G3.05    Representativeness uncertainty assessment missing for higher-level data based on averaging of individual measurements

Gap detailed description

The creation of level-3 (and level-4) data by averaging non-uniformly distributed measurements inevitably leads to representativeness errors, see e.g. Coldewey-Egbers et al., (2015) for the case of a level-3 (gridded monthly means) total ozone data set. The resulting representativeness uncertainty can be larger than the formal uncertainty on the mean. However, estimates of these representativeness uncertainties are rarely included with the data product. Also, the representativeness of the ground-based network should be taken into account when validating such data sets, i.e. the sparse spatial and temporal sampling of the ground network leads to significant representativeness uncertainties in the derived monthly (zonal) means.

Also, in the context of validation of level-2 data, measurements are sometimes averaged after co-location (e.g. Valks et al., 2011, Schneising et al.,2012) without explicit calculation of the representativeness errors and resulting uncertainty.

Activities within GAIA-CLIM related to this gap

No work in this direction is foreseen within GAIA-CLIM.

Gap remedy(s)

Remedy #1

Specific remedy proposed

Studies quantifying the representativeness of averages, e.g. by model-based simulations of averages based on either the limited real sampling or on an ideal, complete sampling. More pragmatically, representativeness uncertainties can also be computed as a function of parametrized measurement inhomogeneity and field variability (Sofieva et al., 2014).

Measurable outcome of success

Success is achieved when level-3 data sets include not only the formal uncertainty on the mean and the variance around that mean, but also an estimate of the representativeness uncertainty on that mean.  The reliability of this reported representativeness uncertainty must than also be validated, e.g. with targeted intensive field campaigns. 

Achievable outcomes

Technological effort required to address this gap depends on the particular product and on whether atmospheric variability is well understood for that ECV (cfr. gap G3.01). For most of the ECVs targeted by GAIA-CLIM, an estimate of the representativeness uncertainty should be achievable at a low cost.  The additional validation required to assess the quality of this representativeness uncertainty estimate may –in absence of existing reference data sets at sufficiently high spatial and temporal sampling–  require a more significant investment, e.g. to conduct intensive field campaigns.


This remedy offers a comprehensive solution to the gap.


The calculation of representativeness uncertainties is not time consuming and can be done on a time scale of weeks or months. For this to become common practice among data product creators may take several years.

Gap risks to non-resolution

Identified future risk / impact

Probability of occurrence if gap not remedied

Downstream impacts on ability to deliver high quality services to science / industry / society

Representativeness uncertainties unknown and unreported.

Very high

Potential over-interpretation of the data due to representativeness uncertainty not being taken into account.


Work package: