Using C-POL Data to Calibrate a Stochastic Parametrization of Organized Convection

As GCMs present uncertainties in the simulation of subgrid scale processes, the development of stochastic approaches has become an important tool in providing skillful representations of atmospheric processes of rather non-deterministic nature, e.g. the development of a stochastic multicloud model (Khouider at al., 2010) to simulate the life cycle of the three most common cloud types (cumulus congestus, deep and stratiform) in tropical convective systems. In this model, these cloud populations interact with each other and with their environment according to intuitive probabilistic rules determined by the large scale atmospheric state and a set of transition timescale parameters. We use a statistical method based on the Bayesian inference theory (De La Chevrotiere, et al. 2015) to estimate these timescale parameters that together with the large scale predictors determine the rates of random transitions between these three cloud types. In contrast to previous studies, we extend the Bayesian method by incorporating further dynamical large scale predictors in the transition rate equations in addition to the thermodynamical predictors. Moreover the Bayesian approach is applied to data derived from CPOL radar from the TWC ICE campaign (Frederick and Schumacher 2007) and from array-averaged products from the DYNAMO field campaign (http://johnson.atmos.colostate.edu/dynamo/products/array_averages/index.html). Sensitivity tests are carried out to evaluate the estimation of the timescale parameters. As a possible validation strategy we intend to fix those timescale parameters as input parameters in a coupled multicloud-one column GCM model and compare the resulting time series for the large scale predictors and cloud populations with the observational data.