Analysis of the temporal behavior of convection in the tropics of the European Centre for Medium-Range Weather Forecasts model

Extended (180-day) high resolution (T106) perpetual January and July integrations of the European Centre for Medium-Range Weather Forecasts (ECMWF) model have been analyzed in terms of the spatial and temporal characteristics of the model's convective activity in the tropics. The model's outgoing longwave radiation (OLR) is used as a surrogate for convective activity, consistent with similar studies based on satellite observations. The 3 hourly temporal sampling is sufficient to allow diagnosis of intradiurnal and interdiurnal variability; the length of the integrations is adequate for identifying lower-frequency, intraseasonal phenomena. Wherever possible, use is made of results from surface or satellite observations of the temporal characteristics of convection to verify the model results. At intradiurnal time scales the model captures the amplitude and phase of the diurnal harmonic over both land and sea. The largest amplitudes occur over the summer continents, with contrasting phases of maximum OLR depending on the presence of convective activity. Over the oceans the model shows a coherent structure to the diurnal cycle associated with regions of convection. Analysis of synoptic (2 to 10 days) and low-frequency (greater than 10 days) variability shows that in many instances the model agrees well with observations. For both seasons the modelmore » simulates westward moving phenomena over the oceans, whose phase speed is reasonable. In July these easterly waves display well-defined periodicities, in agreement with observations, while in January they are more episodic. Low-frequency variability is more prevalent in January, particularly over the convectively active regions of the eastern hemisphere. In general, this variability has a larger spatial scale than the synoptic variability; its periodicities, some in excess of 30 days, are typical of intraseasonal time scales. 56 refs., 10 figs.« less