OMI Operations - Ozonehole monitoring

OMI

OMI was launched at July 15 2004 after which the LEO phase started. The LEO phase ended September 2004 and nominal operations started. During the day side of the orbit radiance measurements are taken with tree different settings:

• tropics (tropical lattitudes -> very light: low gain factors)
• midlat (north and south mid lattitudes -> medium light: medium gain factors)
• arctic (north and south arctic lattitudes -> low light: high gain factors)

OMI and the ozonehole

By the time OMI started taking data according to the nominal baseline the ozonehole season had started at the south pole and OMI UV1 CCD was experiencing a lot of saturated ADC pixels. During the ozonehole season there is an ozone deficit. Less light gets absorbed by ozone and more light is detected by OMI. Furthermore, when OMI flies towards the tropics the amount of light increases. Therefore, the pixel saturations starts at the end of the arctic measurement period. The ADC saturation ends (temporarily) when OMI swithces to midlat settings (lower gain settings). But as the light increases, many UV1 pixels get saturated again. Since the ozonehole does not have an infinite width the pixel saturation ends before OMI switches to tropical settings.

To prevent the UV1 pixels from saturating one can do three things:

• Use 'ozonehole' settings (lower gain factors)
• End the arctic measurements and start measuring using the midlat lower gain factors earlier by:
  • 1 minute earlier (dtA=-1min)
  • 2 minutes earlier (dtA=-2min)
  • 3 minutes earlier (dtA=-3min)
• End the midlat measurements and start measuring using the tropic lower gain factors earlier by:
  • 2 minutes earlier (dtM=-2min)
  • 4 minutes earlier (dtM=-4min)
  • 6 minutes earlier (dtM=-6min)

During the 2004 ozonehole season ozonhole settings with lower gain settings were used for the south arctic measurements. But this was not sufficient to prevent the ADC saturation at the height of the ozonehole season. In addition, the arctic measurements were ended 1/2/3 minutes earlier and this method turned out to be a better way to prevent ADC saturated pixels. Therefore, it was decided for future ozonehole seasons to end the arctic measurements at an earlier time and not use special 'ozonehole' gain settings.

During the 2005 ozonehole season it turned out that ADC saturation also occured during the midlat measurements. The problem starts at around 5 minutes before the end of the midlat measurements and ends before the end of the midlat measurements. In addition to shifting the south arctic measerment end by 1/2/3 minutes the south midlat measurement end was shifted by 2/4/6 minutes. When there remain ADC saturated pixels during the midlat measurments the midlat measurements are skipped and the south arctic measurements are directly followed by the tropical measurements.

Ozonehole monitoring

Preferably one likes to change settings before any pixels are saturated. There are two ways to monitor the pixel performance:

• Use Level 0 data in DATOS to check the pixel values manually
  • Advantage: data available after 6 hours
  • Disadvantage: Manual work
• Use Level 1 data on TLCF to check the pixel values automatically
  • Advantage: automated job
  • Disadvantage: data available after a day.

Monitoring of the Ozonehole using L0 data

L0 data is downloaded from the datastorage system MOS. The DATOS program is used to browse through the images and check the maximum pixel values. In figure 2 the increase in light in the detector due to the presence of the ozonehole is clearly vissible. Which orbits will suffer from ADC saturated pixels is determined by the shape of the ozonehole and the earth surface beneath it and is subject to change as the shape of the ozonehole changes from one day to the next. Going through all images of the day is painstakingly slow.

Monitoring of the Ozonehole using L1 data

The L1 data (hdf files) are stored at the TLCF. The automated process works in the following way:

• Find data of yesterday (day: <dd>-<mm>-<yyyy>)
• Find the last south arctic image (lsa), last south midlat image (lsm), last lsm - 30s, lsm - 60s, ... lsm - 330s and lsm - 360s
• Define a pixel box in UV1
• Determine the sum value , maximum value and the number of bad (ADC saturated) pixels in the box
• Write out these values for each orbit of the day to an array file.
• Read in the array files of all days up until yesterday
• Plot for each orbit the total pixel value for the last south arctic image and the south midlat images.
• Plot for each orbit the number of bad (ADC saturated) pixels for the last south arctic image and the south midlat images. Note: ADC saturated pixels are flagged as bad pixels, but a bad pixel is not necessaryly an ADC saturated pixel. The bad pixels that occure in every orbit are no ADC saturated pixels related to the ozonehole presence.
• Determine for each day a mean total pixel value and plot the trend
• Determine a total pixel value above which pixel saturation might take place (this threshold is determined experimentaly).

The automated job runs every day (only during the ozonehole season). The plots can be viewed to monitor how fast the ozonehole grows and to estimate when the total pixel values will cross the "ADC saturation" threshold. For the estimate the forecast predictions from the temis website can be used.

South arctic ozone throughout the year

Note that in 2008 (and all other years) ADC saturated pixels also occur outside the ozonehole season. It occurs mostly in the last minute to minute and a half of the south midlat measurements. In figure 3 the mean pixel value versus time for each last south midlat image is shown for the whole of 2008. It is clear that there are also many orbits with ADC saturated pixels outside the ozonehole season.

In figure 4 the number of bad and ADC saturated pixels versus time for each last south midlat image is shown for the whole of 2008. It is clear that the number of ADC saturated pixels can also be large outside the ozonehole season.

In figure 5 the daily mean pixel value versus time for each last south midlat image is shown for the whole of 2008.

To view plots for other moments in the orbit please see the following files:

• Mean pixel value versus time
• Number of bad channels versus time
• Daily mean value versus time