Processbased Evaluation of Largescale Models to Gain Confidence in Cloud Feedback Determination

Cloud-climate feedbacks can be simulated with general circulation models (GCMs). To gain confidence in the results from these simulations, the reliability of the physics of climate models (e.g., the representation of turbulence, convection, aerosols, and clouds) must be improved (Illingworth and Bony, this volume). For this purpose, a well-recognized methodology, which forms the basis of the GEWEX cloud system studies, can be employed: the physics of climate models within a single-column framework are compared with observations from field experiments and/or LES or CRM simulations driven by observed forcings (Browning et al. 1993; Randall et al. 2003). The resulting parameterizations are then evaluated in a full 3-D GCM with global datasets to assess whether an improvement of cloud representation has been achieved. Such an approach can be very powerful in pointing out deficiencies in model parameterizations and in improving model parameterizations.

Satellite observations provide particularly well-suited datasets to evaluate GCM processes, because of their global coverage and the availability of large statistics (Illingworth and Bony, this volume). Recent spaceborne active remote-sensing data providing information about the vertical distribution of cloud-related quantities (CALIPSO, CloudSat), as well as precipitation radar (TRMM) and advanced passive instrument retrievals, are especially valuable to enhance the understanding of processes and enable the evaluation of climate models. This, in turn, has enhanced our understanding of processes and enabled the evaluation of climate models. The development and application of satellite simulators in climate models that predict sensor responses assures the comparability of simulations with actual observations (Webb et al. 2001; Bodas-Salcedo et al. 2008; Chepfer et al. 2008). A persistent shortcoming, however, is the lack of high-resolution water vapor retrievals and its probability density function (PDF), which forms the basis of statistical cloud parameterization schemes and has a large impact on cloud processes. Potentially, differential absorption and Raman lidar technology should help improve this situation in the future.

The general consistency of climate models with the global atmospheric models used for NWP suggests that it is possible to evaluate climate models in the NWP framework. Comparison of the representation of clouds in NWP models (or GCMs run in NWP mode) with observations (e.g., satellite data or networks of instrumental sites such as Cloudnet; Illingworth et al. 2007) can identify strengths and shortcomings in NWP parameterization schemes. Parameter values that are applied in parameterization schemes used in an NWP model run with data assimilation may be rejected if they lead to systematic and unrealistic tendencies in short-term forecasts (Rodwell and Palmer 2007). This technique might be adapted to identify features or errors in short-term cloud simulations that map directly onto features in long-term cloud radiative feedbacks.

Systems to predict the climate evolution on a decadal timescale are currently under development. Once in place, they may permit hindcast simulations to be performed, from which information about cloud-climate feedbacks could be inferred.

Despite the fundamental importance of physical parameterizations to improve climate model simulations and reduce uncertainties in climate projections, very few people are actively involved in the development, evaluation, and improvement of physical parameterizations. Progress in developing reliable model-based projections of future climate change could be greatly enhanced by these activities, for which considerable time and energy are required.

0 0

Post a comment