In order to exemplify the capability of the presented diurnal cycles of cirrus clouds to validate climate models the mean cirrus coverage derived by COCS retrieved for 2010 is now compared to the latest operational forecast model of the European Center for Medium range Weather Forecast (ECMWF), the Integrated Forecast System (IFS cycle 36r1, 36r2, and 36r4). In general, the ECMWF forecasts cover the time period from July 1985 to the present. In September 2006 a new feature, ice supersaturation in the upper troposphere, was included in the IFS [Tompkins et al., 2007]. This new feature produced some changes in the statistics of upper tropospheric humidity and cloud coverage. In particular, there is an increase in upper-tropospheric humidity and a decrease in high-level cloud coverage and cloud ice amounts.
The ECMWF provides global weather forecasts twice daily (00:00 and 12:00 UTC) with output time steps of three hours for the first three days. The model uses a resolution of about 25 km at the equator and a vertical resolution of about 15 hPa.
Lamquin et al. [2009] stated, that ECMWF uses an assimilation scheme that does not
Figure 5.11: Cirrus coverage retrieved by COCS (red) and high cloud coverage of ECMWF IFS (blue) for the two convective dominated regions SAC (left) and MED (right).
account for ice supersaturation in the upper troposphere. Hence, data assimilation leads to analyses that underestimate the true occurrence and range of ice supersaturation. Since the analyses serve as initial conditions for the forecast runs, the forecast model needs some time for spinup of the supersaturation field. Studies of upper tropospheric humidity should not use forecast humidity data from the spinup phase since they are unreliable. This initial study of the global supersaturation spinup behaviour shows that at least twelve hours are necessary before the model forecasts humidity correctly and thus cloud data in the upper troposphere. Therefore, the IFS high cloud coverage forecasts of the ECMWF operational data archive for the year 2010 are used with a spinup time of
Of course the temporal resolution of three hours cannot show all details of the diurnal cycle retrieved by COCS with 15 min resolution, but it gives a first hint on whether the cloud coverage calculated by the ECMWF shows similar structures. Please note that ECMWF IFS and COCS data are now analysed as a function of UTC and no longer as a function of LT.
Again the regions dominated by convection, SAC and MED, are depicted first (Fig. 5.11). On the left, the plot shows the two curves of the cirrus coverage retrieved by COCS for SAC (red), and the high cloud coverage retrieved from ECMWF IFS forecasts (blue) for the same region. COCS calculates a slightly higher cirrus coverage (∼ 3 %), but both curves show a similar diurnal cycle with the convective maximum in the late afternoon between around 17:00 to 18:00 UTC and the minimum in the morning (at around 08:00 UTC). On the right, the MED region shows higher differences in the detected coverage. COCS detects around 9 % more cirrus clouds. Nevertheless, both algorithms represent a similar diurnal cycle with less variations than in the SAC region, and again a maximum dominated by convection between 15:00 and 17:00 UTC is found, while the minimum is also found in the morning at around 08:00 UTC.
In the two Southern Atlantic regions, SAR1 and SAR2, the cirrus coverage is present
Figure 5.12: Cirrus coverage retrieved by COCS (red) and high cloud coverage of ECMWF IFS (blue) for the two South Atlantic regions SAR 1 (left) and SAR 2 (right).
with no or very low influence of convection and without the influence of air traffic as it is the case in NAR. The cirrus coverage of both regions is depicted in Fig. 5.12. On the left, the smaller region SAR1 show the double-wave frequency in the cirrus coverage detected by COCS with two minima (at 02:00 and 14:00 UTC) and two maxima (at 07:00 and 21:00 UTC). The ECMWF shows slightly less cirrus coverage (∼ 3−5 %) and follows the diurnal cycle detected by COCS in case of the first maximum in the morning and the second minimum in the afternoon, but it misses the second maximum in the late evening
and the first minimum shortly after midnight.
A larger difference in cirrus coverage is found for the bigger South Atlantic region, SAR2, on the right of Fig. 5.12. COCS detects an around 7 % higher cirrus coverage, but the diurnal cycle is present in the ECMWF forecast as well as in COCS, but shows less variations in the time after the minimum at 12:00 UTC. ECMWF and COCS detect a maximum cirrus coverage at around 05:00 to 06:00 UTC.
Fig. 5.13 depicts cirrus coverage from the ECMWF forecasts and COCS for NAR. While
Figure 5.13: Cirrus coverage retrieved by COCS (red) and high cloud coverage of ECMWF IFS (blue) for the North Atlantic region NAR.
the value of cirrus coverage generally shows an offset of 4−5 % at around midnight, COCS detects significantly more cirrus coverage with a strong diurnal cycle. Two maxima at 07:00 and 19:00 UTC are present with coverage up to 46 % and 45 % respectively. The minima in cirrus coverage are located at 00:00 (43.5 %) and 13:00 UTC (44 %). The ECMWF forecast shows one maximum at midnight with 39 % high cloud coverage and one minimum at 12:00 UTC with 38 %.
In summary, the natural cirrus coverage forcasted by the ECMWF and its diurnal cycle is well represented in the cirrus coverage retrieved by COCS. In general COCS overestimates the cirrus coverage or the IFS underestimates the high cloud coverage. Only NAR, which is strongly influenced by air traffic due to the North Atlantic air corridor, shows a strongly modified diurnal cycle.
Figure 5.14: Mean air traffic density in Europe and the North Atlantic for the NAR region is marked by the solid black box. The black thick curve represents the 75◦ satellite zenith angle of MSG-8/9 [Graf et al., 2009].