Diagnostic Computation of Moisture Budgets in the ERA-Interim Reanalysis with Reference to Analysis of CMIP-Archived Atmospheric Model Data*

Abstract The diagnostic evaluation of moisture budgets in archived atmosphere model data is examined. Sources of error in diagnostic computation can arise from the use of numerical methods different from those used in the atmosphere model, the time and vertical resolution of the archived data, and data availability. These sources of error are assessed using the climatological moisture balance in the European Centre for Medium-Range Weather Forecasts Interim Re-Analysis (ERA-Interim) that archives vertically integrated moisture fluxes and convergence. The largest single source of error arises from the diagnostic evaluation of divergence. The chosen second-order accurate centered finite difference scheme applied to the actual vertically integrated moisture fluxes leads to significant differences from the ERA-Interim reported moisture convergence. Using daily data, instead of 6-hourly data, leads to an underestimation of the patterns of moisture divergence and convergence by midlatitude transient eddies. A larger and more widespread error occurs when the vertical resolution of the model data is reduced to the 8 levels that is quite common for daily data archived for the Coupled Model Intercomparison Project (CMIP). Dividing moisture divergence into components due to the divergent flow and advection requires bringing the divergence operator inside the vertical integral, which introduces a surface term for which a means of accurate evaluation is developed. The analysis of errors is extended to the case of the spring 1993 Mississippi valley floods, the causes of which are discussed. For future archiving of data (e.g., by CMIP), it is recommended that monthly means of time-step-resolution flow–humidity covariances be archived at high vertical resolution.