1: \documentclass[12pt,preprint]{aastex}
2:
3: \begin{document}
4:
5: \title{Spatial Variations of Galaxy Number Counts in the Sloan Digital Sky Survey. II. Test of Galactic Extinction in High Extinction Regions}
6:
7: \author{Naoki Yasuda, Masataka Fukugita}
8:
9: \affil{Institute for Cosmic Ray Research, University of Tokyo, Kashiwa, 277-8582, Japan}
10: \email{yasuda@icrr.u-tokyo.ac.jp}
11:
12: \and
13: \author{Donald P. Schneider}
14: \affil{Department of Astronomy and Astrophysics, 525 Davey Laboratory, Pennsylvania State University, University Park, PA 16802}
15:
16: \begin{abstract}
17:
18: Galactic extinction is tested using galaxy number counts at low
19: Galactic latitude obtained from five band photometry of the
20: Sloan Digital Sky Survey. The spatial variation of galaxy
21: number counts for low extinction regions of $E(B-V) < 0.15$ is
22: consistent with the all-sky reddening map of \citet*{SFD} and the
23: standard extinction law. For higher extinction regions of $E(B-V) >
24: 0.15$, however, the map of \citet{SFD} overestimates the reddening
25: by a factor up to 1.4, which is likely ascribed to the departure from
26: proportionality of reddening to infrared emissivity of dust. This result is
27: consistent with the analysis of \citet{AGa} for the Taurus dark cloud
28: complex.
29:
30: \end{abstract}
31:
32: \keywords{dust, extinction --- techniques: photometric}
33:
34: \defcitealias{PaperI}{Paper~I}
35: \defcitealias{SFD}{SFD}
36:
37: \section{Introduction}
38:
39: The Sloan Digital Sky Survey \citep[SDSS:][]{York2000}
40: provides an unprecedented wide-field imaging survey of the sky
41: with homogeneous, few-percent-error photometry. It allows
42: study of the spatial variation of the projected galaxy number density over
43: the wide field. In our previous publication
44: \citep[Paper~I]{PaperI}, it was shown
45: that the variation of the galaxy number counts is accounted
46: for by the Galactic extinction and large-scale galaxy clustering.
47:
48: Galaxy counts provide a clean test of the Galactic extinction
49: integrated out to the edge of the Milky Way, provided that the
50: galaxies are sampled to sufficiently faint magnitude and their numbers
51: are sufficiently large \citep[see, e.g.,][]{Hubble,BH}. The test does not
52: need any models or external calibrations to derive extinction. Moreover, using
53: SDSS photometric data, extinctions in five-bands can be determined
54: independently and the extinction curve can be derived. The primary limitation
55: of this technique is that one cannot explore the small scales where spatial variations
56: of the galaxy distribution are strongly affected by galaxy clustering. The mean amount of
57: Galactic extinction can be well documented, and the error for scale
58: larger than a degree can be made smaller than the order that concerns
59: us. The study in \citetalias{PaperI} indeed verified the validity of
60: the reddening map of \citet[SFD]{SFD} on the scale of a degree and the
61: extinction laws of \citet{Cardelli} and \citet{ODonnell} to within 5\%
62: for five optical colours. More recently, a similar but more detailed
63: analysis reveals in very
64: low extinction regions slight underestimates of extinction in the SFD map,
65: which are likely due to the contamination of infrared emission
66: from galaxies in the map of dust far infrared emission \citep{Yahata}.
67:
68: Our study of extinction in \citetalias{PaperI} was limited to
69: regions of low extinction [$E(B-V) < 0.15$] since the galaxies
70: used are derived from the main survey area of SDSS \citep[][DR1]{DR1}
71: that intentionally avoids the region of large extinction to
72: explore extragalactic science. A significant amount of SDSS imaging data,
73: however, has been obtained at low Galactic latitudes for
74: the purpose of photometric commissioning and calibrations. A part
75: of such surveys, that include the Orion region, are published in
76: \citet{OrionDR}. The purpose of the present paper is to extend our
77: study to high extinction regions and examine the validity of the
78: reddening map and the extinction law given by \citetalias{SFD}.
79:
80: We note that some indication was already reported that the SFD
81: reddening map may not be correct for high extinction regions from a
82: study of stars in the background of the Taurus dark cloud complex
83: \citep{AGa}. Arce \& Goodman suggested that extinction derived from SFD is
84: overestimated by a factor of $1.3-1.5$ in regions
85: with $A_V > 0.5$ mag, or $E(B-V) > 0.16$. Our study will test
86: this result in the Orion region.
87:
88: In Section 2 we describe the sample, data selection and the procedure
89: to derive extinction from galaxy number counts. Section 3 presents
90: the result together with some discussion. A summary is given in
91: Section 4.
92:
93: \section{Data and analysis}
94: \label{sec:data}
95:
96: The imaging by SDSS telescope \citep{Gunn98,Gunn06} is carried out with
97: the SDSS $ugriz$ filters \citep{Fukugita1996}.
98: Our analysis is based on the data given by \citet{OrionDR}. We
99: use the data of stripe 82 (southern equatorial stripe) which are
100: derived from photometric scans, runs 211(82S), 259(82N), and 273(82S).
101: This is the only stripe in \citet{OrionDR} for which a north-south
102: pair of stripes, which are needed to form a contiguous
103: area, are observed.
104: The region covers $11\fdg55 < \alpha < 91\fdg55$ and $-1\fdg25 <
105: \delta < +1\fdg25$, corresponding to $125\arcdeg \lesssim l \lesssim
106: 207\arcdeg$ and $-62\arcdeg \lesssim b \lesssim -
107: 10\arcdeg$. Figure \ref{fig:map} shows the region covered by current
108: data set overlaid in the reddening map of \citetalias{SFD} for the
109: Galactic southern hemisphere. The total area of our data set, 200
110: deg$^2$, is about 10\% of that covered by \citetalias{PaperI}
111: derived from SDSS DR1. The mean selective reddening $E(B-V)$ increases
112: from 0.03 at a high Galactic latitude to $\approx 1$ in the Orion
113: region close to the Galactic plane, as seen in Figure \ref{fig:raE},
114: which shows $E(B-V)$ of \citetalias{SFD} averaged
115: over the $2.5\arcdeg$ square regions along the stripe. We note that
116: extinction in dark cloud regions is patchy, and $E(B-V)$ occasionally
117: reaches $>10$ mag for small regions, typically for $15\arcmin$.
118:
119: An object catalogue and five colour band images are given in
120: \citet{OrionDR}. The catalogue is created by photometric pipeline {\tt
121: photo} v5\_4\_25, which is used to produce Data Release 2 \citep{DR2}
122: and later data releases \citep{DR3,DR4,DR5}. We select galaxies primarily
123: using {\tt objType} parameter in the catalogue, but drop objects which
124: are flagged as {\tt EDGE}, {\tt BRIGHT}, {\tt SATURATED}, or {\tt
125: BLENDED}. It sometimes happens that some objects are not detected in
126: all five bands. We judge the detection using logical {\tt OR} of {\tt
127: BINNED1}, {\tt BINNED2}, and {\tt BINNED4} flag set per band, which
128: mean that objects are detected in a $1\times1$, $2\times2$, and
129: $4\times4$ binned image, respectively. The entries that are not
130: detected in a given band are not included in our number counts in that
131: band. There are some entries in the catalogue whose corresponding
132: objects appear to be much too faint for the quoted magnitudes; their
133: surface brightness is indeed faint. The occurrence of this type of spurious
134: detection is common in all colour bands. When the surface brightness
135: distribution is plotted for the galaxy sample, these objects produce a
136: spike at nearly zero surface brightness. Since the reason for the
137: inclusion of such entries in the catalogue is not clear, we apply a
138: surface brightness cut at the level of sky background to exclude
139: those entries in the $r$ band. This procedure automatically removes
140: the surface brightness spikes in the $g$, $i$ and $z$ bands. For the $u$ band, there are many
141: true galaxies that have low surface brightness, so we cannot separate
142: the fake spike from the tail of true galaxies; we simply assume that
143: the removal of the spike in the $r$ band also removes spurious objects in the $u$
144: band. This procedure rejects about
145: 2\% of the $r$ band selected galaxies in the dereddened magnitude range of $17.5 < r <
146: 19.5$. We find that most of these entries are flagged as {\tt
147: nopetro}, but if we would use {\tt nopetro} as the selection criterion
148: we would miss an additional 3\% of objects, which are mostly true
149: galaxies.
150:
151: To account for patchy extinction structure in a heavily
152: reddened region, we divided the total area of 200 square degrees into
153: pixels of $2.3\arcmin\times2.3\arcmin$. This pixel size was adopted to
154: match that of the SFD reddening map ($2.372\arcmin$). There are 135,200
155: such pixels in our area. All pixels are sorted in the order of increasing
156: $E(B-V)$ values of the SFD map and grouped every 4225 pixels, so that
157: the sum of the area of each group amounts to $(2.5\arcdeg)^2$, which is
158: large enough to determine extinction from galaxy number counts.
159: Figure \ref{fig:Ebv} shows the mean $E(B-V)$ value of each group. Half
160: of the groups have $E(B-V) < 0.1$. Since the group of the largest $E(B-V)$
161: spans too wide a range of reddening, we subdivide it into four: namely
162: the last four points have been grouped with 1056 pixels (one quarter
163: of 4225). The last dataset includes pixels with $E(B-V)$ from 1 to
164: 33; it is dropped from our analysis.
165:
166: The present analysis differs from that in \citetalias{PaperI} in that
167: contiguous regions of 2.5 degree squares are considered. Such an
168: analysis is not appropriate to deal with the sample that includes high
169: extinction regions, since extinction is very patchy for dark cloud
170: regions and the average of extinction over a large area does not
171: necessarily match the magnitude offset calculated from galaxy
172: counts\footnote{For example, in an area where half of the
173: region is clear and the other half is extincted by 20 magnitudes,
174: then the mean extinction would be 10mag. However, galaxies can be
175: observed in the clear area and the number of galaxy will be half that
176: expected for the case without extinction. This leads us to infer that
177: the magnitude offset would be about 0.5mag.}. For this reason we
178: divide the area into small regions and assemble the region according
179: to extinction values in the present paper. For low extinction
180: regions, the details of the binning are not important.
181:
182: We use the Petrosian magnitude \citep[see][]{EDR,TS} in our study.
