Next: 10 Getting Started Up: 9 Caveats Previous: 9.3 Status of Scientific

9.4 OLP Version 7: Limitations and Caveats

This section addresses the limitations and caveats of the PHT calibration file Version 4.0 and the OLP Version 7 which have been identified during the scientific pipeline validation process.

9.4.1 Error Calculation and Propagation

Whenever possible uncertainties had been included in the Cal-G files. It turned out, however, that the simple propagation of relative uncertainties as originally specified did not yield reasonable results, as the uncertainties are of different types: statistical and systematic uncertainties and photometric bias. The relative contribution of these different types of uncertainty depends on the details of the measurement mode.

OLP Version 7 is restricted to the pure statistical uncertainty from signal evaluation. This can be seen as a minimal error. For an estimate of the total uncertainty associated with a specific measurement the observer is referred to [15] and [11].

Caveat: The PHT OLP software does only provide an uncertainty from the statistical signal evaluation. It does not give any estimates of the systematic uncertainties of all performed calibration steps. Therefore, OLP 7.0 uncertainties are minimal errors. It is essential to refer to special documentation which provide an overview of the uncertainties introduced step by step. These depend on the exact observing conditions and evaluation steps which may even not be completed inside OLP (e.g. background subtraction in maps or from independent staring measurements).

9.4.2 Chopped Photometry

On the basis of the investigated test cases the chopped photometry of AOTs PHT03 and PHT22 cannot be declared scientifically valid.

The most reliable results have been found for chopped P2 photometry for which in many cases an absolute calibration accuracy of around 20% could be achieved.

This rejection is based on the following findings:

In particular for P3 and C100 a general over-correction of the chopped Photometric fluxes w.r.t. reference photometry from the literature or ISOPHOT staring mode photometry is observed.
In multi-filter modes the scatter of the photometric data points, including non-detection of the source, or discontinuitites in the resulting SEDs larger than 30% are found which do not allow a sound scientific analysis.

Reasons for this behaviour of the chopped modes which need follow-up investigations are supposed to be:

Limitations in the usage of an average default responsivity (the orbit dependence alone appears to be too weak to account for the variation of the responsivity values).
A not yet sufficient signal evaluation for chopped measurements. In particular for C100 the restriction to use the central pixel only in the case of chopped point source photometry seems to be mandatory taking the test results into account.
Not yet final signal correction factors. Here, more calibration data have been acquired.

9.4.3 Extended Source Photometry

Wavelength dependent solid angles for the PHT-C arrays derived from refined beam calculations have been successfully tested and make the surface brightness values for all investigated test cases comparable to COBE annual map results.
The PHT22 AOT extended source photometry can therefore be declared scientifically validated with an absolute accuracy of 30% and a relative filter-to-filter accuracy of 20%.
The PHT03 AOT extended source photometry needs still further refinement of the solid angles and a detailed check-out of the dependence on the aperture size.

Note that the accuracy of COBE annual maps for 140 and 240 ${\mu}$ m is typically 20%. For weekly maps it can be even worse. For wavelengths less equal than 100 ${\mu}$ m the quoted accuracy is in general $\le$ 1%.

9.4.4 Spectroscopic Mode

Under the following conditions significant deviations in flux from earlier (validated) OLP Version 6 results can occur:

The newly introduced orbit dependent dark signal subtraction shows noticeable differences in the resulting fluxes for continuum levels below 6 10^-14 W m² ${\mu}$ m^-1 for SS and 4 10^-14 W m² ${\mu}$ m^-1 for SL, respectively (for staring mode observations).
For faint continuum sources changes in the continuum level of about 50% can occur w.r.t. OLP V6 results. For the investigated test case (NGC 6543) no final conclusion could be drawn by comparison with reference spectroscopy (IRAS LRS and CAM-CVF) and further refined cross calibration is necessary. The line fluxes are in general not affected, if the peak intensity is above the levels where dark current subtraction becomes crucial (see above).

9.4.5 Validity Ranges of FCS Calibration Files

Caveat: A poorer calibration accuracy can be expected for observations for which the headers of the PHT data products indicate that the FCS power settings were outside the validity range given in the header of the P*FCSPOW*.FITS CAL-G files (see section 13.10). The deviation is larger for extrapolations further away from the validity range.

Caveat: The FCS calibration for P3, C100 and C200 before revolution 94 is less well sampled with standard calibration observations, than after revolution 94. Therefore, the quality of the FCS calibration of observations before revolution 94 can be poorer as compared to observations taken later in the mission. This also applies to the FCS flat-field calibration for the C100 and C200 detectors.

9.4.6 Inhomogeneous FCS Illumination of PHT-P Apertures

Caveat: In case the FCS calibration of an observation is performed in a non-standard aperture of this filter strong deviations (factor $\simeq$ 2) of the correct responsivity and thus the derived fluxes can occur. The deviations are strongest for the usage of the largest apertures in combination with P1 filters.

9.4.7 Non-geometric Aperture Beam Size Scaling

Caveat: If observations are performed in apertures, which are different in diameter from the ones used in standard calibration, deviations from the correct flux calibration can occur. This implies especially for extended source or background observations using the largest apertures with P1 and P2.

9.4.8 Non-linearity of Detector Responsivity with Flux

There are meanwhile strong indications from the huge data base of standard star measurements that ISOPHOT detectors do not behave linear in their response with incident flux. By design of the AOTs the PHT uplink logic matches the FCS flux to the expected source flux and consequently the non-linearity should be minimal in many cases. However, in multi-filter AOTs it is possible, that the flux in one filter deviates quite strongly from the flux in the filter in which the FCS calibration is performed (especially for P1). In these cases the filter that deviates most from the inband power used for the FCS calibration (usually the first measurements of an AOT sequence) can show calibration problems due to this effect. Similar problems can occur for maps and for multi-aperture observations, which show a wide dynamic range in flux.

Caveat: Measurements with inband powers far away ( $\geq$ factor 10) from the ones used for the related FCS measurement(s) have poorer calibration accuracy.

9.4.9 Detector Signal Transients

Caveat: The transient correction only works satisfactory for measurements longer than the detector transient time constant at a given illumination. The transients are very often the limiting factor for the achieved accuracy. Measurements flagged by the OLP software as ``plateau data affected by residual drift'' can show strong deviations (residual drift is greater than 5%) from the correct result.

9.4.10 Long Term Detector Drifts

For single pointing measurements with one FCS measurement per detector no attempt is made by PHT OLP 7.0 software to correct for long term detector responsivity drifts, i.e. stability of the detector responsivity during the execution time of an AOT is assumed. For raster pointing observations and sparse maps OLP does a linear interpolation with time of the responsivity values derived from the two bracketing FCS measurements. Detailed investigation of PHT-C maps has shown that the detector long term drift is not linear, but very often shows an exponential behaviour due to switch-on effects. Therefore, models of the detector behaviour or baseline fits would be required for optimum results.

Caveat: PHT OLP does not correct (in case of single pointings) or correct only with a linear interpolation (in case of maps) for long term detector drifts within an AOT.

9.4.11 Missing Measurements and Chopper Correction

The current version of OLP does not perform a consistency check whether the number of measurements which were actually performed does match the number of measurements expected from the EOHA (see Chapter ??, except for an initial check during DERIVE_SPD. In case there are measurements missing in an AOT it is possible that there is a mismacth between the keywords written into the SPD header concerning the chopper correction parameters, w.r.t. the measurement number and the filter number that was actually used in a measurement.

Caveat: In case there are measurements missing from an AOT there may be a mismatch between the chopper correction keywords and the related filter number.

9.4.12 Limitations in Use of PHT Color Correction Factors

Caveat: For the establishment of the color correction factors provided by CAL-G Version 4.0 not yet final PHT filter bandpass data including a wide wavelength range outside the nominal bandpass have been used. For some filters and in the case of extremely rising or falling SEDs this can mean an underestimate by out-of-band contributions. Color correction factors above 2 should be considered as less accurate and factors above 4 only as an indication of the order of the correction, as they clearly depend on the integration and interpolation method used. This applies in particular to SEDs of cold sources in combination with the P1 filters which fall onto the steep Wien branch of these SEDs. In these cases it is recommended to contact the ISO Data Centres for the most recent bandpass transmission and establish the color correction factor yourself.

9.4.13 Empirical PPSF Correction Factors for Array Centered C200 Photometry

During OLP V6 Scientific Validation it had been realised that the theoretically calculated point spread function correction factors (PPSF) for photometry with the source centered on the C200 array were not the correct ones. Hence, empirical correction factors were established by comparing the relative intensities in the pixels for targets which had been measured both in array centered and pixel raster mode. These empirical correction factors compare with the theoretical ones as shown in Table 9.3:

**Table 9.3:** PPSF Factors for Array Centered C200 Photometry
C200	theoretical	empirical
Filter	PPSF	PPSF
C_120	0.900	0.574
C_135	0.864	0.668
C_160	0.810	0.708
C_180	0.782	0.972
C_200	0.736	1.054

A plausible explanation for the reverse behaviour (increasing PPSF with wavelength instead of decreasing PPSF) is diffraction at the edges of the field lenses which are mounted with an inter-pixel gap of about 100 ${\mu}$ m, so that the effect is expected to be strongest for the short wavelength C200 filters.

Detailed investigation of several test cases during OLP V7 Scientific Validation clearly points out that the photometry in the C_180 and C_200 filters always gives too low fluxes. Comparison with reference photometry suggests that the PPSF factors for the two filters must be some 20 to 30% smaller, hence close to the theoretical value. This issue will be followed up in a detailed calibration program to establish refined PPSF correction factors for array centered C200 photometry.

Therefore, the following caveat has to be put onto the CAL-G V4.0 empirical PPSF correction factors for array centered C200 photometry:

Caveat: The correction factors for the filters C_180 and C_200 appear to be too large by 20 to 30%. Photometric fluxes derived for both filters in array centered mode should be corrected towards larger values by this percentage (as the uncorrected flux value is divided by the PPSF factor).

Next: 10 Getting Started Up: 9 Caveats Previous: 9.3 Status of Scientific

ISOPHOT Data Users Manual, Version 4.1, SAI/95-220/Dc