MPA: Meridian Photometer Array
The Meridian Scanners at each MPA site complete a full scan every 30 seconds. The data from each scan are binned into six sets 17 bins. Two such scans are averaged to construct one data block. Each message (record) transmitted contains a header, a status block and two data blocks. The first data block always refers to the even minute.
Each Scanner sends its messages according to a preprogrammed time schedule directly to the transmitter. These are received in Ottawa simultaneously at the VAX/780 and the VAX/730, and filed in the Alpha files.
The data collected on site are partially corrected on site: the hardware counts photons from the single photomultiplier throughout each scan and every hour on the half hour, it measures the dark count. Each signal count is corrected for dark count and for nonlinearities at high count rates.
Conversion to Rayleighs is done on the fly by the subroutine RAY_CON. This subroutine operates on the data portion of the MPA BETA record. It converts the signal channels to Rayleighs (6300, 4861, 4709 and 5577) and the background channels (4800 and 6250) to Rayleigh/Angstrom, one BETA record at a time. It returns the converted data in the array RAYLEIGHS, which contains six scans for up to four stations. If the station is On Duty, geophysical units are returned. If a station is Off Duty, RAYLEIGHS is set =-1., if a station is not present, RAYLEIGHS is set =0.
The constants used are in their basic form: halfwidths, efficiency factors and calibration factors in R/A. This relates the code directly to the instrument calibration data. The H-beta efficiencies are interpolated from the values at 0, 30, 60, 90 deg.
Dark count is subtracted in the instrument, but this has turned out to be a mistake, because from time to time, the shutter does not close properly and the value measured is erroneous. Hence a test for the magnitude of the dark count is made and it is added back in if it is deemed to be too high. If the current DC is greater than three times the average DC (which is computed only from those measurements taken when it was really dark) then it is deemed to be too high. This test is a bit crude but will catch those big values when the shutter was partially open in twilight. At other times, the DC is so low that adding 0.3 count (DC=80), say, does not have much of an effect on the final value. whereas 10 counts (DC=2400) does.
Next, data is converted to Rayleighs or R/A with the appropriate calibration factor. The calibration factors are computed first time in only.
Next, the background correction is applied: the appropriate background channel, multiplied by a "correction" factor. The reason for this is that we only have one channel for 4861, 4709 and 5577 and this value must be appropriately scaled.
Next, the filter efficiency factor is applied. These are only available for 4861 and 4709 but may be needed for other channels if the filter passbands shift. For this reason, all channels have efficiency factors applied. NOTE: Subroutine RALEIGH applied the efficiency factor for 4709 to the data before BG correction.
Next, the van Rhijn factors are applied (4709, 4861 and 5577 only). In this version, the coefficients used are a combination of atmospheric extinction and van Rhijn, first used in the IMS studies.
Finally, there is an option to correct the H beta channel for scattered moonlight. The H-beta signal channel filters are centered closely on the centre of the Fraunhofer absorption. This means that during times when the moon is up, the presence of scattered moonlight minus the centre of the line in the signal channel, can, when the signal is corrected for background, cancel out the signal and produce zero or "negative" intensities. It is an empirical procedure, based on the computed lunar position and the relative illumination of a horizontal surface (relative to full moon). This correction process is imperfect at best, because it is highly dependent on the turbidity of the atmosphere, the presence of clouds and so on. It also depends on the look direction, because H beta is doppler shifted and although this is corrected for, applying a moon brightness correction to this makes the whole thing even more imperfect.
At times it may be desirable to omit some of these correction steps. The conversion mode provides for a number of likely combinations.
Two steps in the current correction process need fine tuning. Since none of the background channels is exactly right for the signal channels, an empirical correction factor can be added. This factor must be estimated using "quiet oval" data and this is yet to be done. Verification of the van Rhijn/extinction factors may be possible with some quiet night data also (for 5577 and 6300).
The constants are structured for a 6 instrument array, the original system configuration. At present, only sites 1,2,3 and 5 exist.
The subroutine processes all the data blocks of the BETA record. Input to the subroutine is the BETA record and the conversion mode. Returned are Rayleighs for channels 2, 4, 5 and 6 and background brightness in Rayleigh/Angstrom for channel 1 and 3 for all MPA for which data are present. The conversion mode is defined as follows:
- the full correction mode, corresponding to the algorithm of the routine RAYLEIGH: BG correction (all channels), van Rhyn (4709, 4861, 5577), filter efficiency (4709, 4861)
- NO correction other than conversion
- Conversion and background subtraction (NO van Rhyn correction)
- As for mode=0 but BG adjusted for lunar brightness
- Conversion to R/A for all channels. No BG subtraction!
- ...Returns RAYLEIGHS with all values = -9.0