183: Petrosian magnitudes are suitable for the analysis given
184: in this paper, since the Petrosian radii are unaffected by the
185: foreground extinction and Petrosian magnitudes measure the same
186: fraction of the flux of galaxies regardless of their foreground
187: extinction. This means that we can correct the extinction properly
188: just by subtracting the extinction value in each band. This does not
189: hold for other galaxy magnitudes such as isophotal or aperture
190: magnitudes.
191: We have to note that Petrosian magnitudes are not
192: corrected for seeing variations unlike model magnitudes
193: which are PSF-convolved model fits.
194: The fraction of total magnitude measured by
195: Petrosian magnitude will change as galaxies become smaller
196: as a result of worse seeing \citep{Blanton}.
197: This effect is only a few percent for the galaxies used in
198: this study ($17.5 < r < 22.0$ or $1.0'' < r_{50} < 1.3''$) and
199: does not affect our results.
200: We refer astrometric calibrations to \citet{Pier} and
201: photometric calibrations of the main survey to
202: \citet{Smith} \citep[see also][]{Hogg,Tucker} and
203: \citet{Ivezic}. Since the secondary standard star ``patches'' are
204: sparse or nonexistent for much of the Orion region, photometric
205: calibrations for the current data use the \"ubercalibration
206: algorithm \citep[See][for details]{OrionDR,Gunn06}.
207:
208: We derive the mean extinction-free galaxy number count (differential
209: count) $\bar{N}(m)$ from the entire sample in low extinction regions
210: of $E(B-V) < 0.1$ by employing the SFD reddening map and the default
211: standard extinction law, $k(r) = A_r/E(B-V)=2.751$
212: \citepalias[Table 6 of][]{SFD}. We take this
213: mean relation as the reference. We then count the number of galaxies for
214: each group without applying the extinction correction, and fit to the
215: reference count by shifting the amount of magnitude $\Delta m$, i.e.,
216: as $\bar{N}(m+\Delta m)$; this $\Delta m$ represents the extinction in
217: the specific region.
218:
219: We first work with the $r$ band counts, but extend the study later to
220: other colour bands. It is desirable to work with number counts at a
221: level as faint as possible, so that the galaxy number density is
222: sufficiently large to minimise the Poisson noise and the spatial
223: distribution of galaxies is sufficiently smooth to minimise the
224: large-scale clustering effects. \citetalias{PaperI} uses the data in
225: the magnitude range of $r = 18.5 - 20.5$. In the current study, this approach is not
226: appropriate; the reddening expected in the SFD map varies from $E(B-V)
227: = 0.03$ to $0.9$ (regions of the largest extinction are
228: discarded). This range corresponds to $A_r = 0.08 - 2.5$, and the
229: count in highly reddened regions falls out of the range set as the
230: reference magnitude band. To avoid this problem we set the range in a
231: way that dereddened magnitude range is the same for all regions. We use the
232: counts whose number density per square degrees per 0.5 mag is
233: \begin{equation}
234: 1.8 < \log N(m) < 2.6,
235: \label{eq:Nm}
236: \end{equation}
237: and fit $N(m)$ to the reference count to derive $\Delta m$. This range
238: corresponds to $r = 17.5 - 19.5$ on the reference galaxy number count,
239: which is one mag brighter than that in \citetalias{PaperI}. This is
240: still reasonably faint, yet photometric measurements are made at a
241: high signal-to-noise ratio and star-galaxy separation is sufficiently
242: reliable even with extinction. This ability to classify objects is particularly
243: important for our study because the
244: contamination of stars becomes more serious at low Galactic latitude
245: where the star density is high and high extinction pushes the objects
246: to fainter magnitudes. The number of galaxies contained in
247: $2.5\arcdeg\times2.5\arcdeg$ area integrated over $1.8 < \log N(m) <
248: 2.6$ is approximately 4100. The expected Poisson noise of 1.6\% is
249: negligible for the present work. It occasionally happens that the
250: faint end of eq. (\ref{eq:Nm}) goes beyond the magnitude at which
251: incompleteness starts ($r=22.0$); in this case we drop the data at the bins that
252: go beyond the incompleteness limit. Figure \ref{fig:ncfig}
253: shows examples of $r$-band galaxy counts in low and high extinction
254: regions. We see in this figure how magnitude offsets are evaluated.
255:
256: The normalisation of the reference number count
257: in the current sample is lower than that derived in
258: \citetalias{PaperI} by 5\% ($\Delta \log N(m)=0.02$)\footnote{
259: It is lower by 7\% ($\Delta \log N(m)=0.03$) than \citet{Yasuda}
260: which is based on the SDSS Early Data Release \citep{EDR}.}.
261: A half of this difference is explained by the zero point offset
262: of photometry between DR1 and \citet{OrionDR}, the latter being fainter
263: by $\approx 0.02$ mag. The other half is due to the present
264: omission of objects that produce this spike at zero surface brightness
265: explained above: we did not apply this cut in \citetalias{PaperI}.
266:
267: We carry out a similar analysis for four other colour bands. The
268: reference range of counts (as log $N$), the corresponding dereddened
269: magnitude range, approximate number of galaxies, and the limiting
270: magnitude are given in Table \ref{tbl:param}. Note that we must
271: choose a range of number of galaxies that is smaller in the bluer bands. In
272: particular, the $u$ counts go quickly out of the reference range in
273: the presence of extinction. After deriving
274: reference number counts in the extinction-free limit of the low
275: extinction galaxy samples, we compute $\Delta m_\lambda$ for the
276: counts in each group. In deriving the reference counts
277: we assume the standard extinction curve, $k(\lambda)=A_\lambda/E(B-V)$
278: with $k(\lambda)=5.155$, 3.793, 2.086, and 1.479 for $u$, $g$, $i$, and $z$.
