On the Uncertainty in Estimates of Atmospheric Heating Due to Data Postprocessing

Abstract

Using a general circulation model (GCM) dataset, the impact of postprocessing interpolation from model sigma to pressure coordinates on a diagnostic analysis of the atmospheric energy balance is examined. Various isobaric resolutions are chosen that correspond to those provided in existing analysis archives generated by the National Meteorological Center and the European Centre for Medium-Range Weather Forecasts. Large differences are found between diabatic heating computed residually in isobaric coordinates versus sigma coordinates. Vertically averaged heating reconstructed from pressure level circulation data is found to be in error locally by 10%-50%, with poorer results occurring at coarser vertical resolution. For the zonally averaged isobaric heat balance, the error exceeds the total zonally averaged heating near the tropopause and near the surface. In the former region, the heating is quite small, whereas in the latter region it exceeds 1 K day-1. In contrast, reconstructing the heat balance directly from the GCM sigma-level circulation data introduces errors generally less than 5% of the model's explicit heating. Although uncertainties exist in our results stemming from the use of a single GCM dataset and a particular vertical interpolation method, they suggest the importance of performing analyses on an assimilating model's coordinate surfaces for the purpose of quantitative climate monitoring.