forecasts and hindcasts of MJO events during CINDY2011/DYNAMO field campaign using a global nonhydrostatic model NICAM

In order to advance our understanding of the mechanisms of MJO, in-situ field observations over the Indian Ocean and MJO assessment by operational products have been conducted in the international project Cooperative Indian Ocean experiment on intraseasonal variability in the Year 2011 (CINDY2011) / Dynamics of the Madden-Julian Oscillation (DYNAMO). One of the major objectives of this project is to improve forecast skill and reproducibility of MJO in numerical models, and post-campaign studies using various kinds of models are executed, as well as real-time forecasts to support field operation. As one of the modeling activities in CINDY2011/DYNAMO project, week-long near real-time forecasts using a regionally stretched version of Nonhydrostatic Icosahedral Atmospheric Model (NICAM) have been operated with a horizontal grid size of 14~30 km over the warm pool region. Although the resolution of the model is too coarse to resolve convective clouds, the forecasts are capable of capturing large-scale dynamical fields and organization of convection by marginally resolving mesoscale convective systems (MCS) with explicit moist processes. In preparation for the field campaign, week-long simulations were conducted during May-August 2011, and near real-time forecast has been in operation throughout the observation period (September 2011-January 2012). Preparatory simulations showed that correlation coefficient of 850 hPa zonal wind field between the forecasts and the objective analysis data (NCEP final analysis) kept 0.6-0.8 after 7-day integration, and the monsoon index in the boreal summer season followed well that in the real atmosphere. In the simulations for the period including intraseasonal variability (ISV) associated with Indian monsoon onset in 2011, several cyclonic disturbances were generated as the ISV progressed over Arabian Sea, Bay of Bengal, and the South China Sea. The simulations predicted genesis of these cyclonic disturbances, although location, timing, and tracks of these disturbances were not accurately simulated for the integration time beyond a few days. The near real-time forecast during the CINDY2011/DYNAMO intensive observation period (IOP) reproduced multi-scale structures of convection in the onset phase of MJO over the Indian Ocean that occurred in the second half of October 2011; the simulations clearly reproduced westward-propagating convective systems in an eastward-propagating large scale convective envelope. It is also noteworthy that eastward-propagating shallow convective systems that passed over the observation array prior to the active phase of MJO were marginally simulated in the near real-time forecasts. Preliminary analysis suggests that these convective signals were associated with the passage of equatorial Kelvin waves (Yamada et al.'s poster). Convective patterns that frequently appeared during the IOP, such as double ITCZ-like organization and those associated with easterly waves were fairly simulated. The near real-time forecasts also captured successive tropical cyclogenesis over the Arabian Sea (Kelia and 04A) that occurred in the end of October (during the MJO phase 2-3). However, rapid growth of cyclones occurred generally too early in the model. The correlation score in the CINDY2011/DYNAMO period was generally lower than that in the summer season (0.5-0.65 after 7-day integration). As to MJO forecast skill, phase propagation of MJO was fairly predicted, but with systematic errors in the amplitude; under-prediction for phase 6-8 and 1, and over-prediction for phase 2-5. These forecast errors may be largely attributed to the use of horizontal resolution as coarse as 14-30 km with explicit moist processes. Nevertheless, the forecast results suggest that this framework is useful as supporting material for field observations and can be also used for investigation of multi-scale organization of convection in the onset phase of MJO and associated tropical cyclogenesis. Budget analysis of heat and moisture in the MJO onset phase and multi-scale convective events during the CINDY2011/DYNAMO IOP using simulation results is under way. Results of hindcast experiments using the full version of NICAM are also reported in the presentation.