|
Titel |
Improved cloud parameterization for Arctic climate simulations based on satellite data |
VerfasserIn |
Daniel Klaus, Klaus Dethloff, Wolfgang Dorn, Annette Rinke |
Konferenz |
EGU General Assembly 2015
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 17 (2015) |
Datensatznummer |
250103927
|
Publikation (Nr.) |
EGU/EGU2015-3349.pdf |
|
|
|
Zusammenfassung |
The defective representation of Arctic cloud processes and properties remains a crucial
problem in climate modelling and in reanalysis products. Satellite-based cloud observations
(MODIS and CPR/CALIOP) and single-column model simulations (HIRHAM5-SCM) were
exploited to evaluate and improve the simulated Arctic cloud cover of the atmospheric
regional climate model HIRHAM5. The ECMWF reanalysis dataset “ERA-Interim”
(ERAint) was used for the model initialization, the lateral boundary forcing as well as the
dynamical relaxation inside the pan-Arctic domain. HIRHAM5 has a horizontal
resolution of 0.25°and uses 40 pressure-based and terrain-following vertical levels. In
comparison with the satellite observations, the HIRHAM5 control run (HH5ctrl)
systematically overestimates total cloud cover, but to a lesser extent than ERAint. The
underestimation of high- and mid-level clouds is strongly outweighed by the overestimation
of low-level clouds. Numerous sensitivity studies with HIRHAM5-SCM suggest
(1) the parameter tuning, enabling a more efficient Bergeron-Findeisen process,
combined with (2) an extension of the prognostic-statistical (PS) cloud scheme,
enabling the use of negatively skewed beta distributions. This improved model setup
was then used in a corresponding HIRHAM5 sensitivity run (HH5sens). While the
simulated high- and mid-level cloud cover is improved only to a limited extent, the
large overestimation of low-level clouds can be systematically and significantly
reduced, especially over sea ice. Consequently, the multi-year annual mean area
average of total cloud cover with respect to sea ice is almost 14% lower than in
HH5ctrl. Overall, HH5sens slightly underestimates the observed total cloud cover but
shows a halved multi-year annual mean bias of 2.2% relative to CPR/CALIOP
at all latitudes north of 60Ë N. Importantly, HH5sens produces a more realistic
ratio between the cloud water and ice content. The considerably improved cloud
simulation manifests in a more correct radiative transfer and better energy budget in the
atmospheric boundary layer and results also in a more realistic surface energy budget
associated with more reasonable turbulent fluxes. All this mitigates the positive
temperature, relative humidity and horizontal wind speed biases in the lower model levels. |
|
|
|
|
|