Development of the Observing Requirements for the GSN

The initial selection of the GCOS Surface Network stations is described in a study by Thomas Peterson, Harald Daan, and Philip Jones published in the Bulletin of the American Meteorological Society. The AOPC continues to add or remove stations from the list as circumstances require.

Developers of the GSN

The Atmospheric Observation Panel for Climate (AOPC) was established to define the requirements for meteorological observations for GCOS. AOPC is building on work began by the WMO Commission on Climatology (CCl), and the WMO Commission on Basic Systems (CBS). Both CCl and CBS also review and assist with the implementation of the observing requirements that are identified by the AOPC.

Selecting the Stations

To select the approximately 1000 stations from among the existing 10,000 or so stations making surface observations, criteria were formulated to assess  spatial coverage, availability, and homogeneity and quality of data. It was also felt that priority should be given to those stations that were listed in the WMO Reference Climatological Stations (RCS) lists and where monthly data were distributed as CLIMAT reports. It was also considered desirable to include stations that are part of the GCOS Upper-Air Network (GUAN), Regional Basic Synoptic Network (RBSN), and Global Atmosphere Watch (GAW) networks.

The network selection algorithm was optimized for temperature observations.

Selection Criteria

The network was selected to provide a homogeneous array of high quality stations of sufficient density and coverage for climate purposes. Of interest was the detection of the spatial patterns and scales of global temperature change at the surface and of global atmospheric circulation changes. The network could also be used as one basis for the quality control of denser temperature networks and for other variables (e.g., precipitation) and in providing observations for the monitoring of extremes and variability of weather. 

The criteria used for selection (according to published papers by Jones and others) were that the station record was long, homogeneous , that the temperature data were available, and that the station was currently reporting. Urban centers were avoided. To achieve reasonable spatial representation, approximately one station per 5°x5° square of the world was chosen. This resulted in a network with of the order of 800 stations. In fact there were found to be 779 5°x5° grid cells around the globe which contained at least one station.

As part of the selection process a computer algorithm which synthesized the criteria and information was used for a more objective result.