Climate models exhibit exceedingly large disparities among themselves and with observations for precipitation, which is a key component of the hydrological cycle and water availability. This is arguably the single biggest issue in the credibility of climate models for almost any use. Even for global mean precipitation in the latest generation of models the spread is well over 20%, fairly uniformly distributed, and all values are larger than the “observed” value. Regional differences can be as large as the mean rainfall itself. Some of these discrepancies result from realization-dependent natural variations, which can now be assessed using large ensembles of model runs. Observations of the hydrological cycle contain major uncertainties and are often produced in a piecemeal fashion. However, a combined water and energy balance approach is now viable in analyzing reanalyses and model data, so that conservation law constraints can be utilized to assess the uncertainties, a key aspect addressed here.

The first goal is to enable improved evaluation of model precipitation by better determining the observed precipitation and its uncertainties associated with measurement and sampling errors and natural variability, along with those of the other components of the hydrological cycle. New Global Precipitation Measurement (GPM) observations will help after the launch of the GPM core observatory in 2014, along with other products (e.g., GRACE (Gravity Recovery and Climate Experiment), space-based soil moisture, runoff). Atmospheric reanalyses will be used to determine atmospheric transports and divergences and help define surface E-P (evaporation minus precipitation) from the atmospheric moisture budget. Ocean (re)analyses will be used to provide salinity. Precipitation will be assessed in the context of the entire water cycle and information from sub-daily variations will be a particular objective. The second goal is to evaluate all aspects of precipitation characteristics (amount, frequency, intensity, duration, diurnal cycle, annual cycle, variability on multiple time scales, and extremes) in the CMIP5 models in general and the CESM suite of models at NCAR in particular to isolate the main regions, processes, and times where major issues are apparent. We plan to develop a number of new metrics that enable holistic process-oriented model evaluations that should help diagnose the deficiencies in underlying model physics leading to model improvements. The third goal is to extensively explore interannual and decadal variability of both the mean and characteristics of precipitation using the NCAR large ensemble of model runs with a focus on natural modes of variability such as ENSO and the Pacific Decadal Oscillation that have been shown as key components of the recent hiatus in global mean temperature rise. The final subsidiary goal is to work through the NCAR “Water Cycle Across Scales” project with modelers on possible new parameterizations, for instance of convection, to improve characteristics of precipitation in terms of spatial structures and temporal variability.