$) & $6.0\!\times\!10^{-10}$ & $1.2\!\times\!10^{-10}$ & $1.3\!\times\!10^{-10}$ & $1.5\!\times\!10^{-10}$ & $1.5\!\times\!10^{-10}$\\

460: Flux: zodi (nW/m$^2$/sr) & 3 (5.8 $\micron$) & 873 & 1192--1250 & 996--1026 & 1256--1327 & 1179--1243\\

461: Flux: cirrus (nW/m$^2$/sr) & 3 (5.8 $\micron$) & 7.8 & 36. & 3.6 & 2.6 & 2.6 \\

462: %\sidehead{Channel 4 (8 $\micron$)}

463: $P_{\rm SN}$ $({\rm nW}2{\rm m}^{-4}{\rm sr})$ & 4 (8 $\micron$) & $4.4\!\times\!10^{-10}$ & $1.0\!\times\!10^{-10}$ & $1.1\!\times\!10^{-10}$ & $1.1\!\times\!10^{-10}$ & $1.4\!\times\!10^{-10}$\\

464: Flux: zodi (nW/m$^2$/sr) & 4 (8 $\micron$) & 1723 & 2411--2500 & 2012--2054 & 2509--2614 & 2454--2562\\

465: Flux: cirrus (nW/m$^2$/sr) & 4 (8 $\micron$) & 33.8 & 16.1 & 16.1 & 10.1 & 10.1\\

466: \enddata

467: \tablenotetext{a}{$9.0\!\times\!8.4$ arcmin for Channels 2 and 4.}

468: \tablenotetext{b}{$8.8\!\times\!8.4$ arcmin for Channels 2 and 4.}

469: \tablenotetext{c}{For Channel 1. Other channels may vary by as

470: much as 0.5 hr, except for the QSO1700 region where $\langle

471: t_{\rm obs}\rangle = 9.2$ hr.}

472: \end{deluxetable}

473: \clearpage

474: 

475: \begin{figure}

476: \plotone{f1.eps} \caption{CIB fluctuations from the analyzed

477: fields. The fluctuation spectrum was evaluated by averaging over

478: the concentric rings center at given $q$ and the errors shown

479: correspond to the cosmic variance for the estimates, i.e. the

480: Poissonian errors of relative amplitude $N_q^{-1/2}$ with $N_q$

481: being the number of independent Fourier elements used in

482: determining the power at $q$. Top: the CIB fluctuations were

483: evaluated for the shot-noise levels approximately corresponding to

484: the $P_{\rm SN}$ of the QSO1700 data. Black error bars show the

485: QSO1700 fields results from \cite{kamm}. Red symbols correspond to

486: the HDFN fields and blue symbols denote the CDFS data results.

487: Triangles correspond to E1 and circles to E2. The lines show the

488: fluctuations due to shot-noise from the remaining sources. Bottom:

489: Same as top shown for the lowest levels of $P_{\rm SN}$ reached

490: with the new data. The blue and red dotted lines show the

491: estimated cirrus fluctuations for the CDFS and HDFN fields

492: respectively. } \label{fig:results}

493: \end{figure}

494: 

495: \end{document}

496: