1: \documentclass[12pt,preprint]{aastex}
2:
3: %\usepackage{lscape}
4: %\slugcomment{This draft \today}
5:
6: \shorttitle{Dust Properties of TMC-1C}
7: \shortauthors{Schnee, S., Kauffmann, J., Goodman, A. \&
8: Bertoldi, F.}
9:
10: \begin{document}
11:
12: \newcommand{\nthp}{N$_2$H$^+$(1-0)}
13: \newcommand{\kms}{km s$^{-1}$}
14:
15: \title{The Effect of Noise in Dust Emission Maps on the Derivation of
16: Column Density, Temperature and Emissivity Spectral Index}
17:
18: \author{S. Schnee$^{1,4}$, J. Kauffmann$^{2,3,*}$, A. Goodman$^1$ \& F. Bertoldi$^3$}
19:
20: \affil{$^1$Harvard-Smithsonian Center for Astrophysics, 60 Garden
21: Street, Cambridge, MA 02138 \\
22: $^2$Max-Planck Institut f\"ur Radioastronomie, Auf dem H\"ugel
23: 69, 53121 Bonn, Germany \\
24: $^3$Argelander Institute for Astronomy, University of Bonn, Auf
25: dem H\"ugel 71, 53121 Bonn, Germany \\
26: $^4$Department of Astronomy, California Institute of
27: Technology, MC 105-24 Pasadena, CA 91125 \\
28: $^*$present address: Harvard-Smithsonian Center for
29: Astrophysics, 60 Garden Street, Cambridge, MA 02138}
30:
31: \email{sschnee@cfa.harvard.edu}
32:
33: \begin{abstract}
34: We have mapped the central 10\arcmin$\times$10\arcmin\ of the dense
35: core TMC-1C at 450, 850 and 1200 \micron\ using SCUBA on the James
36: Clerk Maxwell Telescope and MAMBO on the IRAM 30m telescope. We show
37: that although one can, in principle, use images at these wavelengths
38: to map the emissivity spectral index, temperature and column density
39: independently, noise and calibration errors would have to be less than
40: $\sim$2\% to accurately derive these three quantities from a set of
41: three emission maps. Because our data are not this free of errors, we
42: use our emission maps to fit the dust temperature and column density
43: assuming a constant value of the emissivity spectral index and explore
44: the effects of noise on the derived physical parameters. We find that
45: the derived extinction values for TMC-1C are large for a starless core
46: ($\sim$80 mag $A_V$), and the derived temperatures are low ($\sim$6 K)
47: in the densest regions of the core, using our derived value of $\beta
48: = 1.8$.
49: \end{abstract}
50:
51: \keywords{stars: formation --- dust, extinction --- submillimeter}
52:
53: \section{Introduction}
54: Efforts to determine the mass and temperature of starless
55: cores from sub-millimeter and millimeter observations are hampered by
56: uncertainties in the emission properties of the dust grains, such as
57: the emissivity spectral index. Although in principle it should be
58: possible to calculate the column density of dust, the emissivity
59: spectral index of the dust, and the dust temperature from observations
60: at three or more wavelengths, in practice this has never been done for
61: a starless core. A similar analysis has been done for circumstellar
62: disks, e.g. \citep{Beckwith91, Mannings94, Mathieu95} in which
63: temperature gradients, disk masses and spectral indices are
64: calculated. In this paper we explore the levels of uncertainty in the
65: derived dust temperature ($T_d$), emissivity spectral index ($\beta$)
66: and column density ($N_H$) resulting from datasets of either three or
67: four noisy emission maps at different wavelengths. We then apply this
68: analysis to the starless core TMC-1C.
69:
70: TMC-1C is a starless core in the Taurus molecular cloud, at an
71: approximate distance of 140 pc \citep{Kenyon94}. It was shown that
72: TMC-1C is a coherent core, meaning that its velocity dispersion is
73: roughly constant, at slightly more than the sound speed, over a radius
74: of 0.1 pc \citep{Barranco98, Goodman98}. The velocity field of TMC-1C
75: shows evidence of solid body rotation, at 0.3 \kms\ pc$^{-1}$
76: \citep{Goodman93}, and the \nthp\ spectrum reveals the signature of
77: sub-sonic infall \citep{Schnee05a}. The mass derived from 450 and 850
78: \micron\ maps alone (13 M$_\odot$) is several times the virial mass,
79: and the density profile is similar to that of a Bonnor-Ebert sphere
80: \citep{Schnee05a}.
81:
82: Here we use data taken with SCUBA (at 450 and 850 \micron) and
83: MAMBO (at 1200 \micron) to make maps of the dust column density and
84: temperature, and to estimate a constant value for the emissivity
85: spectral index of TMC-1C. Although the high signal to noise at 850
86: and 1200 \micron\ make this set of maps one of the best yet available
87: for a starless core, we show that the noise is still too high to
88: reliably map variations in the emissivity spectral index of TMC-1C.
89:
90: \section{Observations} \label{OBS}
91: \subsection{SCUBA} \label{SCUBAOBS}
92: We observed a 10\arcmin$\times$10\arcmin\ region around TMC-1C using
93: SCUBA \citep{Holland99} on the JCMT. Our maps, especially at 450
94: \micron, benefitted from exceptionally stable grade 1 weather. We
95: used the standard scan-mapping mode, recording the 850 and 450
96: \micron\ data simultaneously \citep{Pierce-Price00, Bianchi00}. Three
97: chop throw lengths of 30\arcsec, 44\arcsec, and 68\arcsec\ were used
98: in both the right ascension and declination directions. The JCMT has
99: FWHM beams of 7.5\arcsec\ at 450 \micron\ and 14\arcsec\ at 850
100: \micron, which subtend diameters of 0.005 and 0.01 pc, respectively,
101: at the distance of Taurus. Pointing during the observations was
102: typically good to 3\arcsec\ or better. The data reduction for the
103: SCUBA data is described by \citet{Schnee05a}. The absolute flux
104: calibration is uncertain at levels of $\sim$4\% at 850 \micron\ and
105: $\sim$12\% at 450 \micron. The rms noise in the 850 \micron\ map is
106: 8.6 mJy/beam, and 13 mJy/beam in the 450 \micron\ map, measured in
107: regions with no significant emission.
