5 Second step: Adding GSC2.2 and UCAC1

The GSC2.2 has also been divided in strips, to be cross-matched with 2MASS. These strips are no more related to the DENIS strips, but to strips with much smaller extension ( $6^{\prime}$) in order to optimize the time needed to cross-match the catalogues. We chose a searching box of $10^{\prime\prime}$. It was not possible to compare the GSC2.2 and 2MASS magnitudes in the same way as for the DCMC vs. 2MASS cross-matching (as in Sect. 3.1.3.) because there is no common magnitude between the two surveys. As a consequence, we kept only the associations to the nearest neighbour and then we cut the resulting distribution at a distance of $4^{\prime\prime}$ (Fig. 6). The mean positional offset between matches is $0.45^{\prime\prime}$ and the modal offset is $0.25^{\prime\prime}$.

Figure 6: Results of the cross-matching between the GSC2.2 and the 2MASS catalogues. Number of objects as a function of the distance to the nearest neighbour. The bin size is 0.1 $^{\prime \prime }$.

At this stage, both the DCMC and the GSC2.2 are cross-matched with the 2MASS catalogue. Thus the link between the DCMC and the GSC2.2 can be done using the 2MASS common point sources. Common entries in DCMC/2MASS and GSC2.2/2MASS have been merged. We have now six resulting files:

• sources present in DCMC, 2MASS and GSC2.2;
• sources present in DCMC and 2MASS only;
• sources present in GSC2.2 and 2MASS only;
• sources in 2MASS only;
• sources in the DCMC only;
• sources in the GSC2.2 only.

We run another cross-matching process on the last two files to find sources which are present in both DCMC and GSC2.2. For small positional differences the associations are likely true whereas at larger distances they are generally random associations, see the distribution of the sources as a function of the distance from the nearest neighbour (Fig. 7). The mean positional offset between matches is $0.61^{\prime\prime}$ and the modal offset is  $0.35^{\prime\prime}$.

Figure 7: Results of the cross-matching between the DCMC and the GSC2.2 sources not present in 2MASS. Number of objects as a function of the distance to the nearest neighbour. The bin size is 0.1 $^{\prime \prime }$.

However, because about 1% of the DCMC sources suffer from astrometric problems, we cannot exclude that there are some true associations for sources at larger distances. Unfortunately there is no magnitude in common between the DCMC and the GSC2.2 catalogue and we cannot adopt the same strategy as in Sect. 3.1.3. We decided to keep only the associations with distances smaller than $2^{\prime\prime}$, to avoid polluting the MC2 with too many false associations. On the other hand, we are losing a few associations, located in the poorly calibrated regions of the DCMC (see the strip like features on the left plot of Fig. 8). A comparison of the slopes of the histograms in Figs. 6 and 7 does not seem to indicate any excess of false matches for distances between 2 $^{\prime \prime }$ and 4 $^{\prime \prime }$. Sources belonging to both DCMC and GSC2.2, but not 2MASS, are shown on the middle plot of Fig. 8. They correspond mainly to sources falling in the yet empty gaps of the 2MASS data. The 2MASS scanning strategy covered the sky with tiles 6 degrees long in Dec and $8.5^{\prime}$ wide in RA. These patterns remain visible on the right plot of Fig. 8, showing the spatial distribution of the MC2 sources belonging to the 2MASS catalogue only, thus denoting different sensitivity limits.

Figure 8: Spatial distribution of some cross-matched sources. Left panel: sources of the MC2 belonging to the DCMC catalogue only. The DENIS strip structure remains visible, because some of them suffer from field distortion all along the borders. Middle panel: sources of the MC2 present only in the DCMC and the GSC2.2 catalogues. Most of them fill the gaps of the 2MASS catalogue. Right panel: sources of the MC2 belonging to the 2MASS catalogue only.

Table 2 summarizes the results obtained so far during the process to build up the MC2, which contains more than 6 million sources for the LMC. The optical/IR database contains 1 968 360 sources: IR from 2MASS (629 212), DCMC (177 414), or both (1 161 734). Among the remaining sources, 4 million of them are only detected in the GSC2.2. It is of great astronomical interest to get as many wavelengths as possible for each star, but this should not lead to disregard sources detected only with one survey and not with the other ones. Keeping non-associations in the MC2 helps keeping track of the internal discrepancies and different sensitivity limits of each catalogue.

  

 

Table 2: Distribution of the MC2 sources, prior the inclusion of UCAC1.

2MASS	DCMC	GSC2.2	Number of sources
			1 161 734
			54 584
			629 212
			151 215
			65 709
			4 064 181
			177 414
Total			6 304 049

The procedure to add the UCAC1 is quite different. We cross-matched the UCAC1, without splitting it in strips, with the MC2 at its present stage. This is possible because the UCAC1 is a small catalogue and our program is fast enough to process it in one run. Another advantage is that the UCAC1 is automatically cross-matched with the DCMC-only sources and the GSC2.2-only sources.

Figure 9 shows the same as Figs. 6 and 7 for UCAC1 and MC2 sources. We decided to keep all these associations, even the ones for sources at distances larger than $1^{\prime \prime }$, because these sources also display a larger proper motion compared to the average source in the catalogue (Fig. 10). This might as well be the cause of the large distance derived during the association process. The mean positional offset between matches is $0.17^{\prime\prime}$ and the modal offset is $0.15^{\prime\prime}$. About 42 UCAC1 sources do not have a MC2 counterpart, which means that 99.9% of the UCAC1 catalogue is linked to the MC2 and 4.2% of the MC2 has a UCAC1 counterpart.