Improving the Presentation and Reliability of OMI Ozone Retrievals

Send to Kindle

[Unsolicited report to the OMI ozone teams at the Goddard Spaceflight Center and KNMI (Royal Netherlands Meteorological Institute).]
20 September 2011


Forrest M. Mims III


Two revisions to the presentation of Aura OMI ozone overpass data will vastly improve the value of the data and the steps required for its analysis:

1. All ozone retrievals contaminated by row anomalies should be removed, not merely flagged, from all levels of KNMI and GSFC OMI data products. Contaminated data should be displayed only after correction.

2. The KNMI and GSFC OMI data should be presented on compatible, 365-day linear calendars with blank rows for missing data and a second set of parameter columns for second overpasses on the same day). Ideally, all KNMI and GSFC data should be bundled into the same spreadsheet so they may be more conveniently compared.


The GES DISC provides a remarkable service, especially for those of us who are comparing ground observations with measurements made by various satellites during overpasses.

Since 1990 I have been comparing my ground-based measurements of total ozone, total water vapor and aerosol optical depth (AOD) with a series of both NASA and NOAA space-based instruments. My instruments have been calibrated (Langley method) and compared with standards at the Mauna Loa Observatory every spring since 1992. The TOMS ozone mapper on the NIMBUS 7 provided a calibration baseline so good that I was able to later find and publish a drift in the TOMS ozone data. Later I found anomalies with other TOMS and NOAA's AVHRR instruments. Of special interest was the excellent agreement between my ground-based measurements of AOD with those by the MODIS instrument on Terra and Aqua, especially when comparisons were made within 60 seconds of the peak overpass time.

My long term experience with NASA satellite data has been very positive and includes overpass studies for GSFC in Brazil (1995 and 1997) and at seven forest fires in Western States. Today I write to report a negative experience involving problems with ozone retrievals from the OMI instrument on Aura.

Row anomalies with OMI data have occurred since 2007. Manually identifying and removing the defective ozone data (and presumably other products) on the OMI web pages is difficult, but failing to do so can have significantly adverse consequences for ozone trend studies and comparisons made by ground-based observers.

Beyond the row anomaly issue is the fact that there are two separate OMI ozone products from KNMI and GSFC. As shown in the plots below, the correlation of the GSFC ozone product with a calibrated and very stable Microtops II instruments is much higher than that with the KNMI algorithm. Data users, especially those doing trend studies, need to be informed about such differences. I learned that the OMI ozone data include defects, some quite serious, only because a well calibrated ground instrument measures the total column abundance of ozone better than OMI. Most people do not have this advantage.


KNMI – GSFC: r2 = 0.53

GSFC – MICROTOPS: r2 = 0.73

KNMI – MICROTOPS: r2 = 0.67

Typical past comparisons of Microtops and pre-OMI satellite ozone instruments: r2 > 0.9

Another complicating factor is that the KNMI algorithm product includes some days that are not included under the GSFC algorithm.

These issues have had a major impact on my 22-year trend studies of the ozone layer, for the OMI data cannot be considered as reliable as most prior TOMS data. Simply put, OMI ozone data is often much noisier than TOMS data. (Before confirming this, for well over a year I assumed that my instrument was the problem.) Also affected is the inference of UV-B for this site (using the Canadian AES model) based on OMI ozone retrievals for comparison with full sky UV-B irradiance made here since 1994 using an array of instruments.

If these issues are affecting my work, they may be affecting research by those unfamiliar with the magnitude of the problem and who might make erroneous decisions and even submit publications about erroneous trends in the abundance of total ozone and ground-level UV-B inferred from the OMI data, especially the KNMI product.

With respect, I suggest the following two revisions for the presentation of the online OMI ozone overpass data:

1. It is urgent that the current OMI ozone listings be supplemented with a new spreadsheet that deletes all data that is flagged as erroneous. Until this is done there is a high risk that some students and professional analysts will derive incorrect conclusions about the status of the ozone layer and the trends of its abundance.

2. It is very important to display the OMI ozone data (both KNMI and GSFC) using a linear calendar method (365 rows of data or missing data) so that plots and data comparisons can be easily made.

The linear calendar should include two sets of parallel data to allow for days with two overpasses. (The second set of columns would duplicate the first set of parameters but would contain the data for the second overpass on that day.) Ideally, all KNMI and GSFC data should be bundled into the same spreadsheet so they may be more conveniently compared. This approach has been used for decades to compare the carbon dioxide measured at the Mauna Loa Observatory by two independent instruments operated by Scripps and NOAA.


Presenting only valid data for the various levels completely free of row anomaly contamination and organized in a conventional 365-day calendar format will greatly enhance the utility and value of OMI ozone retrievals. Especially important is presenting both the KNMI and GSFC products in the same 365-day format, ideally merged into the same spreadsheet so that the data can be statistically compared in only seconds instead of hours. This will permit analysts to very quickly spot the differences between the ozone inferred from the KNMI and GSFC algorithms and with both these products and ground-based data.


Row anomalies with OMI data have occurred since 2007, as reported at and in detail at

AES. Atmospheric Environment Service (Canada)

GES DISC. Goddard Earth Sciences (GES) Data and Information Services Center (DISC) (NASA)

GSFC. Goddard Space Flight Center (NASA)

KNMI. Netherlands's Agency for Aerospace Programs

MODIS. Moderate Resolution Imaging Spectroradiometer (aboard Terra and Aqua satellites)

OMI. Ozone Monitoring Instrument (aboard Aura satellite)

TOMS. Total Ozone Mapping Spectrometer (discontinued NASA ozone satellite instrument)

This entry was posted in Amateur Science, Climate, Environment, Meteorology. Bookmark the permalink.

Leave a Reply