279: Note that $k(\lambda)$ varies by a factor of 3.5 across $u$ to $z$,
280: and by a factor 2.5 if we drop the $u$ band. This makes the use of
281: $g,r,i,z$ colours appropriate to study the extinction curve, even if we
282: would exclude $u$ for its poorer photometry.
283:
284: We find the offset in the normalisations of the reference count,
285: $\Delta \log N(m)=0.3,~ 0.03,~ 0.03$, and $0.05$ for $u$, $g$, $i$,
286: and $z$ compared to those in Paper I. The offsets in $g$, $i$, and $z$
287: bands are explained in the same way as for the $r$ band. The offset in
288: $u$ band is large. The shape of the counts at the faint end also
289: differs from that in paper I. These differences originate from the
290: fact that $u$ band surface brightness is faint for many galaxies and
291: the photon collecting efficiency in $u$ band is low, both contributing
292: to poor photometry in this colour band. A comparison of the photometry
293: between DR1 and \citet{OrionDR} for common galaxies shows that the two
294: magnitudes differ typically as much as 0.5 mag randomly\footnote{For
295: stars this large scatter is not seen. The difference of the two
296: magnitudes is no more than twice those in $r$ or $g$ bands; the mean
297: scatter is 0.03 mag at $u=19$.}. There are also some differences in
298: the selection procedure. In the present analysis we required
299: `detection' in the $u$ band in 1$\times$1, 2$\times$2, or 4$\times$4
300: binned image, which we did not do in Paper I. The surface brightness cut
301: in $r$ is exercised in the present analysis. The effect is small for
302: the $r$ band, but this drops 10\% of objects for $18.0<u<20.0$ in the
303: $u$ band count. These effects altogether might induce a large error
304: in for the $u$ band count. We expect, however, a substantial part of
305: the errors are likely to cancel when we deal with relative quantities,
306: like those in our analysis.
307:
308: \section{Results}
309:
310: In Figure \ref{fig:Ar} we show $\Delta m_r$ versus Galactic
311: extinction, $A_r^{\rm SFD}$, calculated from the reddening map of
312: \citetalias{SFD} averaged over each group with the effective area of
313: $(2.5\arcdeg)^2$ assuming the standard extinction curve. For each
314: group, we applied a jackknife method with the data divided into 10
315: samples to estimate statistical errors. The horizontal bars stand for
316: the range of $A_r$ in each group. The data points for which the faint
317: end of the observed magnitude corresponding to eq. (\ref{eq:Nm})
318: becomes fainter than the incompleteness limit are shown by open
319: circles. We note that the scatter of the points is substantially reduced
320: compared to those in \citetalias{PaperI}, where it is dominated by large
321: scale clustering of galaxies.
322: We expect a $\pm 0.11$ mag variation from large scale galaxy
323: clustering if the area considered were taken from contiguous regions
324: along the stripe. The observed scatter (0.04 mag) is consistent with
325: the mean error of fitting of the reference count to the data of
326: $N(m)$, which is 0.04. This reduced scatter obviously arises from
327: the fact that the information of the large scale structure is mostly
328: lost by division and reassembling of the area. This is an advantage
329: of the present procedure from the viewpoint of testing the effect of
330: extinction.
331:
332: Figure \ref{fig:Ar} shows that for $A_r^{\rm SFD} < 0.4$ (or $E(B-V) <
333: 0.15$), $\Delta m_r$ is proportional to $A_r^{\rm SFD}$ when averaged
334: over the scatter. This agrees with what we demonstrated in
335: \citetalias{PaperI}: the galaxy counts are consistent with the reddening map
336: and the standard extinction law. For a high extinction regime,
337: $A_r^{\rm SFD} > 0.4$ or $E(B-V) > 0.15$, however, the data
338: significantly deviate from the identity regression line: $\Delta m_r$
339: is smaller than $A_r^{\rm SFD}$, indicating that $A_r^{\rm SFD}$
340: overestimates the Galactic extinction.
341:
342: This overestimate of Galactic extinction in the SFD map is consistent
343: with the analysis of \citet{AGa,AGb}, who studied extincion in the
344: Taurus dark cloud complex using the number count and reddening of
345: background stars and 100$\mu m$ infrared emission (calibrated with
346: star number counts). They reached the conclusion that the SFD map overestimates
347: the extinction by a factor of $1.3-1.5$ for the region of $A_V > 0.5$.
348:
349: We perform a similar analysis for the $u$, $g$, $i$, and $z$ colour
350: bands. The relations between $\Delta m_\lambda$ and $A_\lambda^{\rm
351: SFD}$ are shown in Figure \ref{fig:Augiz}. For the $g$, $i$ and $z$
352: bands we see a similar departure from the $\Delta m_\lambda=
353: A_\lambda^{\rm SFD}$ line for $E(B-V)\gtrsim 0.15$, while the data for
354: $E(B-V)\lesssim 0.15$ are consistent with the identity line. The
355: departure for the $u$ band appears somewhat different, but we note that
356: the data for $A_u^{\rm SFD}>2$, where a significant departure is observed, are
357: derived from the counts whose faint end extends beyond the incompleteness
358: limit, in addition to the problem of photometry discussed above. We
359: take the results with the $u$ band hereinafter only for the purpose to
360: see the broad, rather than quantitative, consistency.
361:
362: The departure from the identity regression is more clearly
363: demonstrated in Figure \ref{fig:ratio}, where the ratio of $\Delta
364: m_\lambda / A_\lambda^{\rm SFD}$ is plotted as a function of
365: $A_\lambda^{\rm SFD}$ for each band. The mean values are presented as
366: horizontal bars (for numerical values see Table \ref{tbl:ratio}) for
367: the ranges of $E(B-V)=0.05-0.15$, $0.15-0.45$, and $0.40-1.00$. For
368: $0.05 < E(B-V) < 0.15$, these ratios are consistent with unity
369: within $1\sigma$ errors. Apparent departure from unity,
370: $\Delta m_\lambda<A_\lambda^{\rm SFD}$, is visible for higher extinction
371: ranges. The ratio $A_\lambda^{\rm SFD} / \Delta
372: m_\lambda$ is approximately $1.25$ for $E(B-V)=0.15-0.45$,
373: and 1.4 for $E(B-V)=0.45-1.00$ irrespective of the colour bands
374: (again except for the $u$ band).
375:
376: Two alternative interpretations for this observed departure are (1) the
377: reddening function $k(\lambda)$ is non-linear for a large $A_\lambda$
378: or (2) the selective reddening $E(B-V)$ estimated by SFD is not correct
379: for large extinction. To distinguish between the two possibilities,
380: the regressions of $\Delta m_\lambda$ among different colour bands are
381: plotted in Figure \ref{fig:ratio2}. The dotted lines show the
382: relations expected for the standard extinction law, i.e., $\Delta
383: m_{\lambda_1}/\Delta m_{\lambda_2}=k(\lambda_1)/k(\lambda_2)$. Note
384: that the scales of the abscissa and ordinate are not identical. The
385: figure shows that the data points follow the expected relation; no
386: non-linearity is observed in the extinction curve across the regions
387: of low to high extinction.
388:
389: We may estimate $k(\lambda)$ relative to the reference band, which we
390: take to be the $r$ band. Defining $k'_\lambda = k(\lambda)/k(r)$,
391: we can calculate this quantity from the observed relation $\Delta
392: m_\lambda = k'_\lambda \cdot \Delta m_r$. The results from
393: this analysis are shown in Figure \ref{fig:kk}. The $k$
394: values in Table \ref{tbl:kk} are obtained by multiplying $k(r)=2.751$
395: on $k'_\lambda$. The observed values of $k(\lambda)$ are consistent
396: with those of the standard extinction curve.
397:
398: This analysis suggests that $E(B-V)$ of SFD is
399: overestimated for high extinction regions. Using $\Delta m_\lambda /
400: A_\lambda^{\rm SFD}$ from $g$ to $z$ bands, we suggest the true
401: selective reddening being written as
402: \begin{equation}
403: E(B-V)^{\rm true}=E(B-V)^{\rm SFD}\left[0.87-0.13~ {\rm Erf}\left(\frac{E(B-V)^{\rm SFD}-0.19}{0.11}\right)\right],
404: \label{eq:correct}
405: \end{equation}
406: in terms of $E(B-V)^{\rm SFD}$. The prediction of this function is
407: presented in Figure \ref{fig:ratio}.
408:
409: In agreement with the analysis of \citet{AGa} we ascribe the
410: overestimate of $E(B-V)$ by SFD to the inaccuracy of the conversion of
411: $100\mu m$ emission of dust ($D^T$) to selective reddening for high
412: extinction regions. \citetalias{SFD} assume a simple linear relation
413: between reddening and $100\mu m$ emission as $E(B-V) = pD^T$, where
414: $p$ is a parameter.
415: To determine $p$, \citetalias{SFD} use the relation between the
416: intrinsic \bv colour and the Mg$_2$ line index of elliptical galaxies.
417: Of the 389 elliptical galaxies used,
418: only $\approx20$ reside in regions with $E(B-V) > 0.15$, and there are
419: no elliptical galaxies in the region of $E(B-V) > 0.4$. We notice that
420: the fitting of $100\mu m$ vs $E(B-V)$ relation of SFD (see their Figure
421: 6) starts deviating from linear relation for $E(B-V)\gtrsim
422: 0.15$. From their Figure 6, the value of $\delta (B-V)$ is about
423: $-0.08$ mag for elliptical galaxies whose $E(B-V)$s are between 0.2 and
424: 0.4; this corresponds to overestimation of reddening by $1.2-1.4$ when
425: a linear relation is assumed. This is quantitatively consistent with
426: our result. The relation between the reddening and $100\mu m$ emission
427: should be modified to incorporate the non-linearity.
428:
429: We briefly discuss the possible reasons for the departure between
430: $100\mu m$ emission and reddening from a linear
431: relation. Overestimation of reddening can be caused by the
432: overestimation of $100\mu m$ emission. According to the procedure of
433: \citetalias{SFD}, $100\mu m$ emission will be overestimated when color
434: temperature estimated from the ratio of intensities at 100 and 240
435: $\mu m$ is underestimated. If dense clouds have larger dust grains,
436: their equilibrium temperature will be lower even in the same radiation
437: field. This will cause overestimation of $100\mu m$ emission. Could
438: the composition be different? \citetalias{SFD} assumed the emissivity
439: model of $\epsilon_\nu = \nu^\alpha$ with $\alpha = 2.0$. If there are
440: materials with $\alpha = 1.5$, their temperature will be
441: underestimated. This is the same sense as we have seen. How about
442: mixed temperature along the line of sight? From Figure 2 of
443: \citetalias{SFD} the column density will be underestimated when there
444: are two regions at different equilibrium temperatures. This is the
445: opposite sense to what we have seen. From these brief discussion, the
446: difference of grain size and emissivity model in dense clouds could be
447: the cause of overestimation of 100$\mu m$ emission. However, we need
448: more detailed observations to explore the ways of modifying the
449: \citetalias{SFD} map.
450:
451: \section{Summary}
452:
453: We have tested Galactic extinction using galaxy number counts at the
454: low Galactic latitude using the SDSS galaxy sample of
455: \citet{OrionDR} that covers 200 deg$^2$ from low $E(B-V)\approx 0.03$
456: to high $E(B-V)\approx 1$ extinction regions. The variation of galaxy
457: number counts is consistent with Galactic extinction described by the
458: prediction of SFD for low extinction regions of $E(B-V) < 0.15$. For
459: high extinction regions of $E(B-V) > 0.15$, the SFD extinction
460: prescription overestimates the reddening by a factor up to $1.4$,
461: which we interpret as a result of the departure of the linear
462: relation between 100$\mu m$ infrared emission and
463: selective extinction.
464:
465: \acknowledgments
466:
467: We thank Jill Knapp for invaluable suggestions for improving the
468: analysis.
469:
470: Funding for the SDSS and SDSS-II has been provided by the Alfred
471: P. Sloan Foundation, the Participating Institutions, the National
472: Science Foundation, the U.S. Department of Energy, the National
473: Aeronautics and Space Administration, the Japanese Monbukagakusho, the
474: Max Planck Society, and the Higher Education Funding Council for
475: England. The SDSS Web Site is {\tt http://www.sdss.org/}.
476:
477: The SDSS is managed by the Astrophysical Research Consortium for the
478: Participating Institutions. The Participating Institutions are the
479: American Museum of Natural History, Astrophysical Institute Potsdam,
480: University of Basel, Cambridge University, Case Western Reserve
481: University, University of Chicago, Drexel University, Fermilab, the
482: Institute for Advanced Study, the Japan Participation Group, Johns
483: Hopkins University, the Joint Institute for Nuclear Astrophysics, the
484: Kavli Institute for Particle Astrophysics and Cosmology, the Korean
485: Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos
486: National Laboratory, the Max-Planck-Institute for Astronomy (MPIA),
487: the Max-Planck-Institute for Astrophysics (MPA), New Mexico State
488: University, Ohio State University, University of Pittsburgh,
489: University of Portsmouth, Princeton University, the United States
490: Naval Observatory, and the University of Washington.
491:
492: %{\it Facility:} \facility{Sloan ()}
493:
494: \begin{thebibliography}{}
495:
496: \bibitem[Abazajian et al.(2003)]{DR1} Abazajian, K. et al.\ 2003, \aj, 126, 2081
497:
498: \bibitem[Abazajian et al.(2004)]{DR2} Abazajian, K. et al.\ 2004, \aj, 128, 502
499:
500: \bibitem[Abazajian et al.(2005)]{DR3} Abazajian, K. et al.\ 2005, \aj, 129, 1755
501:
502: \bibitem[Adelman-McCarthy et al.(2006)]{DR4} Adelman-McCarthy, J.~K. et al.\ 2006, \apjs, 162, 38
503:
504: \bibitem[Adelman-McCarthy et al.(2007)]{DR5} Adelman-McCarthy, J.~K. et al.\ 2007, \apjs, in press
505:
506: \bibitem[Arce \& Goodman(1999a)]{AGa} Arce, H.~G., \& Goodman, A.~A.\ 1999, \apjl, 512, L135
507:
508: \bibitem[Arce \& Goodman(1999b)]{AGb} Arce, H.~G., \& Goodman, A.~A.\ 1999, \apj, 517, 264
509:
510: \bibitem[Blanton et al.(2001)]{Blanton} Blanton, M.~R. et al.\ 2001, \aj, 121, 2358
511:
