The sensitivity of simulated water isotope distributions to differing climate model cloud parameters.

Jesse Nusbaumer (1,2), David Noone (2,3)

Abstract
The atmospheric hydrologic cycle plays a major role in the earth system. Not only does it have a direct impact on people’s well being through flooding and drought, but it also impacts the climate as a whole by influencing radiation and the atmospheric circulation via the presence of water vapor, clouds, and precipitation. Thus being able to model all aspects of the atmospheric water cycle is vital if one wants to effectively model and predict changes in the climate or earth system. One problem in trying to develop and improve climate models, however, is that a model can match observations well, but for the wrong reasons. Thus any additional observational constraint, particularly for sub-grid scale cloud and precipitation processes, can help improve climate simulations, and ultimately lead to more accurate climate projections. This study looks at the use of water isotope ratios as that additional observational constraint. It is known that water isotope ratios are sensitive to certain environmental conditions during phase changes that other observed quantities are not as sensitive to, and combined with new observational platforms that provide global-scale measurements of water isotopes, provide the ideal data set to test climate models and their parameterizations. In particular, a new isotope-enabled version of NCAR’s Community Earth System Model (CESM) is compared against global water isotope observations, and sensitivity tests are performed to determine which model tuning parameters, particularly those associated with sub-grid scale cloud and precipitation processes, have the largest impact on the simulated isotope ratios. The observations are then used to help determine what the values of certain tuning parameters need to be in order to produce the best match, and what those values say about issues in the underlying model climate, as well as potential scientific issues in CESM itself.