108:
109: The Emerson2 technique used to reconstruct SCUBA scan maps
110: from its component chop throws introduces false structure that can be
111: removed \citep{Johnstone00}. To remove this structure, we convolved
112: the SCUBA images (after masking out pixels with $|S| > 5 \sigma$) with
113: a Gaussian of FWHM twice the size of the largest chop throw, and
114: subtracted this from the original image, as explained in
115: \citet{Reid05}. The resulting image has fluxes nearly identical to
116: the original in regions of high signal to noise, but has fewer bowls
117: of negative emission and other artifacts introduced by chopping and
118: image reconstruction. We only use data from the high signal to noise
119: central region of the maps to estimate the dust properties. We
120: account for the SCUBA error beams by convolving the 850 and 1200
121: \micron\ maps with the 450 \micron\ PSF, and convolving the 450 and
122: 1200 \micron\ maps with the 850 \micron\ PSF before regridding to a
123: common resolution, as explained in detail in \citet{Reid05}.
124:
125: The original bowls of negative flux in our 450 \micron\ map on
126: either side of region `a' in Figure \ref{TMC1CFIG} have peak values
127: around $-150$ mJy, and average values around $-75$ mJy. We believe
128: that these structures are artifacts of image reconstruction. It is
129: possible that structures of similar magnitude (either positive or
130: negative) might be affecting data elsewhere in the map.
131:
132: \subsection{MAMBO} \label{MAMBOOBS}
133:
134: We observed the 1.2 mm continuum emission in November 8, 2002,
135: October 23, 2003, and November 2, 2003 using the 117-channel MAMBO-2
136: array \citep{Kreysa99} at the IRAM 30-meter telescope on Pico Veleta
137: (Spain). The FWHM beam size on the sky was 10\farcs7. The source was
138: mapped on-the-fly, with the telescope subreflector chopping in azimuth
139: by 60\arcsec\ to 70\arcsec\ at a rate of 2 Hz; the total on-target
140: observing time was about 6 hours. The line-of-sight optical depth
141: varied between 0.1 and 0.5. The data were reconstructed using the EKH
142: algorithm in an iterative way that properly reproduces large-scale
143: emission (Kauffmann et al., in prep.). The rms noise in the 1200
144: \micron\ map is 3 mJy/beam, measured in regions with no significant
145: emission. The flux calibration uncertainty is approximately 10\%,
146: which is derived from the rms of calibrator observations across pool
147: observing sessions and the uncertainty in the intrinsic calibrator
148: fluxes.
149:
150: %\section{Method} \label{METHOD}
151: \section{Solving for physical parameters} \label{EQSOLVE}
152:
153: In principal, one can use three measured quantities (e.g. the three
154: flux density maps at 450, 850 and 1200 \micron) to solve for three
155: unknowns (e.g. maps of the dust temperature, emissivity spectral
156: index, and column density). Our maps originally are in units of flux
157: density per beam, with a different beam size for each map. To derive
158: meaningful physical quantities we smooth and rebin our maps to a
159: common resolution of the largest beam, which in our case is 14\arcsec.
160: As a result, the quanity that we work with is flux density per
161: 14\arcsec\ pixel, which is what we present in our maps of TMC-1C.
162:
163: The flux density per beam in each map is given by:
164: \begin{equation} \label{DUSTFLUX4}
165: S_{\nu} = \Omega B_{\nu}(T_d) \kappa_{_{\nu}} \mu m_H N_{H_2},
166: \end{equation}
167: where
168: \begin{equation} \label{BLACKBODY4}
169: B_\nu(T_d) = \frac{2h\nu^3}{c^2} \frac{1}{\exp(h\nu /kT_d)-1}
170: \end{equation}
171: and
172: \begin{equation} \label{KAPPA4}
173: \kappa_{_{\nu}} = \kappa_{230} \left(\frac{\nu}{230 {\rm GHz}}
174: \right)^{\beta}.
175: \end{equation}
176: In Equation \ref{DUSTFLUX4}, $S_{\nu}$ is the flux density per
177: 14\arcsec\ pixel; $\Omega$ is the solid angle of the beam;
178: $B_{\nu}(T_d)$ is the blackbody emission from the dust at temperature
179: $T_d$; $\kappa_{230} = 0.009$ cm$^2$ g$^{-1}$ is the emissivity of the
180: dust grains at 230 GHz \citep{Ossenkopf94}; $m_H$ is the mass of the
181: hydrogen atom; $\mu = 2.8$ is the mean molecular weight of
182: interstellar material in a molecular cloud per hydrogen molecule;
183: $N_{H_2}$ is the column density of hydrogen molecules and a
184: gas-to-dust ratio of 100 is assumed. It should be noted that
185: $\kappa_{230}$ is uncertain by a factor of $\sim$2, and that we
186: assumed that $\kappa_{230} = 0.005$ and $\mu = 2.33$ (which is the
187: mean molecular weight per free particle for an abundance ratio of
188: $N(H)/N(He) = 10$ and negligible metals) in \citet{Schnee05a}.
189:
190: The ratio of two fluxes, because of the common beam size, can
191: be simply expressed as:
192: \begin{equation} \label{RATIOEQ}
193: \frac{S_{\nu_1}}{S_{\nu_2}} = \left(\frac{\nu_1}{\nu_2}\right)^{3+\beta}
194: \left(\frac{\exp[h\nu_2/kT_d] - 1}{\exp[h\nu_1/kT_d] - 1}\right)
195: \end{equation}
196:
197: The dust temperature can be found independently of the dust
198: emissivity spectral index by taking the difference between the ratio
199: of fluxes, if we assume that each line of sight through the core can
200: be characterized by a single temperature and emissivity spectral
201: index:
202: \begin{eqnarray} \label{TEMPEQ}
203: \log\left(\frac{S_{450}}{S_{850}}\right)
204: \log\left(\frac{1200 \micron}{850 \micron}\right) - \nonumber \\
205: \log\left(\frac{S_{850}}{S_{1200}}\right)
206: \log\left(\frac{850 \micron}{450 \micron}\right) = \nonumber \\
207: \log\left(\frac{\exp\left[\lambda_T / 850 \micron \right] - 1}
208: {\exp\left[\lambda_T / 450 \micron \right] - 1}\right)
209: \log\left(\frac{1200 \micron}{850 \micron}\right) - \nonumber \\
210: \log\left(\frac{\exp\left[\lambda_T / 1200 \micron \right] - 1}
211: {\exp\left[\lambda_T / 850 \micron \right] - 1}\right)
212: \log\left(\frac{850 \micron}{450 \micron}\right),
213: \end{eqnarray}
214: where $\lambda_T = hc/kT_d$.
215:
216: Once the dust temperature is determined, the emissivity
217: spectral index can be calculated by:
218: \begin{eqnarray} \label{BETAEQ}
219: \beta =\log\left(\frac{S_{850}}{S_{1200}}
220: \frac{\exp\left[\lambda_T / 850 \micron \right] - 1}
221: {\exp\left[\lambda_T / 1200 \micron \right] - 1}\right)
222: \left/
223: \log\left(\frac{1200 \micron}{850 \micron}\right) \right. - 3.
224: \end{eqnarray}
225:
226: The column density of dust can be derived from the flux at a single
227: wavelength (e.g. at 1200 \micron), the temperature of the dust and the
228: emissivity spectral index of the dust using Equation
229: \ref{DUSTFLUX4}. The equivalent visual extinction can be calculated
230: from the column density $N_H$ using:
231: \begin{equation} \label{AVEQ}
232: A_V = N_H R_V \frac{E(B-V)}{N_H}
233: \end{equation}
234: where $N_H = 2 \times N_{H_2}$, $N_H/E(B-V) = 5.8\times10^{21}$
235: cm$^{-2}$ mag$^{-1}$ is the conversion between column density of
236: hydrogen nuclei (for our assumed gas to dust ratio) and the selective
237: absorption, and $R_V = A_V/E(B-V) = 3.1$ is the total to selective
238: extinction ratio for the low-density lines of sight similar to those
239: for which $N_H/E(B-V)$ has been measured \citep{Mathis90, Bohlin78}.
240: Although we use a constant value of $R_V$, we recognize that this
241: value is uncertain to within a factor of 2 between regions of high and
242: low column density \citep{Mathis90}, and the relation between
243: extinction and column density may be different for dense cores like
244: TMC-1C.
245:
246: \section{Error Analysis} \label{ERROR}
247:
248: In order to understand how the noise and reconstruction artifacts in
249: our emission maps will affect the accuracy of our derived dust
250: temperature, emissivity spectral index and column density we have run
251: a variety of Monte Carlo simulations. We compare the effects of using
252: maps at three particular wavelengths to solve for all three parameters
253: to using the three measurements to fit for two parameters while
254: assuming a fixed value for the third. We also show the improvements
255: brought about by using a fourth wavelength and fitting for all three
256: physical parameters.
257:
258: \subsection{Illustrative Examples} \label{ERRORSOLVE}
259:
260: Here we discuss the uncertainties to be expected from solving for
261: $T_d$, $\beta$ and $N_H$ from three emission maps that have noise or
262: whose calibration is uncertain. To illustrate the method, Figure
263: \ref{EXSPEC} shows the modified blackbody spectrum from dust with $A_V
264: = 50$, $T_d = 10$ K and $\beta = 1.5$, observed with a 14\arcsec\
265: beam. The black dotted line shows the true emission spectrum, with
266: crosses at 450, 850 or 1200 \micron, and the blue/red crosses show the
267: flux overestimated/underestimated at one wavelength by 20\%. The blue
268: and red curves are the fitted spectra that pass through the one
269: blue/red cross and the two black crosses. It is clear that a 20\%
270: error in one measurement creates errors in all three derived
271: parameters that are much larger than 20\%. The derived values of
272: $T_d$, $\beta$ and $N_H$ are labeled in Figure \ref{EXSPEC}. For
273: convenience, we show the derived $N_H$ in units of $A_V$.
274:
275:
276: As can be seen from Equation \ref{TEMPEQ}, noise in any of the
277: three flux maps first translates into uncertainty in the derived
278: temperature. This incorrect value of $T_d$, along with any noise in
279: the 850 and 1200 \micron\ maps then results in uncertainties in the
280: derived values of $\beta$ and $N_H$. From Equation \ref{BETAEQ}, one
281: can show that overestimating the temperature will result in an
282: underestimate of $\beta$, and underestimating the temperature will
283: result in an overestimate of $\beta$. The errors in the derived
284: parameters from incorrectly measuring the flux at one wavelength,
285: while correctly measuring the flux at the other two wavelengths is
286: shown in Figure \ref{ONEFLUX}. The anti-correlation between $T_d$ and
287: $\beta$ is clearly seen. Also apparent is that for dust with
288: ``core-like'' values of $\beta$, $T_d$ and $N_H$, errors in 450
289: \micron\ flux result in smaller errors in the derived physical
290: parameters than errors at 850 and 1200 \micron, which is convenient
291: because the 450 \micron\ maps often suffer from higher levels of noise
292: than maps at 850 and 1200 \micron.
293:
294: \subsection{Deriving the Physical Parameters} \label{ERRORFIT}
295:
296: In order to determine the effect of similar levels of noise in
297: all three wavelengths on the three derived physical parameters, we
298: determine the flux at 450, 850 and 1200 \micron\ from dust at $T_d =$
299: 10, 15 and 20 K and $\beta =$ 1.0, 1.5 and 2.0. The fluxes are then
300: modified by a multiplicative factor $f = 1.0 + \delta$ where
301: $\delta$ is randomly chosen from a normal distribution of mean zero
302: and standard deviation $\sigma$, and each flux is modified by a
303: different $f$. This is repeated 10,000 times for each value of
304: $\sigma$ between 0 and 0.2. We show the effect of noise in all three
305: wavelengths on the derived column density, temperature and emissivity
306: spectral index in Figure \ref{ALLTHREE}. At a signal to noise of 20
307: (5\% error), the expected uncertainties in the derived $N_H$, $T_d$
308: and $\beta$ are approximately 50\%, 80\% and 40\%, respectively, for
309: 15 K dust with $\beta = 1.5$. The median values of the derived
310: parameters stay close to the input values at every value of $T_d$ and
311: $\beta$ tested.
312:
313: The analysis presented here deals with the special case that
314: the signal to noise is wavelength independent. In general, this will
315: not be the case, and one might expect that for a given amount of
316: observing time the signal to noise will be worse at 450 \micron\ than
317: at 850 or 1200 \micron\ due to atmospheric effects. In addition, the
318: relative signal to noise between maps will change from position to
319: position, due to the gradients in the dust temperature, column density
320: and emissivity spectral index. For instance, at the position of the
321: column density peak in TMC-1C, the signal to noise ratios at 450, 850
322: and 1200 \micron\ are 5.3, 27 and 33, respectively, when including
323: both the random noise and the artifacts in the 450 \micron\ image.
324:
325: Although even just 5\% errors in the measured fluxes at three
326: wavelengths make accurate determinations of three physical parameters
327: impossible, adding a fourth wavelength (for instance, at 350 \micron\
328: or 2.7 mm) drastically reduces the effects of noise, as can be seen by
329: comparing Figure \ref{ALLTHREE} with Figure \ref{ALLFOUR}. To
330: properly constrain the dust parameters, at least one of the
331: observations should not be on the Rayleigh-Jeans portion of the
332: emission spectrum, or else $T_d$ and $N_H$ will be degenerate in the
333: fit. For dust at $T_d =$ 10, 15 and 20 K and $\beta$ = 1.0, 1.5 and
334: 2.0 we calculate the flux at 350, 450, 850 and 1200 \micron. As
335: before, the fluxes are then modified by a multiplicative factor $f =
336: 1.0 + \delta$, where $\delta$ is again a variable randomly chosen from
337: a normal distribution with mean zero and standard deviation $\sigma$.
338: This is repeated 10,000 times for each value of $\sigma$ between 0 and
339: 0.5. We allow $\sigma$ to be larger than in the previous Monte Carlo
340: simulation because the effects of noise are smaller in this case. For
341: each set of four fluxes, the column density, temperature and
342: emissivity spectral index are fit and the results are shown in Figure
343: \ref{ALLFOUR}. At a signal to noise of 20 (5\% error), the expected
344: errors in the column density, temperature and emissivity spectral
345: index are all on the order of $\sim$1\%. As the signal to noise gets
346: lower, the median derived temperature and column density decrease, and
347: this effect is larger for warmer cores.
348:
349: \subsection{Fixing One Parameter} \label{FIXING}
350:
351: In the case of a starless core observed at three wavelengths, one can
352: hold one parameter fixed, such as assuming a constant $T_d = 15$ K or
353: $\beta = 1.5$, and use the three flux measurements to fit the
354: remaining two parameters. Figure \ref{FITFIX} shows the result of
355: using noisy flux measurements to fit the column density and either the
356: dust temperature or the emissivity spectral index of a core with $T_d
357: =$ 15 K and $\beta =$ 1.5. We show the results for correctly assuming
358: that $\beta = 1.5$ and erroneously assuming that $\beta =$ 2.0 or 1.0,
359: as well as for correctly assuming that $T_d =$ 15 K and erroneously
360: assuming that $T_d = $ 10 or 20 K.
361:
362: We see that by choosing the correct value of $\beta$ and fitting the
363: $T_d$ and $N_H$, noise in the emission maps on the level of 10\%
364: result in $\sim$1\% uncertainties in the derived temperature and
365: column density. Choosing a value of $\beta = 1.0$, when the proper
366: value is $\beta = 1.5$ results in temperatures that are on average 3 K
367: too high, with a spread of $\sim$3 K, and column densities that are
368: high by 8\%. Overestimating the emissivity spectral index, by
369: assuming that $\beta = 2.0$, results in temperatures that are on
370: average 2 K too low and column densities that are 5\% low.
371:
372: Figure \ref{FITFIX} shows that by choosing the $T_d$ and fitting for
373: the $\beta$ and $N_H$, the true values of $\beta$ and column density
374: are recovered, on average, with an absolute spread of 0.20 in $\beta$
375: and a relative spread of 13\% in column when the signal to noise ratio
376: is 10. Underestimating the temperature by assuming $T_d = 10$ K when
377: the proper value is $T_d = 15$ K results in an overestimate of the
378: emissivity spectral index by 0.6, on average, and an overestimate of
379: the column density by 70\%. Overestimating the temperature by
380: assuming that $T_d = 20$ K results in an underestimate of $\beta$ by
381: 0.3, on average, and an underestimate of $N_H$ by a factor of 30\%.
382:
383: \subsection{A Model Core} \label{TMC1CEX}
384:
385: We apply the error analysis presented in Section
386: \ref{ERROR} to our data on the starless core TMC-1C in order to derive
387: new science beyond what was possible in our earlier paper on this core
388: \citep{Schnee05a}. The random noise in the emission maps, as measured
389: in regions with faint emission, is found to be 13 mJy/beam, 9 mJy/beam
390: and 3 mJy/beam, which corresponds to S/N values of 21, 14 and 15 at
391: 450, 850 and 1200 \micron\ at the (0,0) position of the emission maps.
392: Calibration uncertainties are $\sim$12\%, 4\% and 10\% at 450, 850 and
393: 1200 \micron, respectively, and the image reconstruction artifacts in
394: the 450 \micron\ map peak at 150 mJy/beam.
395:
396: To determine how the random noise and reconstruction artifacts (which
397: are spatially correlated and therefore not truly random) in the
398: observed emission maps affect the derived parameters, we construct
399: synthetic emission maps of a starless core like TMC-1C. The model
400: core has a temperature and column density profile equal to the one
401: derived from the two dimensional temperature and column density
402: profile of TMC-1C, derived from fitting the 450, 850 and 1200 \micron\
403: data and assuming a constant $\beta = 1.5$. The model starless core
404: is made of cylindrical shells seen face on with a central temperature
405: of $\sim$6 K, rising to $\sim$12 K at the edge. The central column
406: density corresponds to $\sim$80 magnitudes of visual extinction,
407: falling to an $A_V$ of $\sim$20 at the edge. Using the equations in
408: Section \ref{EQSOLVE}, we derive the resultant emission maps at 450,
409: 850 and 1200 \micron, add in Gaussian noise of the same magnitude as
410: the random noise and reconstruction artifacts described above, and
411: from these emission maps derive maps of the column density,
412: temperature and emissivity spectral index. The resultant maps are
413: shown in Figure \ref{THREECORE}.
414:
415: The dust emission in our model core (shown in Figure
416: \ref{THREECORE}) at 450 and 850 \micron\ is not a good tracer the dust
417: column density, though the 1200 \micron\ emission map does resemble
418: the column density distribution. Spatial gradients in the dust
419: properties, such as the temperature and emissivity spectral index,
420: need to be taken into account when attempting to find the ``peak'' of
421: the dust distribution, even when using the relatively longer
422: wavelength 1200 \micron\ emission map as a proxy for column density.
423:
424:
425: When we solve for all three physical parameters, we find that
426: the derived $\beta$ of our model cloud has a median value of 1.5
427: (which is the input $\beta$ everywhere in the model) with a standard
428: deviation of 0.7. Given the close correspondence between the input
429: value of $\beta$ and the median value derived for it, we see that a
430: constant value for the emissivity spectral index can be estimated in
431: this manner. Using this constant value for $\beta$ everywhere, we can
432: then use our three flux maps to fit the temperature and column
433: density. The resultant maps of our model constructed in this way are
434: shown in Figure \ref{THREECORE}. Although using the median value in
435: the $\beta$ map to derive a constant value for the emissivity spectral
436: index reduces the impact of statistical uncertainty due to random
437: noise, systematic shifts, such as those created from calibration
438: uncertainties, are not removed.
439:
440: \section{Dust Emission in TMC-1C} \label{DISCUSSION}
441: \subsection{Morphology} \label{MORPHOLOGY}
442:
443: The observed 450 \micron\ emission map of TMC-1C is
444: qualitatively different from the 1200 \micron\ map. The 450 \micron\
445: map shows a condensation at $(50,-150)$ (position ``a'' in Figure
446: \ref{TMC1CFIG}) which is much fainter at 850 \micron\ and nearly
447: absent at 1200 \micron. The condensation at $(-50,150)$ (position
448: ``c'' in Figure \ref{TMC1CFIG}) is prominent at all wavelengths, while
449: the column density peak at $(0,0)$ (position ``b'' in Figure
450: \ref{TMC1CFIG}) is prominent at 850 and 1200 \micron, but not apparent
451: at 450 \micron.
452:
453: An emission peak in the longer wavelength maps, but not
454: prominent at 450 \micron, can be explained by a cold temperature, as
455: seen at the $(0,0)$ position. The 450 \micron\ emission peak not seen
456: at 1200 \micron\ can be explained by the dust in that region having a
457: steep spectral index ($\beta \ge 2$).
458:
459: \subsection{Derived Parameters} \label{DERIVED4}
460:
461: Based on our analysis in Section \ref{ERRORSOLVE}, calibration
462: uncertainties, reconstruction artifacts and noise in our TMC-1C
463: emission maps prevent us from making accurate maps of dust
464: temperature, emissivity spectral index and column density {\it
465: simultaneously}, even though this may well be the highest S/N set of
466: such maps of a starless core to date. Figure \ref{TMC1CFIG} shows the
467: results of an attempt to do so, and as expected the range of
468: temperatures that we see is quite broad ($T_d < 5$ K and $T_d > 25$ K)
469: for a starless core, with a similarly large spread in emissivity
470: spectral index ($\beta < 0$ and $\beta > 2.5$) and column density.
471: Furthermore, the errors and uncertainties in our emission maps create
472: a spurious anti-correlation between the dust temperature and
473: emissivity spectral index, which is also seen in our attempt to derive
474: $T_d$, $\beta$ and $N_H$ from our simple model of a cylindrically
475: symmetric starless core, described in Section \ref{TMC1CEX}. Because
476: the noise and errors in our observed emission maps of TMC-1C will
477: drive an anti-correlation between the derived dust temperature and
478: emissivity spectral index even when such a trend does not exist, our
479: results are consistent with a constant value of the emissivity
480: spectral index. However, we cannot rule out a real anti-correlation
481: such as that observed at 200, 260, 360 and 580 \micron\ in the Orion
482: molecular cloud \citep{Dupac01} and the M17 star-forming complex
483: \citep{Dupac02}.
484:
485: Following the method described in Section \ref{TMC1CEX}, we take the
486: median value of the emissivity spectral index map (Figure
487: \ref{TMC1CFIG}) and derive a value and uncertainty of $\beta = 1.8 \pm
488: 0.5$ for the portion of TMC-1C located within the white contour of the
489: 850 \micron\ image in Figure \ref{TMC1CFIG}. Unless otherwise state,
490: in our subsequent analysis of TMC-1C we use a constant value of $\beta
491: = 1.8$ everywhere in the core. We derive $\beta$ and the 1$\sigma$
492: uncertainty by running 1000 realizations of a Monte Carlo simulation
493: of the observed flux maps in TMC-1C modified by the calibration
494: uncertainties. Using $\beta = 1.8$, we construct dust temperature and
495: column density maps (Figure \ref{TMC1CTDAV}) from a fit to the 450,
496: 850 and 1200 \micron\ images. The column density in Figure
497: \ref{TMC1CTDAV} peaks around the maximum of the 850 and 1200
498: \micron\ emission. The implied visual extinction is quite high,
499: rising above 80 magnitudes in the densest regions. As expected, the
500: regions with the highest column density are also the regions with the
501: lowest dust temperature \citep{Zucconi01}. By using a constant value
502: of the emissivity spectral index ($\beta = 1.8$), the dust temperature
503: that we derive is nowhere significantly higher than 16 K nor lower
504: than 6 K.
505:
506: Our derived value of $\beta = 1.8 \pm 0.5$ is somewhat higher
507: than that measured for amorphous carbon grains ($\beta = 1.2$) by
508: \citet{Mennella98} and is within the range ($\sim 1.2 - 2.5$) measured
509: for silicate grains by \citet{Agladze96}. Dust in interstellar disks
510: are generally observed to have values of $\beta \le 1$, but with
511: considerable spread \citep{Beckwith91,Mannings94}. Observations by
512: \citet{Stepnik03} have shown that $\beta = 1.9 \pm 0.2$ for a dense
513: filament in Taurus, which agrees very well with our estimate of
514: TMC-1C. Graphite and silicate dust grains in the ISM are often
515: assumed to have $\beta = 2$ \citep[e.g.][]{Draine84}.
516:
517: \subsection{Comparison With Previous Results} \label{COMPARISON4}
518:
519: Two-dimensional temperature and column density maps of TMC-1C,
520: along with the deprojected three-dimensional temperature and density
521: profiles, have previously been reported in \citet{Schnee05a} using
522: only SCUBA 450 and 850 \micron\ data. With the addition of the MAMBO
523: 1200 \micron\ map, we are better able to constrain the temperature and
524: density and estimate the emissivity spectral index. The emissivity
525: spectral index that we use here ($\beta = 1.8$) is higher than in
526: \citet{Schnee05a} ($\beta = 1.5$), because in this paper we are able
527: to derive $\beta$, and in our ealier work we had to assume a value.
528: The value that we use in this paper for $\kappa_{230}$ is taken from
529: \citet{Ossenkopf94}, which is larger than the number we used in
530: \citet{Schnee05a}. As a result using larger $\beta$ and
531: $\kappa_{230}$, the temperatures that we derive are somewhat lower and
532: the densities are also lower. We choose here to use the dust opacity
533: appropriate for dust grains with thin ice mantles, evolved in a dense
534: (10$^6$ cm$^{-3}$) region for 10$^5$ years derived in
535: \citet{Ossenkopf94} because this is the consensus value settled upon
536: by the {\it Spitzer} Legacy Project, ``From Molecular Cloud Cores to
537: Planet Forming Disks'' \citep{Evans03}, which will make comparisons
538: with other c2d cores easier in the future. The mass that we derive
539: for TMC-1C, within a radius of 0.06 pc from the column density peak is
540: 6 M$_\odot$, as compared with 13 M$_\odot$ in \citet{Schnee05a}.
541: However, even this new dust-derived mass is higher than the virial
542: mass derived from TMC-1C N$_2$H$^+$ observations.
543:
544: Following the method used in \citet{Schnee05a}, we create
545: deprojected three dimensional temperature and density profiles. We
546: assume that the inner 0.07 pc of TMC-1C can be approximated as a set
547: of nested spherical shells, each with a constant temperature and
548: density. The derived temperature and density profiles are shown in
549: Figure \ref{PROFILES}, along with the profiles calculated using only
550: the 450 and 850 \micron\ data and assuming that $\beta = 1.5$, as in
551: \citet{Schnee05a}. The dust temperature profile that we derive is not
552: much changed from that derived in \citet{Schnee05a}, and is thus still
553: consistent with the dust temperature profiles predicted for externally
554: heated starless cores with Bonnor-Ebert density distributions
555: calculated by \citet{Evans01, Goncalves04, Stamatellos04}. A
556: Bonnor-Ebert profile is a good fit to the TMC-1C density profile at
557: radii greater than 0.004 pc, but as in \citet{Schnee05a}, the density
558: we derive for the innermost point is significantly higher than
559: predicted by a Bonnor-Ebert model. The density profile, shown in
560: Figure \ref{PROFILES}, is consistent with a broken powerlaw, with
561: $n(r) \propto r^{-0.5}$ inside 0.035 pc and $n(r) \propto r^{-2.0}$
562: outside 0.035 pc. This is considerably flatter than the density
563: profile derived using only the 450 and 850 \micron\ data and $\beta =
564: 1.5$, which have powerlaw exponents of $-0.8$ and $-3.1$ inside and
565: outside the break radius, respectively (also shown in Figure
566: \ref{PROFILES}). In \citet{Schnee05a} we report inner and outer
567: powerlaw exponents of $-0.8$ and $-1.8$, also using just the 450 and
568: 850 \micron\ data and $\beta = 1.5$, but using a slightly different
569: position as the center of the nested spherical shells, so the
570: comparison of powerlaw exponents presented in \citet{Schnee05a} and
571: those present here would be unfair. Also note that the central
572: density of TMC-1C that we derive from the 450, 850 and 1200 \micron\
573: maps is a factor of $\sim$5 lower than when derived using just the 450
574: and 850 \micron\ maps (and assuming $\beta = 1.5$), which shows the
575: significant adjustments that can result from utilizing additional
576: emission maps. An improved value of the central density can make a
577: significant impact on the predictions of the dynamical state of a
578: core, and on chemical models.
579:
580: \section{Summary}
581:
582: We have used SCUBA data at 450 and 850 \micron\ and MAMBO data at 1200
583: \micron\ to create maps of the dust temperature and column density in
584: TMC-1C, improving the results presented in \citet{Schnee05a}. In
585: addition, we are able to estimate the emissivity spectral index,
586: finding a value of $\beta = 1.8 \pm 0.5$, based on calibration
587: uncertainties.
588:
589: Our analysis shows that noise and calibration errors in maps at 450,
590: 850 and 1200 \micron\ would have to be less than $\sim$2\% to
591: accurately measure the dust temperature, emissivity spectral index and
592: column density from three emission maps. Although such low levels of
593: noise and calibration uncertainties are not achievable with the
594: current generation of bolometers, we show that the dust temperature,
595: emissivity spectral index and column density can be accurately mapped
596: if they are fitted at four wavelengths, for instance by including the
597: 350 \micron\ SHARCII waveband. Obtaining accurate maps of $T_d$,
598: $\beta$ and $N_H$ are necessary for accurate determinations of density
599: and temperature profiles as well as the core mass, which determine the
600: time evolution and chemistry of the core.
601:
602: The next generation detectors on the JCMT (SCUBA-2) and on APEX
603: (LABOCA), will not need to sky-chop, allowing more large-scale
604: structure to be visible, improved calibration and fewer image
605: artifacts, making more accurate determinations of dust properties in
606: cores possible \citep{Ellis05, Gusten06}.
607:
608: \acknowledgments
609:
610: We would like to thank Phil Myers, Ramesh Narayan, David Wilner and
611: Doug Johnstone for their suggestions, assistance, and insights. The
612: suggestions of our anonymous referee have made substantial
613: improvements to this paper. The James Clerk Maxwell Telescope is
614: operated by The Joint Astronomy Centre on behalf of the Particle
615: Physics and Astronomy Research Council of the United Kingdom, the
616: Netherlands Organisation for Scientific Research, and the National
617: Research Council of Canada. IRAM is supported by INSU/CNRS (France),
618: MPG (Germany), and IGN (Spain). This material is based upon work
619: supported under a National Science Foundation Graduate Research
620: Fellowship.
621:
622: \begin{thebibliography}{}
623: \bibitem[Agladze et al.(1996)]{Agladze96} Agladze, N.~I., Sievers, A.~J.,
624: Jones, S.~A., Burlitch, J.~M., \& Beckwith, S.~V.~W.\ 1996, \apj,
625: 462, 1026
626: \bibitem[Barranco \& Goodman(1998)]{Barranco98} Barranco, J.~A., \&
627: Goodman, A.~A.\ 1998, \apj, 504, 207
628: \bibitem[Beckwith \& Sargent(1991)]{Beckwith91} Beckwith, S.~V.~W., \&
629: Sargent, A.~I.\ 1991, \apj, 381, 250
630: \bibitem[Bianchi et al.(2000)]{Bianchi00} Bianchi, S., Davies,
631: J.~I., Alton, P.~B., Gerin, M., \& Casoli, F.\ 2000, \aap, 353,
632: L13
633: \bibitem[Bohlin et al.(1978)]{Bohlin78} Bohlin, R.~C., Savage,
634: B.~D., \& Drake, J.~F.\ 1978, \apj, 224, 132
635: \bibitem[Draine \& Lee(1984)]{Draine84} Draine, B.~T., \& Lee,
636: H.~M.\ 1984, \apj, 285, 89
637: \bibitem[Dupac et al.(2002)]{Dupac02} Dupac, X., et al.\ 2002,
638: \aap, 392, 691
639: \bibitem[Dupac et al.(2001)]{Dupac01} Dupac, X., et al.\ 2001,
640: \apj, 553, 604
641: \bibitem[Ellis(2005)]{Ellis05} Ellis, M.\ 2005, Experimental
642: Astronomy, 19, 169
643: \bibitem[Evans et al.(2003)]{Evans03} Evans, N.~J., II, et al.\
644: 2003, \pasp, 115, 965
645: \bibitem[Evans et al.(2001)]{Evans01} Evans, N.~J., II,
646: Rawlings, J.~M.~C., Shirley, Y.~L., \& Mundy, L.~G.\ 2001, \apj,
647: 557, 193
648: \bibitem[Gon{\c c}alves et al.(2004)]{Goncalves04} Gon{\c c}alves,
649: J., Galli, D., \& Walmsley, M.\ 2004, \aap, 415, 617
650: \bibitem[Goodman et al.(1998)]{Goodman98} Goodman, A.~A.,
651: Barranco, J.~A., Wilner, D.~J., \& Heyer, M.~H.\ 1998, \apj, 504,
652: 223
653: \bibitem[Goodman et al.(1993)]{Goodman93} Goodman, A.~A., Benson,
654: P.~J., Fuller, G.~A., \& Myers, P.~C.\ 1993, \apj, 406, 528
655: \bibitem[G{\"u}sten et al.(2006)]{Gusten06} G{\"u}sten, R.,
656: Nyman, L.~{\AA}., Schilke, P., Menten, K., Cesarsky, C., \& Booth,
657: R.\ 2006, \aap, 454, L13
658: \bibitem[Holland et al.(1999)]{Holland99} Holland, W.~S., et al.\
659: 1999, \mnras, 303, 659
660: \bibitem[Johnstone et al.(2000)]{Johnstone00} Johnstone, D.,
661: Wilson, C.~D., Moriarty-Schieven, G., Joncas, G., Smith, G.,
662: Gregersen, E., \& Fich, M.\ 2000, \apj, 545, 327
663: \bibitem[Kenyon et al.(1994)]{Kenyon94} Kenyon, S.~J.,
664: Dobrzycka, D., \& Hartmann, L.\ 1994, \aj, 108, 1872
665: \bibitem[Kreysa et al.(1999)]{Kreysa99} Kreysa, E., Gem\"und, H.-P.,
666: Gromke, J.\ 1999 Infrared Phys. Techn. 40, 191
667: \bibitem[Mannings \& Emerson(1994)]{Mannings94} Mannings, V., \&
668: Emerson, J.~P.\ 1994, \mnras, 267, 361
669: \bibitem[Mathieu et al.(1995)]{Mathieu95} Mathieu, R.~D., Adams,
670: F.~C., Fuller, G.~A., Jensen, E.~L.~N., Koerner, D.~W., \& Sargent,
671: A.~I.\ 1995, \aj, 109, 2655
672: \bibitem[Mathis(1990)]{Mathis90} Mathis, J.~S.\ 1990, \araa, 28, 37
673: \bibitem[Mennella et al.(1998)]{Mennella98} Mennella, V., Brucato,
674: J.~R., Colangeli, L., Palumbo, P., Rotundi, A., \& Bussoletti, E.\
675: 1998, \apj, 496, 1058
676: \bibitem[Ossenkopf \& Henning(1994)]{Ossenkopf94} Ossenkopf, V., \&
677: Henning, T.\ 1994, \aap, 291, 943
678: \bibitem[Pierce-Price et al.(2000)]{Pierce-Price00} Pierce-Price, D.,
679: et al.\ 2000, \apjl, 545, L121
680: \bibitem[Reid \& Wilson(2005)]{Reid05} Reid, M.~A., \& Wilson,
681: C.~D.\ 2005, \apj, 625, 891
682: \bibitem[Schnee \& Goodman(2005)]{Schnee05a} Schnee, S., \&
683: Goodman, A.\ 2005, \apj, 624, 254
684: \bibitem[Stamatellos et al.(2004)]{Stamatellos04} Stamatellos, D.,
685: Whitworth, A.~P., Andr{\'e}, P., \& Ward-Thompson, D.\ 2004, \aap,
686: 420, 1009
687: \bibitem[Stepnik et al.(2003)]{Stepnik03} Stepnik, B., et al.\
688: 2003, \aap, 398, 551
689: \bibitem[Zucconi et al.(2001)]{Zucconi01} Zucconi, A., Walmsley,
690: C.~M., \& Galli, D.\ 2001, \aap, 376, 650
691:
692: \end{thebibliography}
693:
694: \clearpage
695:
696: \begin{figure}
697: \centerline{\includegraphics[totalheight=0.8\textheight]{f1.eps}}
698: \caption{The true and calculated modified blackbody spectra of material
699: with $A_V = 50$, $\beta = 1.5$ and $T_d = 10$ K, observed with a
700: 14\arcsec\ beam. The black curve shows the true spectrum, and
701: the black crosses show the flux at 450, 850 or 1200 \micron. The
702: blue and red crosses show the flux overestimated or underestimated,
703: respectively, by 20\%. The blue and red curves show the spectrum
704: of dust that passes through the blue/red cross and the two black
705: crosses. The dust temperature, emissivity spectral index and
706: column density that would be derived from the three data points
707: are also shown.