512: \bibitem[Burstein \& Heiles(1982)]{BH} Burstein, D., \& Heiles, C.\ 1982, \aj, 87, 1165
513:
514: \bibitem[Cardelli et al.(1989)]{Cardelli} Cardelli, J.~A., Clayton, G.~C., \& Mathis, J.~S.\ 1989, \apj, 345, 245
515:
516: \bibitem[Finkbeiner et al.(2004)]{OrionDR} Finkbeiner, D.~P. et al.\ 2004, \aj, 128, 2577
517:
518: \bibitem[Fukugita et al.(1996)]{Fukugita1996} Fukugita, M., Ichikawa, T., Gunn, J.~E., Doi, M., Shimasaku, K., \& Schneider, D.~P.\ 1996, \aj, 111, 1748
519:
520: \bibitem[Fukugita et al.(2004)]{PaperI} Fukugita, M., Yasuda, N., Brinkmann, J., Gunn, J.~E., Ivezi{\'c}, {\v Z}., Knapp, G.~R., Lupton, R., \& Schneider, D.~P.\ 2004, \aj, 127, 3155 (PaperI)
521:
522: \bibitem[Gunn et al.(1998)]{Gunn98} Gunn, J.~E. et al.\ 1998, \aj, 116, 3040
523:
524: \bibitem[Gunn et al.(2006)]{Gunn06} Gunn, J.~E. et al.\ 2006, \aj, 131, 2332
525:
526: \bibitem[Hogg et al.(2001)]{Hogg} Hogg, D.~W., Finkbeiner, D.~P., Schlegel, D.~J., \& Gunn, J.~E.\ 2001, \aj, 122, 2129
527:
528: \bibitem[Hubble(1934)]{Hubble} Hubble, E.\ 1934, \apj, 79, 8
529:
530: \bibitem[Ivezi{\'c} et al.(2004)]{Ivezic} Ivezi{\'c}, {\v Z}. et al.\ 2004, AN, 325, 583
531:
532: \bibitem[O'Donnell(1994)]{ODonnell} O'Donnell, J.~E.\ 1994, \apj, 422, 158
533:
534: \bibitem[Pier et al.(2003)]{Pier} Pier, J.~R.\ 2003, \aj, 125, 1559
535:
536: \bibitem[Schlegel et al.(1998)Schlegel, Finkbeiner, \& Davis]{SFD} Schlegel, D.~J., Finkbeiner, D.~P., \& Davis, M.\ 1998, \apj, 500, 525 (SFD)
537:
538: \bibitem[Smith et al.(2002)]{Smith} Smith, J.~A. et al.\ 2002, \aj, 123, 2121
539:
540: \bibitem[Stoughton et al.(2002)]{EDR} Stoughton, C. et al.\ 2002, \aj, 123, 485
541:
542: \bibitem[Strauss et al.(2002)]{TS} Strauss, M.~A. et al.\ 2002, \aj, 124, 1810
543:
544: \bibitem[Tucker et al.(2006)]{Tucker} Tucker, D. et al.\ 2006, Astron. Nachr., 327, 821
545:
546: \bibitem[Yahata et al.(2006)]{Yahata} Yahata, K. et al.\ 2006, astro-ph/0607098 (sumitted to \pasj)
547:
548: \bibitem[Yasuda et al.(2001)]{Yasuda} Yasuda, N. et al.\ 2001, \aj, 122, 1104
549:
550: \bibitem[York et al.(2000)]{York2000} York, D.~G. et al.\ 2000, \aj, 120, 1579
551:
552: \end{thebibliography}
553:
554: \begin{deluxetable}{ccccc}
555: \tablecolumns{5}
556: \tablewidth{0pt}
557: \tablecaption{Parameters for our analysis \label{tbl:param}}
558: \tablehead{
559: \colhead{Band} & \colhead{Range of counts} & \colhead{Mag range} & \colhead{Num. of gal} & \colhead{Limiting mag}\\
560: \colhead{} & \colhead{log $N(m)$ } & \colhead{} & \colhead{} & \colhead{}
561: }
562: \startdata
563: $u$ & $1.0-1.8$ & $18.0-20.0$ & $800$ & $21.6$ \\
564: $g$ & $1.3-2.1$ & $17.5-19.5$ & $1600$ & $22.4$ \\
565: $r$ & $1.8-2.6$ & $17.5-19.5$ & $4100$ & $22.0$ \\
566: $i$ & $1.8-2.6$ & $17.0-19.0$ & $3800$ & $21.2$ \\
567: $z$ & $1.8-2.6$ & $16.5-18.5$ & $3100$ & $19.8$ \\
568: \enddata
569: \end{deluxetable}
570:
571: \begin{deluxetable}{cccc}
572: \tablecolumns{4}
573: \tablewidth{0pt}
574: \tablecaption{Mean values of $\Delta
575: m_\lambda / A_\lambda^{\rm SFD}$ \label{tbl:ratio}}
576: \tablehead{
577: \colhead{} & \multicolumn{3}{c}{Range of $E(B-V)$} \\
578: \cline{2-4} \\
579: \colhead{Band} & \colhead{$0.05-0.15$} & \colhead{$0.15-0.45$} & \colhead{$0.45-1.00$}
580: }
581: \startdata
582: $u$ & $1.130 \pm 0.063$ & $0.950 \pm 0.019$ & \nodata \\
583: $g$ & $1.095 \pm 0.085$ & $0.829 \pm 0.015$ & $0.718 \pm 0.031$ \\
584: $r$ & $1.017 \pm 0.055$ & $0.818 \pm 0.022$ & $0.734 \pm 0.019$ \\
585: $i$ & $1.028 \pm 0.075$ & $0.791 \pm 0.038$ & $0.694 \pm 0.023$ \\
586: $z$ & $1.079 \pm 0.103$ & $0.800 \pm 0.057$ & $0.681 \pm 0.031$ \\
587: \enddata
588: \end{deluxetable}
589:
590: \begin{deluxetable}{ccl}
591: \tablecolumns{3}
592: \tablewidth{0pt}
593: \tablecaption{The value of $k(\lambda)$ \label{tbl:kk}}
594: \tablehead{
595: \colhead{Colour band} & \colhead{Standard extinction curve} & \colhead{Derived
596: from $\Delta N(m)$}
597: }
598: \startdata
599: $u$ & 5.155 & $5.611 \pm 0.165$ \\
600: $g$ & 3.793 & $3.779 \pm 0.045$ \\
601: $r$ & 2.751 & 2.751 (normalisation) \\
602: $i$ & 2.086 & $1.9748 \pm 0.015$ \\
603: $z$ & 1.479 & $1.397 \pm 0.019$ \\
604: \enddata
605: \end{deluxetable}
606:
607: \newpage
608: \begin{figure}
609: \plotone{f1.eps}
610: \caption{Region for the current analysis overlaid on the reddening map
611: of SFD in the Galactic southern hemisphere with the pole at the centre
612: ($b=-90\arcdeg$). The left edge corresponds to $l=180\arcdeg$ and the
613: top is $l=90\arcdeg$. The stripe runs from $(l,b) = (125\arcdeg, -
614: 62\arcdeg)$ to $(207\arcdeg, -10\arcdeg)$ and the square along the
615: stripe represents $2.5\arcdeg$ square regions.
