The approach to the processing of parallel and serendipity data is essentially the same as for the prime data and, whenever possible, the same algorithms and calibration files are applied (Swinyard et al. 1998, [40]; Burgdorf et al. 1998, [4]). For the SPD level product, the first stage of obtaining the slope of the ramp is not required. However the same engineering conversions are then applied to obtain a photocurrent. Only a small adjustment for the difference in slopes, obtained from 1s (parallel/serendipity) and 1/2s (prime mode) ramps, is also applied at this stage.

**Table 4.1:** *The LWS grating rest position wavelengths.*
Detector	Wavelength	Width of resolution	Dark Current
	[ $\mu$ m]	element [ $\mu$ m]	[ $\times$ 10 $^{-16}$ A]

SW1	46.2	0.29	4.35
SW2	56.2	0.29	1.89
SW3	66.1	0.29	1.91
SW4	75.7	0.29	0.86
SW5	84.8	0.29	1.21
LW1	102.4	0.6	2.22
LW2	122.2	0.6	0.03
LW3	141.8	0.6	0.29
LW4	160.6	0.6	1.74
LW5	178.0	0.6	1.28

4.5.1 Dark current subtraction and drift removal

Once the photocurrents have been obtained, the next stage is to remove the dark current. For prime mode grating data this is done by measuring dark current values prior to the illuminator flashes which take place at the start and end of each observation and subtracting the average. All dark current measurements were checked for trends and it was found that the dark current has remained at a stable value throughout each revolution of the ISO mission. The rare exception being that transient effects after observing bright sources sometimes led to higher than normal values. A similar monitoring exercise was done with parallel and serendipity data. The dark values were defined as the mininum photocurrents consistently obtained and these were implemented as one dark current value per detector. These values (see Table 4.1) were found to be lower that those found in prime mode (see Section 5.4) and were applied as a fixed dark removal in the parallel/serendipity pipeline.

From inspection of prime mode illuminator flashes, the responsivity of the LWS detectors is known to vary during a revolution, the net effect being a linear drift upwards, restored by a bias boost peformed during the handover period in the middle of a revolution, and another linear drift in the second part of the revolution (Lim et al. 1998, [26]). The calibration of the detector responsivity relies on a simple ratio between the response to the illuminators found at the time of a particular observation and that used as a reference. However, as serendipity and parallel observations did not have dedicated illuminator flashes a different approach had to be found. For each half revolution all illuminator flashes were linearly fitted to obtain responsivity drift coefficients for that revolution. The parallel and serendipity data were then calibrated by using the interpolated responses of the detectors. For revolutions where there are no prime mode observations, a standard responsivity drift defined by averaging all revolutions, is applied.

In prime mode the grating or FP is moving constantly hence the detector receives a constantly changing signal. In parallel mode the grating remains at a fixed position and therefore it was possible to apply a transient correction to the data. The wavelength determination was done by lookup table as all the data were taken at the grating rest position and this remained stable throughout the ISO mission. Each data point has a bandwidth of one grating resolution element (see Table 4.1). The units of both parallel and serendipity products are in MJy sr $^{-1}$ as a correction is made for the beam profiles (see Section 5.9 or Lloyd 2000, [27]).

4.5 Processing of the Parallel and Serendipity Modes

4.5.1 Dark current subtraction and drift removal