The data were ``cleaned'' as described in Nandra & Pounds (1994) to remove periods of poor quality data. The process consists of the following stages. First, the SUD and PI_MONI (V1 electrodes or LAC anode wires not illuminated via the collimator) rates were compared, and data from times when they were greater than 3 standard deviations from the best-fit linear relationship were removed from further analysis. Second, the LAC count rates in adjacent spectral channels were compared, and points lying greater than 5 standard deviations from the best-fit linear relationship were also removed (the relatively poor spectral resolution of the LAC ensures that this does not remove valid data). Finally, for background observations, the LAC count rate is expected to vary with the SUD rate (and hence PI_MONI rate). Therefore, points lying greater than 4 standard deviations from the best-fit linear relationship were removed from the background data. Plots are available in both the GINGALAC and GINGABGD databases showing the SUD rate versus the PI_MONI rate. A plot is available in the GINGABGD database showing SUD rate versus LAC top-layer count rate in the energy range 2-10 keV. These plots are produced after data rejection so that the user can verify that the cleaning algorithm has successfully removed poor quality data. Examples of both plots are given in figure 3.
A number of additional plots were produced during the pipeline data selection and cleaning process. First, a plot of the SUD rate versus the time allows the user to evaluate the background level during an observation (the total background scales with the SUD rate). The user may wish to extract data from periods of low background (i.e. low SUD rate). Second, a plot of YELEV versus the LAC top-layer count rate from the pulse height channel 4 is available. There will be no correlation between these two parameters if the data are unaffected by solar X-rays scattered into the Ginga field of view by the Earth's atmosphere (see section 2.1). Third, the GINGABGD database contains a plot of DELTXZ versus the LAC top-layer count rate in the energy range 2-10 keV; this can be used to determine whether a source lies within the Ginga field of view. One might expect some correlation between DELTXZ and the LAC count rate if such a source exists. Whilst these data have not been removed (to allow for maximum use of the database), the user is able to exclude such data.