Generally, the retrieval of RO parameters starts with phase delay measurements and satellite orbit parameters. In the Wegener Center processing scheme a priori information is only involved at the stage of the bending angle initialization via statistical optimization (Gobiet and Kirchengast 2004; Borsche et al. 2006; Foelsche et al. 2008; Gobiet et al. 2007), leading to refractivity, geopotential height, and—by means of the hydrostatic equation—to pressure. Temperature is derived assuming a dry atmosphere, which is valid for upper tropospheric/lower stratospheric levels (e.g., Foelsche et al. 2008). In the mid- to lower troposphere the moist-dry ambiguity inherent in refractivity can only be resolved by means of background information, such as tropospheric temperature and surface pressure (e.g., short range forecast data from numerical weather prediction), for the retrieval of humidity (Kursinski and Hajj 2001). Thus, RO data yield good temperature results in the UTLS, while specific humidity only should be interpreted in the lower and middle troposphere and bearing in mind that background information is included.
Regarding RO data, we thus only consider results above 400 hPa for refractivity, geopotential height, and dry temperature, but for the sake of completeness, we also show the lower levels in Fig. 5. The focus of RO climatologies still lies on large-scale zonal means, but recent satellite missions such as COSMIC (Anthes et al. 2008) will provide an increasingly insight also into regional analysis due to a better spatial coverage by occultation events in consequence of a whole constellation of satellites in orbit.
The most pristine RO parameter closest to bending angle (on the applicability of the latter for climate change detection see Ringer and Healy 2008), refractiv-ity, shown for MAM, is approximately inversely proportional to temperature and a suitable trend indicator between 300 hPa and 100 hPa and between 30 hPa and 10 hPa throughout all seasons and large-scale zonal bands. For (late) winter (DJF and MAM) continental regions in mid and high latitudes of the Northern Hemisphere, the defined criteria for trend indicators are not fulfilled, due to higher variability and thus worse goodness-of-fit values in these regions.
Geopotential height of pressure levels (shown for JJA) can be interpreted as an integrated tropospheric bulk temperature and allows to define the UTLS from about 400 hPa to 20 hPa in almost all—not only large-scale zonal bands—regions as trend indicator area. Similar to temperature, northern hemispheric continental winter regions are excluded from being indicator regions, while for JJA only the southernmost regions (SSA, SAU, and ANT) do not comply with our trend indicator definition.
Thus, refractivity and geopotential height already suffice to cover the whole UTLS with RO accessible trend indicators.
Nevertheless, temperature is the most commonly used and interpreted parameter in climate science. RO temperature trends are generally best represented in the UT (about 400-100 hPa) and LS (about 30-10 hPa). In northern hemispheric winter (DJF, see Fig. 5), the continental regions of the mid and high latitudes again disqualify as trend indicators due to worse goodness-of-fit values as a consequence of higher atmospheric variability.
Specific humidity was only investigated up to 100 hPa since stratospheric water vapor concentrations are not accessible by RO. The emphasize of RO accessible specific humidity is, as specified above, on lower and middle tropospheric levels (up to 300 hPa). There, zonal means (except in the tropics above 700 hPa, which show larger variations) and the Arctic region are good indicators in SON (shown in Fig. 5) and all other seasons; individual IPCC regions generally show too much variability below 300 hPa.
Overall, the GCMs show similar trend pattern in spring and autumn, but the strongest signals are generally found in northern hemispheric summer, when internal variation is less than in other seasons.
Was this article helpful?