3 Observations and data reduction

3.1 Observations

The sample was observed using the TEK 2048 CCD attached to the 0.9 m telescope of CTIO over 3 nights in August 1998. The pixel scale, projected to the sky plane, is 0.396$\hbox{$^{\prime\prime}$ }$. Broad B, V, I and four narrow filters 75 Å wide centered at $\lambda = 6477$ Å, 6649 Å, 6781 Å and 7053 Å were employed. The first narrow filter was used to measure the continuum level adjacent to H$\alpha$ and the remaining three to isolate the H $\rm\alpha + N[II]$ emission, depending on the redshift of the object. Seeing conditions were very good (1.2 $\hbox{$^{\prime\prime}$ }$ to 1.5 $\hbox{$^{\prime\prime}$ }$) and all nights were photometric. For each target object we obtained a set (usually 3 or 5) of short (200s-400s) exposures for each filter. A summary of the log of observations is shown in Table 2.

  

Table 2: Log. of observations and exposure time in each filter$^{\rm a}.$

$^{\rm a}$ For each filter we list the total number of frames multiplied by the exposure time of each frame, in seconds.

3.2 Reduction of the BVI and H$\alpha$ images

The images were corrected for bias and flat-field using standard IRAF routines. Sky subtraction was performed by averaging mean values of the sky, well beyond the galaxy limits, in several boxes on each frame. All images for a particular galaxy were aligned and combined to obtain single B, V, I and H$\alpha$ images. The alignment was performed using the IMALIGN task of IRAF with at least 7 field stars. The typical accuracy was better than 0.05 pixels. The same technique was used to generate the B-V and B-I color images as well as to subtract the continuum emission from the H$\alpha$ images. When combining frames, if the individual images had different seeing, the ones with better seeing were convolved with a Gaussian function in order to match the image with the poorer seeing.

The photometric calibration to the Johnson-Cousins system was made using 4 or 5 standard stars per night selected from Landolt (1992). The usual equations to transform the b, v and i instrumental magnitudes into the V, (B-V) and (V-I) standard magnitudes were used. Estimates of accuracy in the calibrations are $\pm$0.04 mag in V, $\pm$0.06 mag in (B-V) and $\pm$0.06 mag in (V-I).