“Upper Air” as used in the context of GRUAN refers to the combined troposphere and stratosphere. GRUAN places a strong emphasis on the upper troposphere/lower stratosphere region, which has the strongest climate relevance and a large variability. Measurement capabilities for different atmospheric parameters may be strongly altitude dependent and thus the abilities to achieve climate records at different altitude may vary.
Once fully completed GRUAN will be a world-wide network of reference stations. These stations should cover all major climatic regions of our planet. In its initial stage, GRUAN is heavily centered around a number of sites in Europe and the continental United States, with one site in China, several sites in the Western Pacific and only one site in the southern Mid-latitudes. Thus there is an obvious need to establish sites in a number of regions not yet covered by GRUAN.
A reference observations is an observation that gives the best estimate for the measurand as well as the best estimate for the level of confidence in this measurement. The best estimate for the measurand is nothing else than the measurement itself, the estimate for the level of confidence is described by the measurement uncertainty. Therefore, GRUAN will strive to establish uncertainty estimates for every measurement, which implies, that for a profile of an atmospheric parameter, GRUAN will attempt to establish the corresponding profile of the measurement uncertainty. The usefulness of any measurement will be determined by the size of the measurement uncertainty.
Please refer to the publication Immler et al., AMT, 2010 for more details.
The measurement uncertainty will be established by analysing all sources of measurement uncertainty and by combining them to one single error bar for each measurement point. The investigation of all sources of uncertainty will be a time consuming and iterative process and will require a detailed understanding of the instrumentation that is being used as well as a careful consideration of the operational influences that contribute to the measurement uncertainty. Error bars during the early stages of GRUAN are expected to be somewhat crude. As the understanding of the instrumentation improves and as the instrumentation itself improves, uncertainty estimations are expected to improve as well. Thus, it is expected that there will be different versions of GRUAN data, which differ mostly in their uncertainty estimate.
Precision is the recognition that there might be systematic uncertainties in a measurement, which are irrelevant for long term trends as long as the instrumentation does not change. However, for climate records spanning multiple decades, it is expected that the instrumentation will change several times. In particular sounding equipment has gone through a rapid development in the last years and slowing down of this development is not expected. Therefore, the distinction between accuracy and precision has only limited use for long term climate records. A more suitable concept is a best estimate for the level of confidence in a measurement, which may simply be called uncertainty estimation.
Systematic errors have to be corrected. This means that studies are required to identify and quantify systematic biases. The solar radiation correction for the temperature measurement on radiosondes is a classic example for such a systematic bias that requires correction. Since no correction algorithm is perfect, this correction will then introduce a new and hopefully random uncertainty, which is then combined with all other sources of uncertainty.
No. GRUAN stations will have a variety of instrumentation and observational platforms. The GRUAN challenge will be to homogenize the observations of this diverse set of instruments to establish a consistent data set that can be used for climate research, satellite validation and the characterization of the atmospheric column. However, to achieve homogeneity of observations between different stations identical treatment of observations is a must. This means that observational data have to be analysed to the same detailed level and that uncertainty estimations have to be done consistently across all stations. To verify that observations and uncertainty estimates at different sites using different instrumentation leads to consistent results frequent intercomparisons of these instruments are required. This may be achieved either in dedicated intercomparison campaigns at dedicated sites or routine intercomparisons as part of the observational program.