616: The progression of black to white in the figure corresponds to
617: increasing values of $E(B-V)$.
618: \label{fig:map}}
619: \end{figure}
620:
621: \begin{figure}
622: \plotone{f2.eps}
623: \caption{Variation of the mean reddening $E(B-V)$ of SFD averaged over
624: the $2.5\arcdeg$ square region along the stripe in this study.
625: The Orion Complex is located at R.A. = $90\arcdeg$.
626: \label{fig:raE}}
627: \end{figure}
628:
629: \begin{figure}
630: \plotone{f3.eps}
631: \caption{Range of $E(B-V)$ of the SFD map for 32 groups, each group
632: consisting of 4225 pixels that form the effective $2.5\arcdeg$ square
633: region. (The four rightmost points are the groups having one fourth
634: the number of pixels of the others, corresponding to effective
635: $1.25\arcdeg$ square.) Filled circles denote the mean values and bars
636: show the minimum and maximum.
637: \label{fig:Ebv}}
638: \end{figure}
639:
640: \begin{figure}
641: \plotone{f4.eps}
642: \caption{Example galaxy number counts in the $r$-band. Upper panel is
643: for a low extinction ($E(B-V)=0.03$) region and lower panel is for a high
644: extinction ($E(B-V)=0.26$) region. Solid lines are the reference
645: galaxy counts. Solid points (and dashed lines) represent observed
646: galaxy counts, which are shifted to match the reference counts by
647: the amount of $\Delta m$ (denoted by open circles). The bands indicated by
648: dotted lines are the range that is used for matching.
649: \label{fig:ncfig}}
650: \end{figure}
651:
652: \begin{figure}
653: \plotone{f5.eps}
654: \caption{Magnitude offsets $\Delta m_r$ corresponding to the variation
655: of galaxy number counts in groups corresponding to
656: the effective area of $(2.5\arcdeg)^2$
657: plotted against
658: mean extinction $A_r^{\rm SFD}$ calculated from the SFD reddening map
659: and the standard extinction curve.
660: Horizontal bars show the region of $A_r$ in each group and vertical
661: bars are errors obtained by a jackknife estimate.
662: The open circles represent the
663: region where $A_r^{\rm SFD}$ is so large that the faint end of the
664: reference magnitude range falls beyond the incompleteness limit. The
665: dotted line is the identity regressions line $\Delta m_r
666: = A_r^{\rm SFD}$.
667: \label{fig:Ar}}
668: \end{figure}
669:
670: \begin{figure}
671: \plotone{f6.eps}
672: \caption{Same as Figure \ref{fig:Ar}, but for the other four SDSS colour bands:
673: (a) $u$ band, (b) $g$ band, (c) $i$ band, and (d) $z$ band.
674: \label{fig:Augiz}}
675: \end{figure}
676:
677: \begin{figure}
678: \plotone{f7.eps}
679: \caption{The ratio of $\Delta m_\lambda / A_\lambda^{\rm SFD}$ as a
680: function of $A_\lambda^{\rm SFD}$ for five colour bands. The meaning of open
681: circles is the same as in Figure \ref{fig:Ar}. The mean values
682: for different ranges of $E(B-V)$ ($0.05-0.15$, $0.15-0.45$,
683: and $0.45-1.00$) are indicated as horizontal bars, and the empirical
684: fitting function eq. (\ref{eq:correct}) is shown with dotted curves.
685: \label{fig:ratio}}
686: \end{figure}
687:
688: \begin{figure}
689: \plotone{f8.eps}
690: \caption{The regression of $\Delta m_\lambda$ between different
691: combinations of $u$, $g$, $r$, $i$, and $z$. The meaning of open
692: circles is the same as in Figure \ref{fig:Ar}. The dotted lines are
693: relations expected for the standard extinction law.
694: \label{fig:ratio2}}
695: \end{figure}
696:
697: \begin{figure}
698: \plotone{f9.eps}
699: \caption{$k(\lambda)/k(r)$ obtained from galaxy number counts
700: as $\Delta
701: m_\lambda/\Delta m_r$ (data with error bars) and
702: those from the standard extinction curve (crosses).
703: \label{fig:kk}}
704: \end{figure}
705:
706: \end{document}
707: