1: \documentstyle[12pt,epsf,colordvi,aaspp4]{article}
2: %\documentstyle[12pt,aasms4]{article}
3:
4: \begin{document}
5:
6: \title{Good views of the Galaxy}
7:
8: \author{Hiranya V. Peiris}
9: \affil{Princeton University Observatory, Peyton Hall, Princeton, NJ 08544;
10: hiranya@astro.princeton.edu}
11:
12: \begin{abstract}
13:
14: Fitting Galactic structure models to star counts only provides useful
15: information about the Galaxy in some directions. In this paper, we
16: investigate the use of $\chi^2$ goodness-of-fit tests to discriminate
17: between degenerate Galactic structure models, and the implications of this
18: technique for the Galactic spheroid and thick disk components. The axis
19: ratio of the Galactic spheroid and the normalization of spheroid stars
20: with respect to disk stars introduce a degenerate effect which means that
21: Galactic structure models with certain combinations of these parameters
22: are indistinguishable from each other in most directions. We present an
23: analysis of the optimal directions in which these degeneracies can be
24: lifted. Poisson and magnitude errors are taken into account, and an
25: attempt is made to place an upper limit on the systematic error due to
26: separation of spheroid stars from thick/old disk stars. We find that the
27: magnitude range $20 < V < 21$ is the best for lifting most degeneracies,
28: and present the optimal combinations of directions using which this can be
29: achieved. We also give directions in which the signature of the presence
30: of a Galactic thick disk can be most readily identified, and the
31: directions in which contamination from a thick disk can be minimized. It
32: is hoped that forthcoming data from large-scale sky surveys would reveal
33: much about the structure of our Galaxy using star count techniques.
34:
35: \end{abstract}
36:
37: \keywords{Galaxy: structure --- stars: statistics}
38:
39: % ------------------------------------------------------------------------------
40:
41: \section{Introduction} \label{intro}
42:
43: Since we observe our Galaxy from within it, we must use indirect tools
44: such as star counts to probe its structure. The success of this method
45: relies on assuming that the components of the Milky Way Galaxy have stars
46: distributed similarly to those of galaxies of the same Hubble type, and
47: comparing models with observations only in regions where little
48: obscuration is known to be present. Observed ``star counts'' (i.e. the
49: distribution of number counts of stars in apparent magnitude and color
50: within a given area/direction in the sky) are then used to determine
51: parameters in distribution functions whose overall shapes are assumed
52: known.
53:
54: Within the past two decades, Galactic structure models of varying degrees
55: of complexity have been developed (Bahcall \& Soneira, 1980, 1981, 1984,
56: \cite{gil84}, Robin and Cr\'ez\'e 1986a, 1986b, \cite{rei93}). However,
57: during the same period, relatively few sets of observational star count
58: data have been published.
59:
60: Galactic structure models, for a given set of structure parameters,
61: typically predict differential number counts (stars / apparent magnitude
62: bin / area) and frequency-color distributions (stars / color bin / area)
63: for a specified direction. Because the Galaxy consists of at least two
64: components, obeying different density laws and being characterized by
65: different stellar populations, these distributions change with the color
66: range, magnitude range and direction under consideration. It is important
67: to match both the shapes of these distributions and the absolute numbers
68: of stars with observations.
69:
70: Existing models predict differential number counts reasonably well, to
71: faint magnitudes of $V \sim 21$, but recent studies suggest that agreement
72: between observed and predicted frequency-color distributions breaks down
73: at faint magnitudes below $V \sim 19$ (\cite{rei93}). ``Standard''
74: Galactic structure models appear to overestimate the contribution of
75: (blue) spheroid stars substantially, and the contribution of (red) disk
76: stars is correspondingly underestimated in order to match the total number
77: counts. Contrary to prediction, observations show that {\it disk} stars
78: are the majority population to at least $V \sim 21$. However, the small
79: sizes of fields in which data are available mean that comparisons in
80: frequency-color space are complicated by uncertainties due to small-number
81: statistics.
82:
83: A further hindrance to the determination of distribution function
84: parameters arises due to the fact that data are only available in a small
85: number of directions. This presents a problem in trying to fit model
86: parameters which cause degenerate effects in the number counts. For
87: instance, increasing the normalization of the Galactic spheroid and
88: decreasing its axis ratio \slantfrac{b}{a} (or vice versa) represents a
89: degeneracy. This means that in most directions, one cannot distinguish
90: between models with a flattened spheroid plus a low normalization ratio of
91: spheroid stars to disk stars, and those with a high axis ratio plus a high
92: normalization.
93:
94: We anticipate the availability of star count data from large-scale, deep
95: sky surveys in the near future (see \S~\ref{dirn} and \S~\ref{finish})
96: which would solve the problem of excessive statistical noise, and provide
97: data in almost any direction we desire. In this paper, we investigate the
98: possibility of tackling the degeneracy problem by attempting to find two
99: or more directions in which $\chi^2$ goodness-of-fit tests can distinguish
100: between spheroid parameter combinations that yield degenerate results in
101: general.
102:
103: A degenerate situation is defined as one in which, in a general direction,
104: there is no significant difference in results given by a model where a
105: pair of spheroid parameters are characterized by values $(p, q)$ and a
106: model characterized by values $(r, s)$. This means that one cannot learn
107: very much about these spheroid parameters even if these models fit the
108: spheroidal star counts in that direction with reasonable accuracy. The
109: degenerate parameters we examine here are the spheroid axis ratio and the
110: normalization of spheroid stars to disk stars in the solar neighborhood.
111:
112: By using $\chi^2$ tests with a specified set of directions as ``bins"
113: (see e.g. \cite{pre92}), we have investigated the optimal directions for
114: breaking this degeneracy. These tests are based on fitting the total
115: number of spheroid stars in each direction ``bin'', for a given set of
116: directions and a particular magnitude range. Therefore, both the observed
117: number count and color information is used, since the spheroid counts have
118: to be separated from the total counts using the frequency-color
119: distribution, as explained in \S~\ref{seperr}.
120:
121: \S~\ref{model} describes the Galactic structure models we have used in
122: these tests. Details of the directions used in the analysis are found in
123: \S~\ref{dirn}. A description of the $\chi^2$ tests, along with our
124: results, are presented in \S~\ref{chisq}. We conclude, in \S~\ref{finish},
125: that there are two or more directions in which most ``degenerate" model
126: pairs can be distinguished. We find the magnitude range $20 < V < 21$ to
127: be the most useful for this purpose. We also make suggestions for the best
128: directions in which to separate spheroidal stars from old- and thick-disk
129: stars using a frequency-color distribution diagram, and conversely, the
130: directions in which the presence of a thick disk would be most clearly
131: observed.
132:
133: % ------------------------------------------------------------------------------
134:
135: \section{The Models} \label{model}
136:
137: While it is our intention to illustrate the use of this technique in
138: general, for specificity we will use the model that is usually referred to
139: in the literature as the Bahcall-Soneira (B\&S) model. We have taken the
140: realization of the B\&S model in the form of the Export Code that is
141: available on the Internet
142: (http://www.sns.ias.edu/\verb+~+jnb/Html/galaxy.html). This model is
143: described in Bahcall (1986). We have added a few refinements as described
144: below.
145:
146: The disk luminosity function (LF) has been modified between $9.5 < M_V
147: \leq 18.0$ to match the LF recently derived from {\it Hubble\ Space\
148: Telescope} ({\it HST}) M dwarfs (Figure 2, \cite{gou97}). A ``sech$^2$"
149: model has been used for the vertical density distribution function of main
150: sequence disk stars in the same absolute magnitude range, as given in Eq.
151: 3.2 of the same work. The HST LF was derived using two different methods:
152: a maximum likelihood (ML) fit that takes into account both measurement
153: errors and Malmquist bias, and a naive binning method where the total
154: number of stars in a given magnitude bin is divided by the effective
155: volume integrated over that bin. This binning method takes into account
156: neither Malmquist bias nor observational error. The two alternative
157: derivations agree quite well over most of the LF, but at the faint end
158: ($13.5 \leq M_V \leq 18.5$) the ML procedure could by affected by a
159: statistical fluctuation. Because Poisson errors are potentially a much
160: more serious problem at the faint end than Malmquist bias (since the HST
161: survey extends to the ``top'' of the disk), we choose to use the form of
162: the LF derived using the binning method.
163:
164: The spheroid LF has been modified between $7.5 \leq M_V \leq 13.5$ to
165: match another LF derived from {\it HST} star counts (Figure 3,
166: \cite{gou98}). Note that this LF differs in its zero point by a factor of
167: $\sim 2$ with studies of kinematically-selected local spheroid subdwarfs
168: (\cite{dah95}).
169:
170: The disk and spheroid LFs used are shown in Figure~\ref{fig-lf}. In
171: addition to the spheroidal parameters whose values are described in
172: \S~\ref{chisq}, the color-magnitude diagram (CMD) of M13 was adopted for
173: the spheroid CMD.
174:
175: \placefigure{fig-lf}
176:
177: In addition to this two-component model, a three-component model was
178: created with the addition of a thick disk with a scale height of 1200 pc,
179: a 47 Tuc-like CMD, a disk-like LF, and a normalization to disk counts at
180: the solar neighborhood of 2\%. This was used in analyzing the separation
181: error in identifying spheroid counts from disk/thick-disk counts using
182: frequency-color diagrams, as described in \S~\ref{seperr}.
183:
184: The density laws used for the old disk, the spheroid and the thick disk
185: are summarized in Table~\ref{tbl-dist}, which also gives the
186: normalizations used.
187:
188: \placetable{tbl-dist}
189:
190: % ------------------------------------------------------------------------------
191: \section{The Directions} \label{dirn}
192:
193: We selected a representative set of directions in which to carry out the
194: analysis, which are shown in Table~\ref{tbl-dir}.
195:
196: \placetable{tbl-dir}
197:
198: Directions 1 \& 2 were selected following a quantitative analysis of
199: directions in the $(l^{II} = 0\arcdeg \rightarrow 180\arcdeg)$ and
200: $(l^{II} = 90\arcdeg \rightarrow 270\arcdeg)$ planes, as the optimal
201: directions in which to distinguish between the effects of the
202: normalization and the axis ratio of the spheroid. Bahcall and Soneira
203: (1980) explain why these planes are especially important for determining
204: the characteristics of the spheroid. We ran the model at $5\arcdeg$
205: intervals in $b^{II}$ in each of these planes, and calculated the
206: variation with $b^{II}$ of the ratio of spheroid counts to disk counts,
207: and the ratio of spheroid counts for pairs of near-degenerate models: for
208: example, using the notation (disk : spheroid\ normalization\ ratio,
209: spheroid\ axis\ ratio), $(800:1, 1.0):(500:1, 0.8)$. We then looked for
210: directions with a high spheroid : disk ratio (in order to minimize the
211: Poisson error in total spheroid counts as well as the error due to
212: contamination of spheroid counts by disk stars) while at the same time,
213: maximizing the quantity $\mid (800:1, 1.0):(500:1, 0.8) - 1\mid$.
214: Directions 1 \& 2 optimized both these quantities.
215:
216: Directions 3--10 were selected because fields in these directions are
217: being prepared by the Digital Palomar Observatory Sky Survey (DPOSSII; S.
218: G. Djorgovski \& S. C. Odewahn 1998, private communication; Djorgovski et
219: al. 1997, 1998) for star count studies. It would then be possible to see
220: how the techniques discussed here in a theoretical sense work in practice.
221:
222: % ------------------------------------------------------------------------------
223:
224: \section{$\chi^2$ Analysis} \label{chisq}
225:
226: The $\chi^2$ statistic is
227: \begin{equation} \label{chieq}
228: \chi^2 = \sum_{dir=1}^{N} \frac{\left[A_{model}(m_1,m_2,l,b)d\Omega -
229: A_{observed}(m_1,m_2,l,b)d\Omega \right]^2}{\sigma^2},
230: \end{equation}
231:
232: where $dir=(l,b)$ is a given direction, $A(m_1,m_2,l,b)d\Omega$ is the
233: total number of spheroid stars with apparent magnitudes in the range $m_1
234: \leq m \leq m_2$ in the direction $(l,b)$ per projected sky area
235: $d\Omega$, and $\sigma$ is the error, which can have several components as
236: described below. Reduced $\chi^2$ is given by $\chi^2/\nu$ where $\nu$ is
237: the number of degrees of freedom, in this case equal to the number of
238: directions $N$ (i.e. the total predicted counts are {\it not} renormalized
239: to the same total counts as the data). A large value of reduced $\chi^2$
240: indicates that the null hypothesis that the $A_{observed}$'s are drawn
241: from the same population as the $A_{model}$'s is rather unlikely.
242:
243: Since we do not currently have observed number counts in all the
244: directions, we picked one model out of the set covering the investigated
245: parameter-space to be the ``observed" number counts and compared it
246: against each of the rest of the models in the set (which represented the
247: ``theory"). The process was then repeated till each of the models had been
248: picked as the one representing actual data.
249:
250: We define the criterion for lifting the degeneracy of spheroid parameters
251: as obtaining the probability of finding a value of reduced $\chi^2$
252: greater than or equal to the ``observed" value $<$ 10$^{-5}$. That is, we
253: aim to reject the null hypothesis (that the ``observed" and ``theoretical"
254: models are the same) at a confidence level of 0.001\%. The range of models
255: we used spanned the parameter space \((disk : spheroid\ normalization\
256: ratio) \times (spheroid\ axis\ ratio) = (500:1, 800:1) \times (1.0, 0.8,
257: 0.6, 0.4)\). Thus there are 56 possible pairings of models picked as
258: "observed" and "theoretical".
259:
260: $\chi^2$ tests were carried out using the set of directions detailed in
261: \S~\ref{dirn} as ``bins." The directions yielding the highest contribution
262: to the $\chi^2$ value were noted as the optimal directions to use for
263: breaking the degeneracy. It was then checked whether $\chi^2$ tests using
264: just those optimal directions as bins would cause the degeneracy to be
265: lifted.
266:
267: The error term $\sigma$ can have contributions from Poisson error,
268: magnitude error, and the systematic ``separation error'' which arises
269: because spheroid stars cannot be separated cleanly from old/thick disk
270: stars in observations, thereby introducing an error into the total
271: spheroid counts. In general, a magnitude error of order $\sim 0.1^{mag}$
272: contributes less to the error budget than the Poisson error. In a ``good''
273: direction, the separation error should contribute the least to the error
274: budget and the number counts are Poisson-noise-limited. In a ``bad''
275: direction, the separation error can be by far the greatest source of
276: uncertainty. In \S~\ref{nosep}, we first consider the case where the
277: spheroid counts {\it{can}} be cleanly separated, and then in
278: \S~\ref{sepanal} we consider the effect of adding the separation error.
279:
280: \subsection{Analysis excluding separation error} \label{nosep}
281:
282: $\chi^2$ tests were first carried out using the two-component (spheroid +
283: old disk) model, taking into account the Poisson error of the counts and a
284: magnitude error of 0.1$^{mag}$, but assuming that spheroid stars can be
285: separated cleanly from old/thick disk stars in observations.
286:
287: The magnitude error was calculated as follows. Let $N(m_1 \leq m \leq
288: m_2)$ represent the total spheroidal number counts for apparent magnitude
289: range $m_1 \leq m \leq m_2$, and let the error in the zero-point be
290: $\delta m$. Then we calculate $N_+(m_1+\delta m \leq m \leq m_2+\delta m)$
291: and $N_-(m_1-\delta m \leq m \leq m_2-\delta m)$. The error in the total
292: spheroidal number counts due to the zero-point error in magnitude is then
293: given by $Max(\mid N-N_+ \mid, \mid N-N_- \mid)$.
294:
295: \subsubsection{Directions 1 \& 2}
296:
297: Directions 1 and 2 in Table~\ref{tbl-dir} were found to be the best
298: overall for discriminating between ordinarily degenerate models using
299: $\chi^2$ tests.
300:
301: Using the notation $[(model\ taken\ as\ observation),\ (model\ taken\ as\
302: prediction)]$, almost all model pairings were distinguishable from each
303: other in the apparent magnitude range ${19 < V < 20}$. The exceptions were
304: $[(500:1,0.8),\ (800:1,1.0)]$ and $[(800:1,1.0),\ (500:1,0.8)]$. Upon
305: examining the range ${20 < V < 21}$ the degeneracy of these model pairings
306: were also lifted, while the other model pairings also remained
307: non-degenerate.
308:
309: \subsubsection{Directions 3--10 $(19 < V < 20)$}
310:
311: Considering the directions {3--10} which are being prepared by the DPOSSII
312: survey, $\chi^2$ tests in directions 3 and 4 in apparent magnitude range
313: ${19 < V < 20}$ can distinguish between all model pairings with the
314: exceptions shown in Table~\ref{tbl-chi1}.
315:
316: \placetable{tbl-chi1}
317:
318: \subsubsection{Directions 3--10 $(20 < V < 21)$}
319:
320: Considering the model pairings in the apparent magnitude range ${20 < V <
321: 21}$ in the DPOSSII directions, $\chi^2$ tests can distinguish between all
322: model pairings in directions 3 and 4 except for those shown in
323: Table~\ref{tbl-chi2}.
324:
325: \placetable{tbl-chi2}
326:
327: \subsubsection{Discussion} \label{seperr}
328:
329: Observationally, spheroid stars can be separated from old/thick disk stars
330: using the bi-modal distribution that appears in star color-frequency
331: profiles in certain directions at faint magnitude ranges. The blue and red
332: peaks consist of spheroid and disk stars, respectively. This distribution
333: has been interpreted as arising because the disk and spheroid components
334: have different density gradients in a magnitude-limited survey
335: (\cite{bs80}). The sharp density gradient in the disk for directions far
336: from the Galactic plane favors relatively nearby and intrinsically faint
337: (red) stars at faint apparent magnitudes. Conversely, the shallow density
338: gradient in the spheroid favors relatively distant, intrinsically bright
339: (blue) stars at the same faint apparent magnitudes, because the effective
340: volume increases at large distances.
341:
342: $18 \leq V \leq 22$ is the apparent magnitude range where this
343: double-peaked distribution is most pronounced, enabling the cleanest
344: separation of spheroidal stars in observations (\cite{bs80},
345: \cite{rei93}). In this interval, the spheroid counts peak at $B-V \sim
346: 0.5$, with a narrow range in color. This corresponds roughly to the
347: main-sequence turn-off, at $M_V \sim 4.5$. For a given apparent magnitude
348: interval $m_1 \leq m \leq m_2$, only stars within a range of absolute
349: magnitudes $M_1 \leq M \leq M_2$ contribute to the observed number counts.
350: The majority contribution to the blue spheroidal peak comes from stars
351: near the main sequence turn-off, with $+4 \leq M_V +6$ (corresponding to
352: $\sim 0.5 \rightarrow 1.0$ M$_\odot$). Evolved stars with $M_V \leq 3.5$
353: make negligible contribution to deep star counts at this magnitude range.
354:
355: The shallow density gradient in the spheroid leads to a broad distribution
356: of number counts with distance in a given direction, peaking at $\sim 7$
357: kpc in the direction of the North Galactic Pole (\cite{bs80}). Therefore,
358: the fainter the magnitude range under consideration, the better the
359: performance of the $\chi^2$ statistic, because spheroid number counts are
360: still rising at $V \sim 22$. Thus the Poisson errors and magnitude errors
361: at fainter magnitudes are smaller. This is the reason that the magnitude
362: range $20 \leq V \leq 21$ is more useful for lifting degeneracies using
363: the $\chi^2$ tests than the brighter range $19 \leq V \leq 20$. Once we go
364: even fainter to the range $21 \leq V \leq 22$, photometric uncertainties
365: and observational errors, such as star-galaxy separation and contamination
366: of the blue spheroidal population by quasars and compact emission line
367: galaxies (CELGs) increasingly start to affect counts. Thus the
368: intermediate apparent magnitude range $20 \leq V \leq 21$ is the optimal
369: one to use for the purpose of lifting degeneracies by this method.
370:
371: \subsection{Analysis including separation error} \label{sepanal}
372:
373: Figures~\ref{fig-19} and~\ref{fig-20} show frequency distributions of star
374: colors for a particular ``standard" model, in the ten directions given in
375: Table~\ref{tbl-dir}, for magnitude ranges ${19 < V < 20}$ and ${20 < V <
376: 21}$. The maximum error in separating out the spheroid stars arises if
377: there is a thick disk component which tends to fill up the valley between
378: the double-peaks due to spheroid and old disk stars, or in the worst
379: cases, peaks within the blue spheroid peak. The separation error is
380: calculated as the area under the blue peak in total (spheroid + old disk +
381: thick disk) counts, minus the calculated (i.e. model) spheroid counts. The
382: ``blue peak'' is taken to be bounded on the red side by the color
383: corresponding to the minimum point of the total counts.
384:
385: \placefigure{fig-19}
386: \placefigure{fig-20}
387:
388: Now, we consider the effect of including the systematic error due to
389: separating spheroid stars from old/thick disk stars in observations, using
390: the three-component (spheroid + old disk + thick disk) model. The
391: separation error is an overestimate since this contamination in
392: observations can be corrected for to some extent by subtracting model disk
393: counts from the total counts.
394:
395: The contamination due to the thick disk was found to dominate that due to
396: old disk stars in all cases. This error was found to be smaller in the
397: apparent magnitude range ${20 < V < 21}$ than in the range ${19 < V <
398: 20}$. Hence the following analysis was only carried out for the fainter
399: range. Further, since the ratio of separation error to total spheroid
400: counts increases significantly for models with smaller axis ratios, the
401: models with axis ratio 0.4 were excluded from the analysis. Thus, model
402: parameter space was restricted to \((500:1, 800:1) \times (1.0, 0.8,
403: 0.6)\).
404:
405: Upon including the separation error, directions 1 and 2, found to be the
406: best overall in discriminating between model parameters, were found to be
407: virtually useless, since the thick disk counts peaked within the spheroid
408: peak for these directions.
409:
410: In the case of the DPOSSII directions (3--10), all the model pairings
411: which could be distinguished in this apparent magnitude range using
412: $\chi^2$ tests in directions 3 and 4 remained non-degenerate, except
413: $[(800:1,0.6),\ (500:1,0.6)]$, which became degenerate. The rest of the
414: model pairings which had been non-degenerate in directions given in
415: Table~\ref{tbl-chi2} also became degenerate.
416:
417: Table~\ref{tbl-chi3} shows whether these degenerate models can be
418: discriminated using other combinations of directions.
419:
420: \placetable{tbl-chi3}
421:
422: During this analysis, it was found directions 7 and 9 are generally good
423: for minimizing the effects due to the thick disk. Conversely, directions 1
424: and 10 are particularly good for looking for the presence of the thick
425: disk.
426:
427: %------------------------------------------------------------------------------
428: \section{Conclusions} \label{finish}
429:
430: We found that directions 1 ($l^{II} = 0\arcdeg, b^{II} = 40\arcdeg$) and 2
431: ($l^{II} = 90\arcdeg, b^{II} = 40\arcdeg$) in apparent magnitude range $20
432: < V < 21$ were the most effective in distinguishing between effects due to
433: degenerate parameters, assuming that spheroid stars could be separated
434: cleanly from disk and thick-disk stars.
435:
436: In the case that the maximum possible separation error (i.e. uncorrected
437: for disk and thick disk counts) was included in the analysis, directions 3
438: (the North Galactic Pole) and 4 ($l^{II} = 67\arcdeg, b^{II} = 49\arcdeg$)
439: in apparent magnitude range $20 < V < 21$ were the most useful. In those
440: models where these directions failed to lift the degeneracy, directions 7
441: ($l^{II} = 111\arcdeg, b^{II} = -46\arcdeg$) and 9 ($l^{II} = 172\arcdeg,
442: b^{II} = 48\arcdeg$) were the best for minimizing contamination from the
443: thick disk, which dominates the separation error.
444:
445: Directions 1 ($l^{II} = 0\arcdeg, b^{II} = 40\arcdeg$) and 10 ($l^{II} =
446: 61\arcdeg, b^{II} = -37\arcdeg$) were found to be the most effective in
447: detecting the presence of a thick disk.
448:
449: This is an exciting time in which deep, high-quality data-sets covering
450: large areas of the sky are about to become available from much-anticipated
451: surveys such as the Sloan Digital Sky Survey (SDSS, see e.g. \cite{gun93},
452: \cite{gun95}) and the DPOSSII (see \S~\ref{dirn} for references). The use
453: of survey data for star count studies is complicated by the fact that many
454: existing ``standard" Galactic structure models are based on
455: color-magnitude diagrams (CMDs) and luminosity functions (LFs) determined
456: in the standard Johnson-Morgan-Cousins $UBV$ photometric system, and are
457: only useful for comparison with data in the visual band or in photometric
458: systems to which accurate transformations exist. However, in cases such
459: as that of the SDSS, which uses the non-standard $u'g'r'i'z'$ photometric
460: system (\cite{fuk96}), such models will soon be re-created using LFs and
461: CMDs determined in the SDSS photometric bands. The DPOSSII uses three
462: photographic $JFN$ bands calibrated to the Gunn $gri$ bands. Though these
463: are different from the photoelectric/CCD $gri$ bands (Palomar, 4-Shooter,
464: SDSS) they are well-defined. Once the work of creating a model based on
465: the SDSS bands is completed, it should be fairly straightforward to
466: translate the stellar sequences and LFs to the DPOSSII $gri$ bands. Until
467: such models are available, a different approach to star count models, such
468: as the evolutionary stellar population synthesis technique (\cite{fan99})
469: can be used.
470:
471: The possibility of fitting models simultaneously in a multitude of
472: directions with large samples containing number, magnitude and color
473: information should revolutionize studies of Galactic structure using star
474: counts.
475:
476: % ------------------------------------------------------------------------------
477:
478: \acknowledgements
479:
480: The author is very grateful to John Bahcall for his helpful advice. She wishes
481: to thank Andy Gould for providing machine-readable versions of Figure 2 of
482: Gould et al.\ (1997) and Figure 3 of Gould et al.\ (1998); David Spergel and
483: Xiaohui Fan for instructive discussions on Galactic Structure models; and Mark Jackson for useful comments on the manuscript.
484:
485: % ------------------------------------------------------------------------------
486:
487: \clearpage
488: \begin{deluxetable}{ll}
489: \footnotesize
490: \tablecaption{The Assumed Stellar Distributions. \label{tbl-dist}}
491: \tablewidth{0pt}
492: \tablehead{
493: \colhead{Component} & \colhead{Distribution}
494: }
495: \startdata
496: Disk & $n_D = n_D (R_0) \exp [-z/H(M_V)] \exp [-(x-R_0)/h]$ \nl \nl
497: Spheroid & $n_{sph} =
498: n_{sph}(R_0)(R/R_0)^{-7/8}{ \exp [-10.093(R/R_0)^{1/4} + 10.093]}$ \nl
499: & $\times 125(R/R_0)^{-6/8} { \exp [-10.093(R/R_0)^{-1/4} + 10.093]},
500: \ R < 0.03 R_0$ \nl
501: & $\times [1-0.08669/(R/R_0)^{1/4}],\ R \geq 0.03 R_0$ \nl
502: \nl
503: Thick Disk & $n_{TD} = n_{TD} (R_0) \exp [-z/H_{TD}] \exp [-(x-R_0)/h]$ \nl
504: \nl
505: Normalization & $n_D(R_0) = 0.13 pc^{-3}, \ n_{TD}(R_0) = 0.0026 pc^{-3},$ \nl
506: & $n_{sph}(R_0) = 0.00026 pc^{-3}$ or $0.0001625 pc^{-3}$ depending on model
507: \nl
508: \enddata
509: \tablecomments{ Here $z$ is the distance perpendicular to the plane, $x$
510: is the galactocentric distance in the plane, and $h$ is the old disk
511: scale length. Galactocentric distance $R = (x^2 + z^2/ \kappa^2)^{1/2}$,
512: where $\kappa$ is the axis ratio and $1-\kappa$ is the ellipticity. We
513: adopt $R_0 = 8$ kpc and $h = 3.5$ kpc. The old disk scale height $H(M_V)$
514: is given in Bahcall and Soneira (1980). The thick disk is taken to have a
515: scale height $H_{TD} = 1.2$ kpc and a 47-Tuc-like CMD.}
516: \end{deluxetable}
517:
518: % ------------------------------------------------------------------------------
519:
520: \clearpage
521: \begin{deluxetable}{cr@{}lr@{}ll }
522: \footnotesize
523: \tablecaption{A Representative Set of Directions. \label{tbl-dir}}
524: \tablewidth{0pt}
525: \tablehead{
526: \colhead{ Direction } & \colhead{$l^{II}$(1950)} & \colhead{} &
527: \colhead{$b^{II}$(1950)} & \colhead{} &
528: \multicolumn{1}{c}{Comments\tablenotemark{a}}
529: }
530: \startdata
531: 1..... & 0 & \arcdeg & +40 & \arcdeg & BS9\tablenotemark{b} \nl
532: 2..... & 90 && +40 && Between BS4 and BS7\tablenotemark{b}\nl
533: 3..... & \nodata && +90 && Galactic pole (SA57, BS1)\tablenotemark{c,d} \nl
534: 4..... & 67 && +49 && BS16\tablenotemark{c,d} \nl
535: 5..... & 180 && +50 && BS6\tablenotemark{c} \nl
536: 6..... & 180 && +30 && BS8\tablenotemark{c} \nl
537: 7..... & 111 && --$\,$46 && BS11, SA68\tablenotemark{c} \nl
538: 8..... & 167 && --$\,$51 && BS13\tablenotemark{c} \nl
539: 9..... & 172 && +48 && BS14\tablenotemark{c} \nl
540: 10..... & 61 && --$\,$37 && BS17\tablenotemark{c} \nl
541: \enddata
542: \tablenotetext{a}{The field identification BS{\it n} denotes the
543: designation assigned to a given direction in Table 1 of Bahcall and
544: Soneira (1981), where quantities that can be efficiently studied in that
545: direction are listed. The Selected Area number is also given for fields
546: which have been studied.}
547: \tablenotetext{b}{These directions were determined to be the best for
548: distinguishing between effects due to spheroid flattening and
549: normalization using the two-component (old disk + spheroid) standard
550: model.}
551: \tablenotetext{c}{These fields are being made available by the DPOSSII
552: survey for star count studies.}
553: \tablenotetext{d}{These directions were determined to be the best for
554: determining the spheroid parameters out of the DPOSSII directions.}
555: \end{deluxetable}
556:
557: % ------------------------------------------------------------------------------
558:
559: \clearpage
560:
561: \begin{deluxetable}{cc}
562: \footnotesize
563: \tablecaption{Model pairings degenerate in directions 3 and 4 for ${19 < V <
564: 20}$ excluding separation error. \label{tbl-chi1}}
565: \tablewidth{0pt}
566: \tablehead{
567: \colhead{Model pairing\tablenotemark{a}} & \colhead{Min. directions required to
568: lift degeneracy}
569: }
570: \startdata
571: $[(500:1,0.8),\ (800:1,1.0)]$ & 6, 7 \nl
572: $[(800:1,1.0),\ (500:1,0.8)]$ & 6, 7, 10 \nl
573: $[(500:1,0.6),\ (800:1,0.8)]$ & \nodata\tablenotemark{b}
574: \nl
575: $[(800:1,0.8),\ (500:1,0.6)]$ & \nodata\tablenotemark{b}
576: \nl
577: $[(500:1,0.4),\ (800:1,0.6)]$ & \nodata\tablenotemark{b} \nl
578: $[(800:1,0.4),\ (500:1,0.4)]$ & 4, 10 \nl
579: \enddata
580: \tablenotetext{a}{Uses notation {\it [(model\ taken\ as\ observation),\ (model\
581: taken\ as\
582: prediction)]}.}
583: \tablenotetext{b}{This model pairing cannot be distinguished using the DPOSSII
584: directions 3--10 in the apparent magnitude range ${19 < V < 20}$.}
585: \end{deluxetable}
586:
587: % ------------------------------------------------------------------------------
588:
589: \clearpage
590:
591: \begin{deluxetable}{cc}
592: \footnotesize
593: \tablecaption{Model pairings degenerate in directions 3 and 4 for ${20 < V <
594: 21}$ excluding separation error. \label{tbl-chi2}}
595: \tablewidth{0pt}
596: \tablehead{
597: \colhead{Model pairing\tablenotemark{a}} & \colhead{Min. directions required to
598: lift degeneracy}
599: }
600: \startdata
601: $[(500:1,0.8),\ (800:1,1.0)]$ & 6, 10 \nl
602: $[(800:1,1.0),\ (500:1,0.8)]$ & 6, 10 \nl
603: $[(500:1,0.6),\ (800:1,0.8)]$ & 3, 6 \nl
604: $[(800:1,0.8),\ (500:1,0.6)]$ & \nodata\tablenotemark{b} \nl
605: \enddata
606: \tablenotetext{a}{Uses notation {\it [(model\ taken\ as\ observation),\ (model\
607: taken\ as\ prediction)]}.}
608: \tablenotetext{b}{This model pairing cannot be distinguished using the DPOSSII
609: directions 3--10 in the apparent magnitude range ${20 < V < 21}$.}
610: \end{deluxetable}
611:
612: % ------------------------------------------------------------------------------
613:
614: \clearpage
615:
616: \begin{deluxetable}{cc}
617: \footnotesize
618: \tablecaption{Model pairings degenerate in directions 3 and 4 for ${20 < V <
619: 21}$
620: including separation error. \label{tbl-chi3}}
621: \tablewidth{0pt}
622: \tablehead{
623: \colhead{Model pairing\tablenotemark{a}} & \colhead{Min. directions required to
624: lift degeneracy}
625: }
626: \startdata
627: $[(500:1,0.8),\ (800:1,1.0)]$ & 7, 9 \nl
628: $[(800:1,1.0),\ (500:1,0.8)]$ & \nodata\tablenotemark{b} \nl
629: $[(500:1,0.6),\ (800:1,0.8)]$ & \nodata\tablenotemark{b} \nl
630: $[(800:1,0.8),\ (500:1,0.6)]$ & \nodata\tablenotemark{b} \nl
631: $[(800:1,0.6),\ (500:1,0.6)]$ & 7, 9 \nl
632: \enddata
633: \tablenotetext{a}{Uses notation {\it [(model\ taken\ as\ observation),\ (model\
634: taken\ as\ prediction)]}.}
635: \tablenotetext{b}{This model pairing cannot be distinguished using the DPOSSII
636: directions 3--10 in the apparent magnitude range ${20 < V < 21}$ using
637: $\chi^2$ tests including the separation error.}
638: \end{deluxetable}
639:
640:
641: % ------------------------------------------------------------------------------
642:
643: \clearpage
644: \begin{thebibliography}{}
645: \bibitem[Bahcall \& Soneira\ 1980]{bs80} Bahcall, J. N., \& Soneira, R. M.
646: 1980, \apjs, 44, 73
647: \bibitem[Bahcall \& Soneira\ 1981]{bs81} Bahcall, J. N., \& Soneira, R.
648: M. 1981, \apjs, 47, 357
649: \bibitem[Bahcall \& Soneira\ 1984]{bs84} Bahcall, J. N., \& Soneira, R.M.
650: 1984, \apjs, 55, 67
651: \bibitem[Bahcall\ 1986]{bah86} Bahcall, J. N. 1986, \araa, 24, 577
652: \bibitem[Dahn et al.\ 1995]{dah95} Dahn, C. C., Liebert, J. W., Harris, H., \& Guetter, H. C. 1995, ESO Workshop on the Bottom of the Main Sequence and Beyond, ed. C. G. Tinney (Heidelberg: Springer), 239
653: \bibitem[Djorgovski et al.\ 1997]{djo97} Djorgovski, S. G., de Carvalho, R.
654: R., Odewahn, S. C., Gal, R. R., Roden, J., Storolz, P., \& Gray, A. 1997,
655: Applications of Digital Image Processing XX, ed. Tescher, proc.
656: S.P.I.E. 3164
657: \bibitem[Djorgovski et al.\ 1998]{djo98} Djorgovski, S. G., Gal, R. R.,
658: Odewahn, S. C., de Carvalho, R. R., Brunner, R., Longo, G., \&
659: Scaramella, R. 1998, to appear in Wide Field Surveys in Cosmology,
660: ed. Colombi \& Mellier, proc. XIV IAP Colloq., in press.
661: \bibitem[Fan 1999]{fan99} Fan, X. 1999, \aj, 117, 2528
662: \bibitem[Fukugita et al. 1996]{fuk96} Fukugita, M., Ichikawa, T., Gunn, J.
663: E., Doi, M., Shimasaku, K., \& Schneider, D. P. 1996, \aj, 111(4), 1748
664: \bibitem[Gilmore 1984]{gil84} Gilmore, G.F. 1984, \mnras, 207, 223
665: \bibitem[Gould et al.\ 1997]{gou97} Gould, A., Bahcall, J. N., \& Flynn, C.
666: 1997, \apj, 482, 913
667: \bibitem[Gould et al.\ 1998]{gou98} Gould, A., Flynn, C., \& Bahcall, J. N.
668: 1998, \apj, 503, 798
669: \bibitem[Gunn \& Knapp 1993]{gun93} Gunn, J. E., \& Knapp, G. R. 1993, in
670: ASP Conf. Ser. 43, Sky Surveys: Protostars to Protogalaxies, ed. Soifer, 267
671: \bibitem[Gunn \& Weinberg 1995]{gun95} Gunn, J. E., \& Weinberg, D. H.
672: 1995, in Wide Field Spectroscopy and the Distant Universe, ed.
673: Maddox \& Ara\`g on-Salamanca, (World Scientific, Singapore)
674: \bibitem[Press et al. 1992]{pre92} Press, W. H., Teukolsky, S. A.,
675: Vetterling, W. T., \& Flannery, B. P. 1992, Numerical Recipes in C
676: (2$^{nd}$ Edition; C.U.P.)
677: \bibitem[Reid \& Majewski\ 1993]{rei93} Reid, I. N., \& Majewski, S.R.
678: 1993, \apj, 409, 635
679: \bibitem[Robin \& Cr\'ez\'e 1986a]{rob86a} Robin, A., \& Cr\'ez\'e, M.
680: 1986a, A\&A, 157, 1
681: \bibitem[Robin \& Cr\'ez\'e 1986b]{rob86b} Robin, A., \& Cr\'ez\'e, M.
682: 1986b, A\&AS, 64, 53
683: \end{thebibliography}
684: % ------------------------------------------------------------------------------
685: \clearpage
686:
687: \begin{figure} \plotone{f1.ps} \caption{The adopted disk and spheroid
688: luminosity functions (LFs). The LFs adopted are the same form those
689: described in Bahcall (1986) except for the following refinements: the disk
690: LF for $M_V > 9.5$ has been modified to match smoothly the LF derived from
691: $HST$ star counts given in Figure 2 of Gould et al.\ (1997) and the
692: spheroid LF for $M_V > 7.5$ has been modified to match the LF given in
693: Figure 3 of Gould et al.\ (1998), also derived from $HST$ data.
694: \label{fig-lf}} \end{figure}
695:
696: % ------------------------------------------------------------------------------
697:
698: \clearpage
699:
700: \begin{figure}
701: \plotone{f2.ps}
702: \caption{Frequency distribution of star colors for a ``standard" model
703: $(500:1,0.8)$ (\cite{bah86}) plus a thick-disk as described in
704: \S~\ref{model}, in the apparent magnitude range $19 < V < 20$.
705: \label{fig-19}}
706: \end{figure}
707:
708: %------------------------------------------------------------------------------
709:
710: \clearpage
711:
712: \begin{figure}
713: \plotone{f3.ps}
714: \caption{Frequency distribution of star colors for a ``standard" model
715: $(500:1,0.8)$ (\cite{bah86}) plus a thick-disk as described in
716: \S~\ref{model}, in the apparent magnitude range $20 < V < 21$.
717: \label{fig-20}}
718: \end{figure}
719: \end{document}
720:
721: