Improvements to the JPL EDMF Cloud Parameterization by a Suite of Global A-Train Satellite Observations

Often, cloud parameterizations are validated against results from Large Eddy Simulations, where both are initialized by individual idealized profiles obtained from field campaigns. While these profiles provide uniform initial conditions across studies, they also omit fine-scale variations and are targeted to specific environments or cloud types. These validation techniques therefore cannot encapsulate the wide variety of conditions that exist at the planet scale, leading to over-tuned parameterizations. This presentation will demonstrate that globally-available observations of the atmospheric state from instruments in the A-train constellation can be used to constrain internal parameters of the JPL Eddy-Diffusivity/Mass Flux (EDMF) cloud parameterization. The JPL EDMF is implemented within a single column model, which is initialized by atmospheric state profiles from the NASA MERRA. To improve model performance compared to observations, simulated cloud fraction, rain rate, and liquid water content are compared to analogous CloudSat, CALIPSO, and MODIS observations. Performing validation against observations at their native resolution allows diagnoses of not only grid-scale means, but sub-grid measures of variability. The presentation also includes a discussion of some challenges of this technique, most notably the sensitivity of the comparisons to model initial conditions and the choice of when in the simulation to perform the comparison to observations. Improvements in the representation of marine boundary layer clouds are highlighted but the techniques and discussion are applicable to global studies.