708: \label{EXSPEC}}
709: \end{figure}
710:
711: \clearpage
712:
713: \begin{figure}
714: \centerline{\includegraphics[width=4.2in]{f2.xv.ps}}
715: \caption{The derived emissivity spectral index ($\beta$), the ratio of
716: the derived column density to the true column density and the
717: derived temperature as functions of the ratio of the measured flux
718: to the true flux at one wavelength (450, 850 or 1200 \micron), with
719: the other two fluxes measured without error. The black and red
720: curves show the results for dust with $T_d = 10$ K and $T_d = 15$ K,
721: respectively. $\beta = 1.5$ in both cases. The curves are cut off
722: where the derived temperature is greater than 100 K.
723: \label{ONEFLUX}}
724: \end{figure}
725:
726: \clearpage
727:
728: \begin{figure}
729: \centerline{\includegraphics[width=4.0in]{f3.xv.ps}}
730: \caption{Each panel shows the 1$\sigma$ spread in the derived column
731: density, temperature or emissivity spectral index as a function
732: of the percent noise in all three flux maps at 450, 850 and
733: 1200 \micron. A 5\% error in the measured flux corresponds to a
734: signal to noise of 20. The black line shows the median value
735: for the derived parameter, and the greyed area shows the 1 $\sigma$
736: spread in that parameter for dust at 10 K. The blue (dotted)
737: and red (dashed) lines show the median value and 1 $\sigma$
738: spread for dust at 15 and 20 K, respectively, and each panel is
739: repeated for emissivity spectral index ($\beta$ = 1.0, 1.5 or
740: 2.0). $N_H$, $T_d$ and $\beta$ are derived as in Section
741: \ref{EQSOLVE}. The column density is plotted as the ratio of
742: the derived $A_V$ to the input $A_V$, and the temperature and
743: emissivity spectral index are plotted as the difference between
744: the derived parameter and the true value.
745: \label{ALLTHREE}}
746: \end{figure}
747:
748: \clearpage
749:
750: \begin{figure}
751: \centerline{\includegraphics[width=4.5in]{f4.xv.ps}}
752: \caption{Same as Figure \ref{ALLTHREE}, except $N_H$, $T_d$ and $\beta$
753: are fit from {\it four} observations at 350, 450, 850 and 1200
754: \micron.
755: \label{ALLFOUR}}
756: \end{figure}
757:
758: \clearpage
759:
760: \begin{figure}
761: \centerline{\includegraphics[width=4.5in]{f5.xv.ps}}
762: \caption{Each panel shows the 1 $\sigma$ spread in the derived $T_d$,
763: $\beta$ and $N_H$ for dust at $T_d = 15$ K and $\beta = 1.5$
764: from a least squares fit between the predicted and given fluxes at
765: 450, 850 and 1200 \micron. The grey area uses the correct
766: assumption that $\beta = 1.5$ or $T_d = 15$, while the blue
767: (dotted) and red (dashed) lines incorrectly assume that
768: $\beta = 1.0$ and $\beta = 2.0$, or $T_d = 10$ and $T_d = 20$,
769: respectively.
770: \label{FITFIX}}
771: \end{figure}
772:
773: \clearpage
774:
775: \begin{figure}
776: \centerline{\includegraphics[totalheight=0.7\textheight]{f6.xv.ps}}
777: \caption{The top row of panels shows the temperature and column density
778: maps of a model starless core similar to the central 2\arcmin\
779: of TMC-1C. The second row shows the predicted
780: emission maps at 450, 850, and 1200 \micron, using a constant
781: $\beta = 1.5$. We then add Gaussian random noise to the
782: emission maps equal to the measured noise and artifacts in our
783: 450, 850 and 1200 \micron\ maps of TMC-1C (80, 9 and 3 mJy,
784: respectively) and show the resultant fluxes in the third row.
785: Using the altered fluxes, we attempt to solve for $\beta$, $T_d$
786: and $N_H$ at each point in the map, and show the results in the
787: fourth row from the top. When we use the value $\beta = 1.5$
788: everywhere along with the altered 450, 850 and 1200 \micron\
789: fluxes to fit $T_d$ and $N_H$, we derive the maps shown in
790: the bottom row.
791: \label{THREECORE}}
792: \end{figure}
793:
794: \clearpage
795:
796: \begin{figure}
797: \centerline{\includegraphics[width=5.5in]{f7.xv.ps}}
798: \caption{The 450, 850 and 1200 \micron\ emission maps of TMC-1C, along
799: with the derived $\beta$, $T_d$ and $N_H$ (expressed in units
800: of $A_V$ for convenience). The fluxes are in units of mJy per
801: 14\arcsec\ pixel. The derived parameters are shown within the
802: white 850 \micron\ contour. The (0,0) position is at
803: RA=4:41:35.8 DEC=+26:00:42.5 (J2000). Position 'a' is strong
804: at 450 \micron\ but not at 850 or 1200 \micron. Position 'b'
805: is strong at 850 and 1200 \micron\ but not at 450 \micron\, and
806: position 'c' is strong in all three maps.
807: \label{TMC1CFIG}}
808: \end{figure}
809:
810: \clearpage
811:
812: \begin{figure}
813: \centerline{\includegraphics[width=4.0in]{f8.xv.ps}}
814: \caption{The TMC-1C dust color temperature (top) and column density
815: (bottom), derived from a fit to the 450, 850 and 1200 \micron\
816: emission maps, assuming that the emissivity spectral index is
817: constant at $\beta = 1.8$.
818: \label{TMC1CTDAV}}
819: \end{figure}
820:
821: \clearpage
822:
823: \begin{figure}
824: \centerline{\includegraphics[width=4.0in]{f9.eps}}
825: \caption{The density (top) and temperature (bottom) profiles of TMC-1C.
826: The points in red are determined by a fit to the 450, 850 and
827: 1200 \micron\ data, assuming $\beta = 1.8$ and $\kappa_{230} =
828: 0.009$ and the black points are determined using only the 450
829: and 850 \micron\ data assuming $\beta = 1.5$, as in
830: \citet{Schnee05a} and $\kappa_{230} = 0.005$, as assumed here.
831: In both cases, we assume that the central 2 arcminute radius of
832: TMC-1C can be modeled as nested spherical shells of uniform
833: density and temperature.
834: \label{PROFILES}}
835: \end{figure}
836:
837: \end{document}
838:
839: