ASTROMETRY OF INTERNATIONAL CELESTIAL REFERENCE FRAME SOURCES USING THE SECOND US NAVAL OBSERVATORY CCD ASTROGRAPH CATALOG

The Astronomical Journal, 129:2907–2913, 2005 June # 2005. The American Astronomical Society. All rights reserved. Printed in U.S.A. ASTROMETRY OF INTERNATIONAL CELESTIAL REFERENCE FRAME SOURCES USING THE SECOND US NAVAL OBSERVATORY CCD ASTROGRAPH CATALOG1 M. Assafin, P. T. Monken Gomes, and D. N. da Silva Neto, Observatório do Valongo, Universidade Federal do Rio de Janeiro, Rua Ladeira Pedro Antonio 43, CEP 20.080-090 Rio de Janeiro, RJ, Brazil; massaf@ov.ufrj.br A. H. Andrei and R. Vieira Martins2 Grupo e Estudos em Astronomia, Observatório do Valongo, Universidade Federal do Rio de Janeiro, Rua Ladeira Pedro Antonio 43, CEP 20.080-090 Rio de Janeiro, RJ, Brazil; and Observatório Nacional, Ministério da Ciência e Tecnologia, Rua Gal. José Cristino 77, São Cristóvão, 20921-400 Rio de Janeiro, RJ, Brazil; oat1@ov.ufrj.br and 3 J. I. B. Camargo, R. Teixeira, and P. Benevides-Soares Instituto de Astronomia, Geofı́sica, e Ciências Atmosféricas, Universidade de São Paulo, Rua do Matão, 1226 Cidade Universitária, CEP 05508-900 São Paulo, SP, Brazil; julio.camargo@obs.u-bordeaux1.fr Receivved 2004 November 10; accepted 2005 February 19 ABSTRACT We present results of a pilot investigation on the astrometry of International Celestial Reference Frame (ICRF) sources using small- to medium-sized telescopes and the second US Naval Observatory CCD Astrograph Catalog (UCAC2). For this purpose, 31 ICRF sources were observed, mostly south of the equator, during 1997–2000. We used the automated 0.6 and 1.6 m Cassegrain telescopes equipped with CCD detectors located at Laboratório Nacional de Astrofı́sica, Brazil. The source positions were referred to UCAC2, with fainter 0.6 m telescope stars serving as a reference frame to the reductions of the 1.6 m telescope CCD fields. Observations were made in the V band in a compromise between the 579–643 nm bandpass (between V and R) of UCAC2 and the bluer ICRF sources. To ensure that UCAC2, with its magnitude bandpass system, is a reliable reference catalog for our V-band CCD frame reductions, we have also compared it against an independent set of star positions with similar characteristics obtained in the V band with the Valinhos CCD Meridian Circle, Brazil. Average values and errors for the optical
radio position offsets using the 0.6 m telescope were +4  8 mas (41 mas) and +1  8 mas (42 mas) for right ascension and declination, respectively. (Parenthetical values refer to standard deviation, i.e., to the typical error of a single measurement given the quantity of sources.) For the 1.6 m telescope, offsets were
12  9 mas (45 mas) and +8  9 mas (46 mas). An expected random error that increases with magnitude and affects the positions of the fainter 0.6 m telescope secondary stars is verified. No systematic errors were found within the attained position precision, including differential color refraction. External comparisons with independent telescope/catalog sets of precise source positions were also made, showing consistent results within the respective errors. Key word: astrometry — catalogs — quasars: general — reference systems 1. INTRODUCTION forward the optical representation of the ICRS. Proper motions may then be derived for the first time for the fainter stars in these sky regions. Such programs contribute directly to the determination of the actual alignment between the HCRS and the ICRS through comparisons with the ICRF. In principle, because of the faint magnitudes of the ICRF sources, large instruments are needed for the task. However, large telescopes with time dedicated to astrometry are hard to find nowadays. On the other hand, according to reports of the IAU Commission 8 working group on ‘‘The Future Development on Ground-Based Astrometry,’’ the contribution of small- to medium-sized instruments to astronomy is growing (Stavinschi & Kovalevsky 2001). Dedicated small instruments are starting to play an important role in long-term photometry programs, such as monitoring variable stars, microlensing events, etc. The same can be said about astrometry programs concerning Galilean satellites, Earth-grazing objects, the semidiameter of the Sun, etc. In this context, an ongoing long-term observational program on the astrometry of southern ICRF sources is being carried out at Laboratório Nacional de Astrofı́sica (LNA), Brazil, using two Cassegrain telescopes of 0.6 and 1.6 m diameter equipped with CCD detectors; these are not specific, dedicated astrometric Since astrometric space missions will not produce scientific results until the next decade, ground-based astrometric observation programs remain important. One fundamental task is the alignment maintenance of the Hipparcos Catalog Reference System (HCRS; Rickman 2002) with respect to the International Celestial Reference System (ICRS). Another objective is the densification of the optical reference frame toward fainter magnitudes. Both these tasks have an impact on astrometry at other wavelengths, on astrometry with large and next-generation deep sky telescopes, and even on the framework of space missions themselves. In particular, programs for the optical observation of International Celestial Reference Frame (ICRF ) sources help push 1 Based on observations obtained at Laboratorio Nacional de Astrofı́sica ( LNA), Rua Estados Unidos, 154 Bairro das Nações, P.O. Box 21, 37500364 Itajubá, MG, Brazil. 2 Currently associated researcher at Institut de Mecanique Celeste et de Calcul des Ephemerides, Observatoire de Paris, 77 Avenue Denfert Rochereau, 75014 Paris, France. 3 Currently visitor at Observatoire de Bordeaux, 2 Rue de l’Observatoire, F-33270 Floirac, France. 2907 2908 ASSAFIN ET AL. instruments. Here we present results of a pilot investigation for 31 ICRF sources, mostly south of the equator. The second US Naval Observatory CCD Astrograph Catalog (UCAC2; Zacharias et al. 2004) is the primary reference frame for these observations. The goal of the investigation is to access the astrometric performance of these nonlarge telescopes in conjunction with UCAC2 in order to establish the prospects for the success of the aforementioned program. Observations of common fields made with the Valinhos CCD Meridian Circle (VCMC; Instituto de Astronomia, Geofı́sica, e Ciências Atmosféricas, Universidade de São Paulo) contributed to the investigation of the UCAC2 performance. In x 2 we describe the observations and reduction procedures. In xx 3 and 4 we discuss the results and draw overall conclusions. 2. OBSERVATIONS AND REDUCTIONS 2.1. LNA Telescopes: Observations and Reductions A total of 31 ICRF radio sources distributed in the range
90    þ20 were observed, mostly between 1997 and 2000. Observations were carried out with small- to medium-sized, nondedicated telescopes: two fully automated Cassegrain telescopes with diameters of 0.6 m (f /13.5, f ¼ 8:1 m) and 1.6 m (f /10, f ¼16 m), located at the LNA site (k ¼þ45 34 0 57 00 ,  ¼
22 32 0 04 00 , h ¼ 1870 m). The telescopes were equipped with thin, back-illuminated CCD detectors (1024 pixel ; 1024 pixel; pixels are of 24 m size), resulting in CCD fields of 5 0 ; 5 0 , 1 pixel ¼ 0B3 for the 1.6 m telescope and 10 0 ; 10 0 , 1 pixel ¼ 0B6 for the 0.6 m telescope. The reference frame for the 0.6 m telescope CCD field reductions was furnished by the UCAC2 catalog of compiled positions from observations made with the USNO Twin Astrograph, profiting from a red filter in the 579–642 nm spectral bandpass and corrected red lens, and many other early epoch catalogs (Zacharias et al. 2004). For the 1.6 m telescope, the reference frame came from the positions of fainter secondary stars derived from the 0.6 m telescope CCD field reductions. All observations were made with the V filter (Johnson system) as a compromise between the bandpass adopted for UCAC2 and the bluer ICRF radio sources. Each field was observed at least three times, with telescope shifts between exposures to avoid sampling the same pixels on the star images. All exposures were guided. Exposure times ranged from 90 up to 300 s, allowing efficient use of telescope time and signal-to-noise ratios between 3 and 5. The magnitude limit for the LNA site was estimated as V ¼ 21. The same sources were observed at each telescope on time intervals not longer than 2 yr, so as to avoid problems due to unknown proper motions on the fainter noncataloged stars. An incidental gap in the sample distribution at 3h   8h occurred. Measurement and reductions followed the procedures described in detail in Assafin et al. (1997a, 1997b). All identified objects in all CCD frames for each telescope had their (x, y) measurements determined by bidimensional Gaussian fits. An overlapping technique adapted from the method described in Benevides-Soares & Teixeira (1992) was applied to transform the (x, y) values of the observed fields into a single averaged (x, y) instrumental frame for each telescope. This is done in a fashion very similar to classical polynomial adjustments between measured and standard coordinates, but in this case the many (x, y) values of the same field are adjusted against each other in an iterative manner. After the (x, y) transformation, the ( ,  ) values were then obtained by the usual reduction methods, using a classical six-constant model to relate measured and standard coordinates with regard to the reference catalog. In the case of the 1.6 m telescope, the reference frame came from the 0.6 m telescope field reductions, formed by positions of secondary Vol. 129 field stars located within 5 0 ; 5 0 of the source position. In the case of the 0.6 m telescope, the reference catalog used was UCAC2. No proper motions were applied here because the time interval between observations and the UCAC2 average epoch is less than 2 yr. Magnitudes were computed on the basis of the fitting of image profiles and relate to the magnitude system of UCAC2 by means of a simple logarithmic relation and the adjustment of a zero point—they are not photometrically derived from flux measurements and should be used with care. Figure 1 shows the (x, y) measurement error estimates from the Gaussian fits as a function of the magnitude (UCAC2 instrumental magnitude system) for each telescope. The behavior of the (x, y) Gaussian measurements with magnitude is as expected, with errors increasing toward the faint end. For a given magnitude value, the better performance of the 1.6 m telescope with regard to the 0.6 m telescope is explained by the larger aperture and favorable pixel scale (greater diameter and twice the focal distance). In the case of the 0.6 m telescope, about 65% of the stars have their (x, y) measurement errors below 15 mas (magnitudes up to ~16). This includes most UCAC2 reference stars but not many secondary stars. Toward the faintest limits, say up to the 19th magnitude, the error is greater than 40 mas on that telescope, but then only one observed source was that faint. Besides, statistics are less reliable there, since this faint magnitude range represents only 6% of the total 0.6 m telescope measurements. In the case of the 1.6 m telescope, most stars lay below 15 mas, reaching 20 mas and above only in extreme cases for magnitudes over 20 mag, where no sources were observed. Finally, the small error peaks in x and y around the 17th magnitude on the 0.6 m telescope graphics are the contribution of pixel-undersampled, fainter stars from a small group of short 90 s CCD exposures. Similar peaks around the 11th magnitude are caused in turn by pixel saturation on brighter stars from a small group of deep 300 s CCD exposures. Considering reference stars alone, all other computed error estimates such as (O
C) residuals, standard errors, etc., display no dependencies with magnitude or location on the CCD frames for both telescopes. On average, 30 reference stars on the 0.6 m telescope and 28 stars on the 1.6 m CCD fields were used, with typically less than 10% outliers being excluded in the reductions. Averages for the errors commented in the following are listed on Table 1. While (x, y) Gaussian errors refer to stellar image scales, (O
C) residuals from the x, y transformation of individual (x, y) measurements contain astronomical, atmospherical, and instrumental errors at much larger areas on the CCD frames. The formal standard deviations of the residuals of the x, y transformations were 18 and 15 mas for the 0.6 and 1.6 m telescopes, respectively, on both coordinates. A random increase of residuals toward fainter magnitudes occurs in the same fashion as with Gaussian errors. The similar residuals that were found for the (x, y) transformations reflect the star distributions with magnitude for each telescope and the respective Gaussian (x, y) error distributions shown on Figure 1. This indicates that no further significant degradation occurred because of other atmospherical or instrumental effects over the entire CCD fields. It also ensures the good intrinsic astrometric quality of the 0.6 m telescope frames within the range of its magnitude limits. The standard deviations for the unweighted ( ,  ) reductions were 31 and 30 mas for the 0.6 and 1.6 m telescopes, respectively, for both coordinates. They agree with the expected errors of the reference catalog for the 15th magnitude—the average of the UCAC2 reference stars used in the reductions. The standard errors for the right ascension and declination of the sources were No. 6, 2005 ASTROMETRY OF ICRF SOURCES 2909 Fig. 1.—The (x, y) measurement error estimates from Gaussian fits with respect to magnitude (UCAC2 instrumental magnitude system). For the 1.6 m telescope, each point on average represents 55 stars over 2413 measured objects, and for the 0.6 m telescope each point on average represents 77 stars over a total of 4383 objects. computed directly from the variance-covariance matrix in the ( ,  ) least-square adjustments following Eichhorn & Williams (1963). They were, on average, 9 mas for both coordinates and telescopes. In all, the error estimates for the (x, y) measurements and ( ,  ) reductions were similar for both instruments. Thus, we must consider a possible random error propagation in magnitude from the Gaussian fits and from the (x, y) transformations, affecting the positions of the fainter 0.6 m telescope stars, which serve as a reference frame for the 1.6 m field reductions. Random error propagation from the 0.6 m telescope ( ,  ) solutions themselves should also be expected (see more comments on this issue in x 3). our V-band CCD frames, since they were referred to UCAC2, a catalog from observations on the 579–643 bandpass between V and R. That is, we needed a set of independent V-band star positions similar to UCAC2 in epoch, magnitude range, and position precision. For that, we observed 17 common fields taken with the V filter with VCMC (k ¼ þ46 58 0 03 00 ,  ¼
23 00 0 06 00 , h ¼ 850 m). The observations with VCMC were carried out in the period 1998–2000 in drift-scanning mode, resulting in fields of 130 extension in declination and 30 minutes in right ascension. On average, 10 fields were observed per source, with an overlapping technique being used to reduce positions. All the thousand stars identified in these fields had positions referred to the Tycho-2 catalog, the primary reference frame of UCAC2 itself. Only those portions of the sky corresponding to 100 around the source positions were used. Visual magnitudes were also calculated, giving values ranging from 8 to 15 for the VCMC stars 2.2. UCAC2 and its Magnitude Bandpass System Applied to V-Band Observations We needed to investigate whether unaccounted magnitude/ color effects might have eventually affected the reductions of TABLE 1 Error Estimates: (x, y) Measurements and ( ,  ) Reductions 0.6 m Telescope Errors ( , ) Reductions (mas) 1.6 m Telescope Errors ( , ) Reductions (mas) (x, y) Measurements (mas) (x, y) Measurements (mas) Statistics N   E E Ex Ey N   E E Ex Ey Average ........................ Dispersion .................... 30 30 7 31 8 9 6 9 6 18 6 17 5 28 31 8 29 7 9 7 9 6 15 7 14 6 Note.—Error estimates for the (x, y) measurements and ( ,  ) reductions for the sample of 31 ICRF sources. N stands for the average number of reference stars; Ex and Ey are the error estimates of (x, y) measurements for the entire CCD frames [standard deviations of the residuals from the (x, y) overlapping transformation adjustments; see text for explanation];  and  are the formal standard deviations from the (O
C ) residuals of reference stars from the ( ,  ) reductions; and E and E are the formal standard errors of the source positions from the variance-covariance matrix of ( ,  ) reductions. 2910 ASSAFIN ET AL. Vol. 129 dependent fields (17) are considered. In all, the analysis above indicates that UCAC2 behaves similarly to any V catalog, not introducing any systematic errors in our reductions regarding magnitude or color, on both equatorial coordinates. 3. RESULTS Fig. 2.—Position offsets (nonabsolute), VCMC
UCAC2, against V magnitudes (Valinhos instrumental system) for right ascension. The error bars refer to the formal standard deviations within each bin (average of 84 stars per bin). A similar magnitude equation was found by Dominicci et al. (1999) and confirmed by Camargo et al. (2003). used. A detailed description of the instrument and astrometric procedures is given in Viateau et al. (1999). As a starting point, the VCMC positions of all 17 common fields were directly compared against their corresponding UCAC2-band positions given in UCAC2. For the 594 common stars, the VCMC
UCAC2 position differences were, on average, +77 mas (69 mas) and
16 (74 mas) for right ascension and declination, respectively. (Formal standard deviations are given in parentheses.) The right ascension average difference is significant and is due to a magnitude equation. Plots of position offsets with respect to magnitude gave a magnitude equation for right ascension, with (nonabsolute) right ascension position offsets systematically increasing toward fainter magnitudes. This feature is displayed in Figure 2. It plots the right ascension position offsets (nonabsolute) as VCMC
UCAC2 against V magnitudes (Valinhos instrumental system). Error bars stand for the formal standard deviations within each bin. An analogous magnitude equation was found and described by Dominicci et al. (1999) and confirmed by Camargo et al. (2003) when comparing position offsets between the VCMC and the Bordeaux CCD Meridian Circle. Thus, the magnitude equation can be safely ascribed to the VCMC exclusively. No dependencies on the position offsets were found with regard to color (V
R), even after eliminating the magnitude equation from the VCMC right ascensions. We must then conclude that no magnitude/color effect arises from using UCAC2 field stars. This conclusion is further supported by a direct comparison against the positions of the VLBI sources. In this case, we reduced the 0.6 m CCD fields with the available VCMC positions. Only those CCD frames with at least 10 VCMC reference stars available were reduced. This yielded the positions of six ICRF sources. Average values and errors for the VCMC
radio position offsets were +89  29 mas (71 mas) and
31  29 mas (72 mas) for right ascension and declination, respectively. (Values in parentheses refer to standard deviations.) The VCMC-based optical
radio right ascension offset equals the previously found VCMC
UCAC2 average difference in right ascension, implying no mismatch between the UCAC2 and radio frames. The direct VCMC
UCAC2 average declination difference also matches the VCMC-based optical
radio offset within the combined errors. Besides, the direct average declination difference (
16 mas) significance is found to be low when the standard deviation (74 mas) and the number of in- Table 2 list the results for all 31 ICRF sources observed at the 1.6 and 0.6 m LNA telescopes. Optical
radio position offsets refer to reductions made with the 0.6 and 1.6 m telescope frames. Offsets in parentheses were omitted from the statistics given at the bottom of the table because they refer to reductions with five or less reference stars only, which is much less than the average number of 30 for both telescopes. The average values are listed together with the respective formal standard errors (in parentheses) and standard deviations. Epochs and magnitudes refer to the mean 0.6 and 1.6 m telescope respective values. The number of reference stars used in the reductions is also given. No observations and therefore no positions were obtained with the 1.6 m telescope for sources 1253
055 and 1921
293 at the time of this publication. The bright, optically extended source1228+126 presents a jetlike feature extending up to about 1500 northwest from the center in all 0.6 m telescope CCD images. Tentative reductions with manual removal of pixels visually associated to the feature were attempted. Two results with and without pixel removal are given on Table 2. In spite of our efforts, no reliable results were achieved for this source, since the values are much larger than the typical optical
radio offsets obtained. 3.1. Astrometric Performance of the Telescopes The optical
radio positions offset averages and standard errors for the 0.6 m telescope were +4  8 mas (41 mas) and +1  8 mas (42 mas) for right ascension and declination, respectively. (Values in parentheses refer to standard deviations.) For the 1.6 m telescope, the offsets were
12  9 mas (45 mas) and +8  9 mas (46 mas). Since the dominating source of systematic error (i.e., offsets) comes from the primary reference stars’ zero points on right ascension and declination in each field, it is bound to be the same for both telescopes because the 1.6 m telescope data is derived from the 0.6 m astrometric solution. At any rate, the averages are statistically null, indicating no misaligment between UCAC2 and the ICRF at the 10 mas level. Statistics for the 0.6 m results refer to 24 optical
ICRF position offsets, with 24 other offsets for the 1.6 m telescope comparisons. The results are in agreement with the UCAC2 position errors at the observation epoch. As given by the standard deviations, the 1.6 m telescope positions are consistent with the ICRF within 45 mas and those for the 0.6 m telescope are consistent within 41 mas. The similarity of the standard deviations for such different telescopes supports the strategy of observing secondary stars with the 0.6 m telescope and binding them to the UCAC2 stars in order to have them provide a fainter, numerous, and accurate representation of UCAC2 in the small CCD frames from the 1.6 m telescope. This enables us to bridge the magnitude gap and extend the program to the faintest end of the optical counterparts of the ICRF sources, while being bound to the UCAC2/ Hipparcos reference system. It is also verified that no larger CCD nor mosaics of frames are required, since Table 2 shows equal O
C standard deviations ( , ) for both the faint nonUCAC2 stars at the 1.6 m telescope and the UCAC2 stars at the 0.6 m telescope. On the other hand, since the gain of precision on the centroid determination (Ex, Ey) from the 0.6 to the 1.6 m telescope is modest for the brighter ICRF sources optical counterparts, the determination of their position solely by the 0.6 m No. 6, 2005 ASTROMETRY OF ICRF SOURCES 2911 TABLE 2 Results for 31 ICRF Sources for the 1.6 and 0.6 m LNA Telescopes 1.6 m Telescope 0.6 m Telescope UCAC2/0.6 m Based Opt
Rad (mas) UCAC2 Based Opt
Rad (mas) ICRF Source Epoch Magnitude
cos 
 N
cos 
 N 0047
579 ..................................... 0237+040 ...................................... 0919
260 ..................................... 0920
397 ..................................... 1032
199 ..................................... 1034
293 ..................................... 1129
580...................................... 1144
379...................................... 1228+126 ...................................... 1228+126 ...................................... 1253
055 ..................................... 1320
446 ..................................... 1329
665 ..................................... 1354+195 ...................................... 1355
416 ..................................... 1435
218 ..................................... 1510
089 ..................................... 1511
100...................................... 1514+197 ...................................... 1514
241 ..................................... 1538+149 ...................................... 1656+053 ...................................... 1921
293 ..................................... 1925
610 ..................................... 2037
253 ..................................... 2052
474 ..................................... 2128
123 ..................................... 2204
540 ..................................... 2251+158 ...................................... 2255
282 ..................................... 2326
477 ..................................... 2328+107 ...................................... Average ......................................... Standard deviation ........................ 95.6493 98.9096 98.8233 98.0671 97.8877 99.3370 99.3411 97.7918 99.5014 99.5014 99.5014 98.9151 98.4904 97.4712 97.7493 99.3383 99.3685 97.4397 99.3657 95.8767 97.4438 97.0644 97.6575 97.9986 97.9959 98.2041 96.5904 96.5644 98.3123 96.1945 96.1259 98.3123 17.3 17.9 17.9 18.0 19.0 18.1 18.4 17.8 15.2 15.6 15.6 19.1 18.8 16.4 15.2 18.5 16.9 18.3 18.7 15.4 17.1 16.7 16.4 19.7 19.3 17.5 15.3 18.1 16.0 16.6 16.0 18.2 +25
15 +42
2 +67
11
40
36 ... ... ... +50
60 (
77) +34 (
131)
20
90
10 (+9)
54
25 ...
45
47
51
54 (+90)
40 +82 +50
26
12 (9) 45 +48 +6
26 +30 +0
7 +22 +28 ... ... ... +88 +75 (
141) +26 (
16)
35
48 +17 (+3) +46
34 ... +10
66 +43
111 (
179)
9 +42 +48
29 +8 (9) 46 8 6 24 37 11 22 113 24 ... ... ... 62 93 (5) 52 (5) 30 18 12 (5) 15 44 ... 18 33 21 10 (4) 8 7 7 9
18 ... +78 +49 +26 +1 +58
78 (
206) (
359) +47 +67 ... (
69) +27 ...
30
37
20 +32
15
17 +2 ...
13
16
77 ... +20
18 +05 +28 +4 (8) 41 +14 ...
21 +5 +52 +03 +63 +18 (+95) (+188) +19 +79 ... (
67)
08 ...
33 +2
8
35 +23 +14
41 ...
44 +25
47 ...
25 +21 +57
113 +1 (8) 42 6 ... 43 37 10 09 172 22 (5) (5) 12 57 ... (5) 24 ... 23 21 8 24 13 22 132 ... 22 9 15 ... 12 6 9 16 Notes.—Results for 31 ICRF sources for the 1.6 and 0.6 m LNA telescopes. Optical
ICRF radio position offsets (Opt
Rad) refer to reductions made with the UCAC2 catalog (UCAC2 based, 0.6 m telescope) and a secondary catalog (UCAC2/0.6 m based, 1.6 m telescope). The average values are listed together with the respective formal standard errors (in parentheses) and standard deviations. Epochs (1900 plus table value) and magnitudes refer to the mean 0.6 and 1.6 m values. N stands for the number of reference stars used in the reductions. No observations and therefore no positions were obtained at the 1.6 m telescope for sources 1253
055 and 1921
293 at the time of this publication. Two results for source 1228+126 are given (0.6 m telescope), with (first result) and without (second result) taking into account a jetlike feature in the images. telescope is validated, which considerably speeds up the realization of the 1.6 m telescope program. The samples investigated here encompass well-imaged sources for both telescopes. For the 10% of the sources successfully observed only at the 1.6 m telescope (see Table 2), the optical
radio offsets (included in the statistics) remained within expected values and did not affect the comparison of results between the telescopes. Likewise, the standard deviation on the 1.6 m telescope optical
ICRF positions of sources dimmer than 19 mag is 43 mas. Accordingly, the position offset distributions of both telescopes are point-to-point similar within 97% or better, applying nonparametric Spearman rank-correlation statistics. This means that the UCAC2 reference frame was still preserved in the positions derived from the 0.6 m telescope reductions and passed forward to the 1.6 m telescope CCD frames. The overall behavior of the optical
radio position offsets with respect to right ascension and declination is shown by the vector map of Figure 3, which refers to the 1.6 m telescope results. The 0.6 m telescope offsets give a similar plot. Arrows point to the relative optical position. For better visualization, right ascensions of two sources between 0h and 3h are plotted from 24h on. Although in general the offset sizes vary, it seems from Figure 3 that for small regions in the sky represented by adjacent vectors similar directions prevail. We generated a random distribution of offset vectors and took the angles between vectors of adjacent sources. To check whether the 0.6 and 1.6 m local distributions really did not behave randomly, we compared their respective angle averages with those expected of a random distribution. After that, we applied the test of equal means using Student’s t-statistics. We found for both instruments a 90% or better probability that the vector directions for adjacent sources are indeed not aleatory. Here an interpretation of this result is unclear because of the small sample at hand. A deeper investigation of the issue of optical
radio position offset distributions, based on a broader observed sample in the context of examining dense astrometric catalogs such as the USNO-B1 2912 ASSAFIN ET AL. Fig. 3.—Vector map of optical
radio position offsets with respect to right ascension and declination for the 1.6 m telescope results. The 0.6 m telescope offsets give a similar plot. Arrows point to the relative optical position. For a better visualization, the right ascensions of two sources between 0h and 3h are plotted from 24h on. (Monet et al. 2003) and the Two Micron All Sky Survey (Cutri et al. 2003), is given in da Silva Neto et al. (2005). We also investigated the presence of differential color refraction, since the sources are bluer than the average reference stars in the CCD fields. Indeed, the size and distribution of the optical
radio declination offsets versus declination (observations were made close to the meridian) were not incompatible with differential color refraction with a V filter (Kovalevsky 1997). Linear fits of these offsets against declination were then tried for both telescopes. Nominally, the obtained correction reduced the standard deviation of the optical
ICRF offsets to 37 mas. However, when the errors on the determination of linearfit coefficients are added to the error budget, no net gain is verified with respect to the adherence to the ICRF. The conclusion of the LNA program will bring a much larger number of observations, which shall clarify the issue. Finally, no correlation between the optical
ICRF radio positions with source brightness is found. 3.2. Comparison with Independent Instrument/Catalog Sets of Source Positions To further test the astrometric performance of our telescopes and UCAC2, we compared our results with other independent Vol. 129 instrument/catalog sets of source positions. For that, the list of 327 sources from Zacharias et al. (1999) was an interesting choice. In that case, short-focus astrographs and long-focus reflectors were used on plate/CCD observations. The positions were directly referred to the HCRS from the astrograph fields. To account for the unknown proper motions of the secondary stars linking short- and long-focus fields, solar motion and Galactic rotation were considered. The paper listed positions with and without these kinematic corrections. Another list considered is given by Assafin et al. (2003), which is independent with regard to telescope (0.9 m telescope at Cerro Tololo Inter-American Observatory [CTIO]) and reduction procedures but which shares the same reference catalog, UCAC2. The comparison between these lists of optical positions is shown on Table 3. Only sources without optical
radio offsets marked in parentheses in Table 2 were considered. Table 3 displays differences, namely, this work minus other lists. Zacharias [C] and Zacharias [N] refer to the optical positions of 20 common sources in Zacharias et al. (1999) with [C] and without [N] kinematic corrections, respectively. CTIO refers to the optical positions of eight common sources in Assafin et al. (2003). Averages and formal standard deviations for the position differences are furnished. The adherence to the ICRF of our UCAC2-based LNA positions has already been shown by the averages at the beginning of x 3.1. It is further supported by the statistically null differences between the common positions of ourselves (LNA) and CTIO. However, the difference between the common LNA and Zacharias et al. (1999) positions is statistically significant in declination. Such a difference does not originate from the positions here obtained, since for the 20 common sources our offsets, LNA
ICRF, are (for the 1.6 m telescope observations)
11  10 mas (46 mas) and
02  10 mas (43 mas) for right ascension and declination, respectively. This is interpreted as indicating that the accuracy of the UCAC2 positions supersedes previous catalogs and specific lists, such as the set of radio source positions by Zacharias et al. (1999) in this case. 4. SUMMARY AND CONCLUSIONS We present results of a pilot investigation on the astrometric performance of small- and medium-sized telescopes and the UCAC2 catalog. For this, 31 ICRF sources, mostly south of the equator, were observed during 1997–2000 at the 1.6 and 0.6 m Cassegrain telescopes at LNA, Brazil. The source positions were referred to UCAC2. Observations were made in the V band in a compromise between the 579–643 nm bandpass of UCAC2 and the bluer ICRF sources. An investigation was made to certify the UCAC2-band positions of UCAC2 as a good reference frame for the CCD fields observed in the V band. For this, UCAC2 was compared with an independent set of star positions of similar epoch, magnitude range, and precision in the V band. This set of TABLE 3 Comparison of Results with Other Optical Position Lists Us
Zacharias [C] (mas) Us
Zacharias [ N] (mas) Us
CTIO (mas) Statistics
cos 

cos 

cos 
 Average .................................. Standard deviation ................. +3 (18) 80
29 (10) 45
1 (15) 68
53 (10) 45
3 (18) 53
15 (21) 60 Notes.—Comparison of results with other optical position lists: this work minus another list. Zacharias [C] and Zacharias [N] refer to the optical positions of 20 common sources in Zacharias et al. (1999) with [C] and without [N] corrections, respectively, for solar motion and Galactic rotation. CTIO refers to the optical positions of eight common sources in Assafin et al. (2003; see discussion in x 3.2). No. 6, 2005 ASTROMETRY OF ICRF SOURCES positions was obtained with the Valinhos CCD Meridian Circle, Brazil. A magnitude equation was found in the study but associated entirely to the VCMC positions, a result also reported independently by Dominicci et al. (1999) and confirmed by Camargo et al. (2003). No other magnitude or color effects were found that would otherwise have limited the use of UCAC2 in the reductions. The final average values and errors of the optical
radio position offsets for 24 sources using the 0.6 m telescope were +4  8 mas (41 mas) and +1  8 mas (42 mas) for right ascension and declination, respectively. (Values in parentheses refer to standard deviation.) For the 1.6 m telescope, 24 other sources gave offsets of
12  9 mas (45 mas) and +8  9 mas (46 mas). Although, as expected, the stars’ (x, y) measurement errors increase toward fainter magnitudes, it was verified that the sets of positions are still statistically consistent with the UCAC2 frame being properly passed from the 0.6 to the 1.6 m telescope positions. A linear adjustment was tried on the optical
ICRF declination offsets against declination in favor of a possible differential color refraction for a V filter (Kovalevsky 1997); this nominally reduced the standard deviations of the optical
ICRF offsets in declination. However, when the errors on the coefficients of the linear fit are considered, there is no net gain. The correction was therefore not applied and the effect will be further investigated when a larger sample becomes available. Other external comparisons with independent instrument/catalog sets of high-precision 2913 source positions, namely, the lists of optical positions given by Zacharias et al. (1999) and Assafin et al. (2003), show results consistent with the respective errors. In all, using the UCAC2 as the primary reference frame one can expect that individual sources positions will be on the HCRS within at least 40 mas. An ongoing long-term program investigating the optical astrometry of southern ICRF sources is being carried out at both telescopes at LNA. Because of the limited number of available sources, a detailed investigation of zonal tendencies and their origins and of differential color refraction shall only be attempted after the conclusion of the entire program. Nevertheless, the results so far obtained indicate the great potential of small- to medium-sized Cassegrain telescopes for accurate astrometric work, particularly on the astrometry of the optical counterpart of extragalactic radio sources. M. A. acknowledges the Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (grant E-26/170.686/2004) for their support. D. N. S. N. is grateful to the Brazilian Conselho Nacional de Desenvolvimento Cientı́fico e Tecnológico (grant 303950/ 2003-0). A. H. A. thanks the Centre National del la Recherche Scientifique for contract QAF183803. R. V. M. is thankful to the Brazilian Coordenação de Aperfeiçoamento de Pessoal de Nivel Superior (grant 0449/04-0). REFERENCES Assafin, M., Vieira Martins, R., & Andrei, A. H. 1997a, AJ, 113, 1451 Eichhorn, H., & Williams, C. A. 1963, AJ, 68, 221 Assafin, M., Vieira Martins, R., Andrei, A. H., & Veiga, C. H. 1997b, AJ, 113, Kovalevsky, J. 1997, in Modern Astrometry, ed. I. Appenzeller et al. ( Berlin: 2329 Springer), 40 Assafin, M., Zacharias, N., Rafferty, T. J., Zacharias, M. I., da Silva Neto, D. N., Monet, D., et al. 2003, AJ, 125, 984 Andrei, A. H., & Vieira Martins, R. 2003, AJ, 125, 2728 Rickman, H. ed. 2002, Transactions of the IAU Vol. XXIVB (San Francisco: Benevides-Soares, P., & Teixeira, R. 1992, A&A, 253, 307 ASP), 33 Camargo, J. I. B., Ducourant, C., Teixeira, R., Le Campion, J.-F., Rapaport, M., Stavinschi, M., & Kovalevsky, J. 2001, EAS Newsletter, 22, 8 & Benevides-Soares, P. 2003, A&A, 409, 361 Viateau, B., et al. 1999, A&AS, 134, 173 Cutri, R. M., et al. 2003, 2MASS All-Sky Catalog of Point Sources ( Pasadena: Zacharias, N., Urban, S. E., Zacharias, M. I., Wycoff, G. L., Hall, D. M., Caltech) Monet, D. G., & Rafferty, T. J. 2004, AJ, 127, 3043 da Silva Neto, D. N., Andrei, A. H., Assafin, M., & Vieira Martins, R. 2005, Zacharias, N., Zacharias, M. I., Hall D. M., Johnston, K. J., de Vegt, C., & 429, 739 Winter, L. 1999, AJ, 118, 2511 Dominicci, T., Teixeira, R., Horvarth, J., Medina Tanco, G., & Benevides-Soares, P. 1999, A&AS, 136, 261