Data quality flags for GLOBE aerosol data

GLOBE sun photometer data reported to GLOBE's data entry system are subjected to a number of quality checks. This causes some data to be rejected. However, there are many cases where problems with sun photometer data are not obvious. In these cases, the aerosol optical thickness (AOT) values may still be wrong. As a result, post-processing needs to be applied to the data in order to assess their quality. Here are some data quality guidelines:

  1. Negative AOT values are not allowed. Such values are mathematically possible, but physically impossible. Sun photometer data that result in negative AOT values are not accepted by the GLOBE data processing system.
  2. Green-channel AOT values should never be less than red-channel values. Under some extreme atmospheric conditions, the red and green AOT values can approach each other. Then, green AOT values might be slightly smaller than red values. GLOBE processing of AOT values does not remove the contribution due to absorption of sunlight by ozone. Accounting for ozone reduces green-channel and red-channel AOT values by about 0.01 and 0.03, respectively. So, under clear-sky conditions, it is possible for red-channel AOT values returned by GLOBE to be a little larger than green-channel values. GLOBE sun photometer data processing does not "know" about this restriction, if values otherwise look reasonable.
  3. Within a set of 3-5 measurements of each channel, voltages should not differ by a significant amount. "Significant" is a matter of interpretation, and sometimes AOT values that pass GLOBE checks still exhibit too much variability within a set of data. This is usually a sign of poor measurement technique or contamination by clouds between observer and sun. (Contamination of measurements by thin Cirrus clouds is a common problem.)

The Science Team has established a 5-level data quality system, as described in the following table.

FlagInterpretation
1Values that have been verified by comparison with an independent and reliable source (very rare!)
2Tightly clustered values that are consistent with complete metadata, with no apparent data anomalies
3Values that have no serious apparent anomalies, but more loosely clustered, with incomplete metadata, and/or unusual values that cannot be independently verified
4Values that have apparent errors, including gross inconsistencies with metadata, apparent cloud contamination, and/or too much variability within the set of measurements
5Values based on data that are obviously in error, and should not have been accepted into the GLOBE system