SpringerLink
Forum Springer Astron. Astrophys.
Forum Whats New Search Orders


Astron. Astrophys. 363, 517-525 (2000)

Previous Section Next Section Title Page Table of Contents

6. Uncertainties in the photometric redshifts
and spectral types

A potential source of uncertainty in estimating the photometric redshifts and spectral types of galaxies is the possibility that the models might be degenerate (i.e. different synthetic SEDs, corresponding to different redshifts and galaxy types producing the same result). Furthermore, the photometric errors in the observed SEDs are likely to affect the final estimate of both the photometric redshifts and the spectral types of their respective galaxy. Also, due to the relative similarity of the model SEDs for starburst and irregular galaxies (Fig. 2), the accuracy with which the spectral types for these systems are predicted, needs to be established.

To investigate the above problems, a Monte Carlo simulation is performed. A simulated catalogue is generated to resemble the observed HDF survey, with UBVI magnitudes, known redshifts ([FORMULA]) and spectral types. The galaxies are randomly selected to have SEDs similar to the synthetic SEDs for the four types of galaxies (elliptical, spiral, starburst and irregular) considered in Sect. 5.1, shifted in redshift space. Random Gaussian noise, resembling photometric errors, are then added to the simulated SEDs. The simulated catalogue has a magnitude limit of I=28 mag and an apparent magnitude distribution similar to the observed HDF survey.

The photometric redshift code is used to predict the redshifts ([FORMULA]) and spectral types of individual galaxies in the simulated catalogue and to compare them with their input values. Two set of simulations are carried out, assuming different observational errors (i.e. [FORMULA] in the Gaussian noise distribution). The difference between the input (simulated) and the output (predicted) redshifts is presented in the histogram in Fig. 4, with the spectral types compared in Table 4.

[FIGURE] Fig. 4. Histogram of [FORMULA] for the simulated HDF catalogue. The continous line is the result for the photometric errors ([FORMULA], [FORMULA], [FORMULA], [FORMULA]) corresponding to (0.15,0.10,0.05,0.05) and the dotted line corresponds to (0.25,0.15,0.10,0.10). Both the histograms peak around zero, implying that the redshifts in the simulated catalogue can be produced to within [FORMULA].


[TABLE]

Table 4. Results from the simulation of spectral types. Fraction of the galaxies of different types, correctly classified between the input and output simulated catalogues are listed.
Notes:
1) Ratio of the number of galaxies of a given spectral type in the output catalogue which are correctly classified, to the total number of galaxies of the same spectral type in the input catalogue.


The [FORMULA] distribution (Fig. 4), including all the four types of galaxies, shows a distinct peak at zero, indicating that the redshifts for the simulated galaxies are well re-produced within [FORMULA]. The objects, located at the tails of the distribution in Fig. 4, are all galaxies classified as spirals, irregulars or starbursts for which, the degeneracy of their SEDs appears to be more serious (see below). When increasing the photometric errors in the simulated SEDs, the agreement between the simulated (input) and predicted (output) redshifts decreases but the distribution still peaks around zero (dotted line in Fig. 4).

The spectral types of the simulated galaxies are compared with the predicted types in Table 4. This gives the ratio of galaxies of a given spectral type, which are correctly classified in the simulation (i.e. galaxies for which their spectral type in the input catalogue were successfully re-produced), to the total number of galaxies of the same type in the input catalogue. In both simulations, we re-produce the spectral types for all the ellipticals in the input catalogue with no mis-identifications. However, due to the relative similarity of the UV-to-optical part of the synthetic SEDs for the spirals, irregulars and starburst galaxies (Fig. 2), we can recover respectively [FORMULA], [FORMULA] and [FORMULA] of the spectral types of galaxies in the input catalogue.

The conclusion from the simulation here is that the photometric redshifts are, on average, well produced (within the expected accuracy) and are not affected by the degeneracy in the template SEDs. Furthermore, the spectroscopic type classification for ellipticals is reliable with 100% re-produced (i.e. no mis-identifications) from the input catalogue while, for the spirals, irregulars and starbursts, there is a slight degeneracy in predicting their spectral types.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 2000

Online publication: December 11, 2000
helpdesk.link@springer.de