In this study, cloud and precipitation properties associated with the MJO are investigated for 32 days starting from December 15, 2006, based on a joint analysis of a global cloud-resolving model (GCRM) simulation and satellite measurements. The GCRM simulation used in this study is provided by the Nonhydrostatic ICosahedral Atmospheric Model (NICAM). Radiative transfer calculations are applied using Satellite Data Simulator Unit to the NICAM output to simulate 14- and 94-GHz radar reflectivities. The synthesized "observations" make the NICAM outputs directly comparable with the TRMM and CloudSat measurements. We focus on the contrast in cloud and precipitation characteristics between the wet (or convectively active) and dry (inactive) phases of the MJO. The MJO wet and dry phases are identified by a time-longitude bandpass filter applied to the TRMM observation. The same bandpass filter is projected on the NICAM simulation to separate the MJO phases in the model output.
The right figure shows contoured frequency by altitude diagrams (CFADs) constructed from the TRMM PR and a hypothetic 14GHz radar flying around the NICAM world. CFAD is the 2D histogram of probability distribution as functinos of radar reflectivity (abscissa) and altitude (ordinate). The CFAD implies that deep convection prevails in the wet phase while shallow clouds are more dominant in the dry phase. The model synthesized CFADs exhibit a qualitatively similar pattern, although snow signals (above approximately 5 km in altitude) are overestimated. The GCRM therefore tends to overly produce deep convection particularly in the MJO dry phase. The CloudSat CFAD (left figure) shows that weak echoes are observed near 10 km or higher even in the MJO dry phase. This high, weak echo corresponds to cirrus clouds, prevailing over the tropics. The simulated 94GHz CFAD is overall successful in reproducing the observed characteristics. Puzzling, however, is that the simulated 95GHz CFAD lacks signals in the domain with altitudes higher than 8 km and reflectivities higher than 0 dBZ, as if it were sharply cut off (marked in red). Signals in this domain are ascribed to snow in deep convective clouds. The lack of snow signals seemingly contradicts the excessive snow production as we have seen above.
A key to resolve the apparent contradiction lies in snow microphysics. In the radiative transfer calculation conducted to simulate the CFAD, we assumed a snow particle size distribution (PSD) exactly as prescribed in the bulk microphysical scheme implemented in the NICAM. This particular PSD, widely adopted by cloud resolving models, relies overly on large particles as snowfall becomes heavier. If particle size is comparable to or larger than the radar wavelength, the efficiency of microwave scattering no longer improves with increasing particle size. In other words, a few huge snowflakes do not produce radar echo as strong as many small snowflakes combined even if the total mass of snow is equivalent. This fact is demonstrated by the figure on the right. Three curves in the left panel are snow mass spectra for a given total mass of 0.01, 0.1, and 1 g/m3. Let us modify the PSD so that a proportion of small particles becomes higher (dashed line) than the original distribution (solid line). The resulting 94-GHz scattering crosssection (right panel) increases by almost an order of magnitude for large snow water contents. As such, a change to snow microphyics, while other parameters remain fixed, could largely control 94-GHz radar reflectivity.
The last figure shows the simulated 94-GHz CFAD again but with a modified snow microphysics. The sharp cutoff we found with the original snow PSD is now completely gone. Replacing large snowflakes with a number of small particles results in an increase of radar reflectivity. The previously empty domain has been filled in. Although the simple modification we tried here is far from optimal, the current methodology would be a possible pathway to diagnose and refine cloud-resolving models.
This study was conducted in collaboration with Prof. Masaki Satoh at Center for Climate System Research, University of Tokyo and Dr. Hiroaki Miura at Frontier Research Center for Global Change, Japan Agency for Marine-Earth Science and Technology (JAMSTEC).Publications related to this topic