| description abstract | As part of its mandate to oversee the design of measurement networks for future weather and climate observing needs, the North American Atmospheric Observing System (NAOS) program hypothesized that replacing some of the existing radiosonde stations in the continental United States (CONUS) with another observing system would have little impact on weather forecast accuracy. The consequences of this hypothesis for climate monitoring over North America (NA) are considered here by comparing estimates of multidecadal trends in seasonal mean 500-mb temperature (T) integrated regionally over CONUS or NA, made with and without the 14 upper-air stations initially targeted for replacement. The trend estimates are obtained by subsampling gridded reanalysis fields at points nearest the 78 (126) existing CONUS (NA) radiosonde stations and at these points less the 14 stations. Trends in T for CONUS and NA during each season are also estimated based on the full reanalysis grid, but regardless of the sampling strategy, differences in trends are small and statistically insignificant. A more extreme reduction of the existing radiosonde network is also considered here, namely, one associated with the Global Climate Observing System (GCOS), which includes only 6 (14) stations in CONUS (NA). Again, however, trends for CONUS or NA based on the GCOS sampling strategy are not significantly different from those based on the current network, despite the large difference in station coverage. Estimates of continental-scale trends in 500-mb temperature therefore appear to be robust, whether based on the existing North American radiosonde network or on a range of potential changes thereto. This result depends on the large spatial scale of the underlying tropospheric temperature trend field; other quantities of interest for climate monitoring may be considerably more sensitive to the number and distribution of upper-air stations. | |
