2 Observations and data reduction

ISOPHOT Serendipity Survey (ISOSS) measurements were obtained with the C200 detector (Lemke et al. 1996), a 2$\times$2 pixel array of stressed Ge:Ga with a pixel size of 89.4 $^{\prime \prime}$. A broadband filter (C_160) with a reference wavelength of 170 $\mu$m and a width of 89 $\mu$m was used. The highest slewing speed of the satellite was 8$^{\prime}$/s. During each 1/8 s integration time 4 detector readouts were taken, i.e. the maximum read out distance on the sky was 15 $^{\prime \prime}$ yielding one brightness value per arcminute (see Figs. 1 and 2). During the ISO lifetime, about 550 hours of ISOSS measurements have been gathered, resulting in a sky coverage of approximately 15%.

Figure 1: IRAS 100 $\mu$m map with the slew paths of the 4 C200 pixels overplotted. The symbols used for the individual detector pixels are the same as in Fig. 2. Here, their colour coding indicates qualitatively the measured intensities. The intensity peaks in Pixel 1 and 4 have no correspondence in the IRAS map. The star symbol marks the N-body ISO-centric position of Ceres at the time of the slew. Measurements within 5$\arcmin$ from the detector centre are marked here and in Fig. 2 as black crosses. The detector and pixel apertures as well as the scan direction are indicated. The slew TDT number is 09380600 (see also Table 2).

Figure 2: The calibrated pixel intensities as a function of slew length. Intensities of Pixel 2-4 are shifted downwards in steps of 20  . The slew section shown corresponds to Fig. 1. Black crosses mark measurements with detector positions closer than 5$\arcmin$ to the calculated asteroid position. Pixel 1 and 4 cross Ceres almost centrally.

  
2.1 Data analysis

A standard data processing was applied using the ISOPHOT Interactive Analysis PIA (Gabriel et al. 1997) Version 7.2 software package. The detailed processing steps are given in Stickel et al. (2000). Special care had to be taken to correct gyro drifts between subsequent guide star acquisitions of the star tracker. For point sources, the deglitched and background-subtracted signals of the 4 pixels were phase-shifted according to the position angle of the detector and co-added. Source candidates were searched for in this co-added stream by setting a cut of 3 $\sigma$ of the local noise. Then, the source position perpendicular to the slew was determined from a comparison between signal ratios with a Gaussian source model. The flux was afterwards derived from 2-D Gaussian fitting with fixed offset position.

In the case of long slews, the surface brightnesses were derived from a measurement of the on-board Fine Calibration Source (FCS) preceding the slew observation. For short slews the default C200 calibration was used. To tie point-source fluxes derived from ISOSS to an absolute photometric level, dedicated photometric calibration measurements of 12 sources, repeatedly crossed with varying impact parameters were compared with raster maps on the same sources (Stickel et al. 1998a). The comparison between slew and mapping fluxes showed that for brighter sources the slewing observations miss some signals, probably due to transient effects in the detector output (Acosta-Pulido et al. 2000) in combination with detector non-linearities. For sources brighter than 30 Jy, Stickel et al. (2000) found signal losses of 50%, although the true losses were not well established due to a lack of reliable sources. For fainter sources (<10 Jy) the flux loss in the slews is only 10-20%.

2.2 Source extraction methods

The SSOs were encountered at different slew speeds, which can be characterized by "fast'': above 3$^{\prime}$/s; "moderate'': 1.5$^{\prime}$/s < speed < 3$^{\prime}$/s; "slow'': below 1.5$^{\prime}$/s; "stop'': at slewends, similar to a staring observation. But aspects like the background level, the detector history and impact parameters also play a crucial role in source extraction and flux calibration methods.

   
2.2.1 Method 1

Method 1 is based on an automatic point source extractor (Stickel et al. 2000) for all slewing speeds above 1.5$^{\prime}$/s and non-saturated crossings. All source candidates were cross correlated with the list of SSO candidates (see Fig. 3) and the associations found were carefully examined. Flux loss corrections (see Sect. 2.1) have to be applied.

   
2.2.2 Method 2

This is a method used for all slewing speeds, but taking only the pixel with the highest signal and converting it to flux density as if the source was centred. This method leads to upper and lower flux limits only. The lower limits, designated by ">'' or "$\gg$'', are connected to clear detections. The quality of the lower estimate depends on the impact parameter. Useful upper limits have only been given for direct hits where no detection signal was seen. The upper limit then corresponds to the 3$\sigma$-value of the background noise. Flux loss corrections (Sect. 2.1) have to be applied as for Method 1. Note: it is assumed that 64% of the flux density of a pixel centred point source falls onto this pixel.

   
2.2.3 Method 3

This method has been used at slewends, if the source was inside the detector aperture:

a)

Using the signals of all 4 pixels at the very end of the slew and converting them to flux densities assuming the source is centred on the detector. Note: only 53% of the flux of an array centred point source is detected, of which 21% fall on pixel 1, 24% on pixel 2, 32% on pixel 3 and 23% on pixel 4 (Laureijs 1999).

b)

Like case 3a, but source centred on one pixel. Note: in total 74.3% of the source flux are seen by the 4 pixels: 64% in the source pixel, 2$\times$4.2% in the two adjacent pixels and 1.9% in the diagonal pixel.

The statistical errors are computed from the weighted results of the 4 pixels.

Not all of the ISO scientific targets are to be found in the end-of-slew data: firstly, the 4 ISO instruments view separate areas of the sky. Slew end position (ISOPHOT) and target position (other instrument) can therefore differ by up to 20$^{\prime}$. Secondly, many observing modes, especially for ISOPHOT, started off-target for mapping purposes or to avoid strong detector transients.