GEOS-5 atmospheric modeling and diagnostics
October 26, 2012 15:26:12
Description of Problem
An overarching goal of the GMAO atmospheric modeling effort is to develop a single atmospheric model suitable for data assimilation, weather forecasting and climate simulation. Climate simulation includes atmosphere only, coupled ocean atmosphere, and coupled chemistry-climate modes. The model’s collection of physical parameterizations is of central importance to the success of the GMAO’s modeling effort.
Part of this year’s effort was focused on the final model changes that led to improved atmosphere-ocean coupled climate simulations, and resulted in the release of the model to be used for decadal climate prediction. This year’s effort was also focused on the analysis and development of the model in data assimilation mode and at higher resolution.
Scientific Objectives and Approach
The basic approach is through multiple simulations in climate, weather forecasting and assimilation modes, analysis of model results based on comparison with observations, and new experiments based on hypotheses to explain model-validation inconsistencies. This includes the development of innovative ways in which to compare model and observations, and innovative ways to inform the model’s parameterizations using observations.
The Fortuna 2_2 and 2_3 versions of the GEOS-5 GCM were developed, tested and released. Results of coupled atmosphere-ocean climate simulations with the new version exhibited improvements over the previous version, in particular as regards the cloud cover and net surface radiation, the latter of which is critical for accurate coupled simulations. Towards this end adjustments were made to the moist processes, in particular to the vertical profile of the critical relative humidity criterion for phase changes of in-cloud and precipitating liquid and ice, and to the cloud forcing algorithm for high latitude regions.
Examination of high resolution simulations with GEOS-5 was performed by GMAO personnel, in climate mode and in data assimilation mode, and several issues were identified. These included a dry bias in the tropics near 800mb, a high bias in the minimum sea level pressure reached during tropical storms and hurricanes, a high bias in the surface winds over the Southern ocean, and a warm bias in summertime land surface temperatures. Several changes to model parameterizations were made during the course of the year to address these issues.
A study of the critical relative humidity for condensation (RHcrit) using sub-grid scale variations of water vapor from a 10-km simulation with GEOS-5 was performed to examine the dependence of RHcrit on horizontal resolution. It was found that the sub-grid scale variability decreased with increasing horizontal resolution, and this translated into a higher RHcrit. Simulations and data assimilation runs with the resolution-dependant RHcrit showed improvements in atmospheric moisture levels relative to observations.
The remaining issues identified were addressed by several changes to the model parameterization of surface layer turbulence. The warm bias over land in summertime was substantially reduced by removing the laminar sublayer over vegetated land. The laminar sublayer acted to inhibit the turbulent transfer of heat from the surface, causing warm temperatures, and its removal increased the turbulent transfer of heat away from the surface. The minimum sea level pressure and maximum winds reached during tropical storms and hurricanes were improved by adding an upper limit to the ocean roughness at extreme surface wind speeds, based on many existing studies. The improvement was documented with a series of 10 hurricane and typhoon forecasts. The ocean roughness parameterization in the medium wind regime was also updated, increasing the ocean roughness based on modern observationally-based studies. This resulted in the reduction of surface winds in the Southern Ocean.
The development effort for the GEOS GCMs physical parameterizations during the coming year will focus primarily on the identification and removal of model behavior which is discontinuous in nature. Analysis of GCM and single column simulations has pointed to errors specifically connected to the discontinuities in the behavior of the turbulence and moist parameterizations. The aspects of the parameterizations that will be specifically addressed include the assumption of a ‘top-hat’-shaped sub-grid scale distribution of water vapor and sudden onset of phase changes in the prognostic cloud scheme, and an abrupt end to buoyant ascent in both the convective and turbulence parameterizations. The model parameterizations will also be examined to remove behavior that is inherently dependent on vertical resolution. Removal of these discontinuities is anticipated to reduce model error and enable a more continuous response to parameter changes during model tuning endeavors. In addition to the correction of these known errors, development during the coming year will include the replacement of the current version of the cumulus parameterization with the more modern RAS-2, which includes the simulation of a cumulus downdraft and a modified mass flux entrainment profile for shallow convection.
In addition to the model development effort, the coming year will be focused on the completion of technical memoranda and journal papers presently in preparation. These publications include a technical memorandum documenting the changes in parameterizations and the connections to the improved climate simulation. A journal paper will follow, focused on explaining the improvements in tropical precipitation and global stationary wave pattern in the context of existing theory about tropical organization and teleconnections. A journal paper reporting on the study of model subgrid scale variability of total water and the connection to the parameterization of critical relative humidity will also be completed and submitted.
Other Publications and Conferences
Molod, A. (2010), NASA/GSFC GEOS-5 Contribution to CGILS, paper presented at the CGILS Workshop, Stonybrook, N.Y., March, 2, 2010