UCLA NEWSLETTER (No. 87)
UCLA TROPICAL METEOROLOGY AND CLIMATE NEWSLETTER (No. 87)
SECTION A (September 13, 2009)
by M. Satoh and K. Oouchi
(Received September 7, 2009; Revised September 10, 2009)
NICAM is a global non-hydrostatic model, abbreviation of the Nonhydrostatic ICosahedral Atmospheric Model developed at the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) and Center for Climate System Research (CCSR), the University of Tokyo. The first global cloud-resolving simulation was run using NICAM with mesh size approximately 3.5 km. NICAM simulations show many realistic behaviors of tropical convective systems. It resolves structures of multi-scale convective systems from meso-scale circulations with O(10 km) to large-scale organized cloud systems of a Madden-Julian Oscillation with O(10,000 km). Diurnal cycles of deep convection, intra-seasonal variability, tropical cyclones, and monsoon circulations are also reasonably reproduced. We introduce recent results from the NICAM simulations and summarize the present status and future plans of the NICAM project.
The global cloud resolving model (GCRM) is now here in the world. Under the present state of arts of massive parallel computers, the resolution of the atmospheric global models can be drastically increased to a few kilometers. With those resolutions, deep convective clouds are directly represented without using cumulus parameterizations. We refer such high-resolution numerical models without cumulus parameterization to as Cloud-Resolving Models (CRMs), although we appreciate that numerical convergence of moist models with explicit convection requires much higher resolutions. Global atmospheric modelers had demanded for CRMs covering the global domain for long time, since behaviors of general circulation models (GCMs) crucially depend on specific implementations of cumulus parameterization. NICAM is the first GCRM run with mesh size approximately 3.5 km over the globe without using cumulus parameterization.
The development of NICAM started since the year 2000 at Frontier Research System for Global Change (FRSGC) of JAMSTEC under the strong leadership of the director Dr. Taroh Matsuno (Matsuno et al. 2009; FRSGC is reorganized into FRCGC in 2004 and then to Research Institute for Global Change (RIGC) in 2009.) NICAM is characterized by icosahedral geodesic grids and the non-hydrostatic systems (Satoh et al., 2008). NICAM has been coded from scratch as a new atmospheric global model, because it has a different grid system and is based on different governing equations from those used in current general circulation models: spectral or latitude-longitude grids with hydrostatic equations. The first global cloud-resolving simulation with mesh size 3.5 km is performed in 2004 under the aqua planet condition (Tomita et al., 2005). It shows systematic eastward propagation of super cloud-cluster like signals with multi-scale structure of convective systems embedded in the large-scale organized convective systems. Following this encouraging result, a hindcast simulation of a Madden-Julian Oscillation (MJO) event is performed and it revealed roles of multi-scale convective systems on the MJO (Miura et al., 2007). In the following sections, we further describe recent results from the NICAM simulations and future plans of the NICAM project.
We generally use combinations of different mesh sizes with about 3.5 km, 7 km, and 14 km. Generally, nonhydrostatic models are run with mesh size less than approximately 5 km, and it further requires mesh size less than 1 km to resolve convective cells. Probably, one may think that only 3.5km-mesh experiments are categorized as GCRM experiments, or they may be called global cloud-system resolving or cloud-permitting, if they like. However, interestingly, we found many similarities between experiments with these different resolutions for behaviors of large-scale organization of convective systems. Based on these experiences, we supplementary use the coarse mesh model about 7 km or 14 km for conducting sensitivity experiments, ensemble experiments, and also longer integrations. We of course note that systematic biases are found for experiments with coarser resolutions: stronger precipitation (Tomita et al., 2005; Iga et al., 2007; Oouchi et al., 2009b), phase lag of diurnal cycle of deep convection (Tomita et al., 2005; Sato et al., 2009), and faster propagation speed of the large-scale organized convective systems (Tomita et al., 2005; Liu et al., 2009). Cloud properties become better and closer to observations as resolution increases (Inoue et al., 2008). Even though, experiments with the mesh size about 7 km and 14 km are computationally efficient, and we think that the use of these resolutions is acceptable for specific purposes of studies, as described below in this article. At this moment, the longest simulation with NICAM is for five months at the mesh interval about 14 km, and for three months at the mesh interval about 7 km. For the highest resolution, we performed one week simulation with the mesh interval about 3.5 km.
As mentioned in Introduction, the performance of NICAM as a GCRM was initially assessed using a series of aqua-planet experiments (APE, Kusatsu Issue of UCLA TROPICAL METEOROLOGY NEWSLETTER, 2007). The encouraging results regarding the tropical convections and disturbances in the APE run stress the importance of cloud-resolving framework for studying tropical convective systems and pursuing its potential in forecast of convective disturbances such as tropical cyclones and MJOs. The APE run was an idealistic run of GCRM and serves as a useful benchmark against which the following experiments with the inclusion of real-world components (sea surface temperature (SST), topography, vegetation) were evaluated (called "real-world run"). Summarized in the following are the highlights of the results from the two sets of real-world runs that have shown promising potential of NICAM in simulating various tropical cloud systems, large-scale disturbances that are interactive with the ensemble of the cloud systems. In a boreal summer run, time range was extended to seasonal length (June-October), thereby expanding the scope of target phenomena from diurnal through intra-seasonal range. Because of the limited space of the letter article, it is not possible to cover all the aspects of the simulated results. For those interested in details, please refer to the original articles listed in the NICAM-hiki (http://nicam.jp/ ; See 3-3).
The 2006 MJO experiment was the first boreal winter run with the topography, land process, and specified sea surface temperature to investigate whether NICAM is capable of simulating an MJO event (Miura et al., 2007). MJO is one of the important moist modes as well as a benchmark phenomenon in the modeling of tropical atmosphere. Its successful simulation had been a challenge in the conventional GCMs.
The 2006 experiments were conducted for one-month period (15 December 2006 - 15 January 2007) at the longest duration, with the mesh sizes approximately 7 km, and 14 km. The highest resolution experiment with the mesh size approximately 3.5 km was conducted for a week (25 Dec. - 1 Jan.). The results revealed that NICAM simulates the MJO with phase velocity as observed (about 5 m/s) and internal cloud structures. The timing of jump over the New Guinea Island of the Maritime continent was also simulated very well; this jump of MJO was found to be contributed by westward-propagating disturbances that are ubiquitous in the tropical Pacific. The spontaneous organization of eastward-propagating MJO is hypothesized to be prompted by the westward-propagating disturbance that helps transport moisture necessary for buildup of MJO-associated convections. The disturbances are likely to include equatorial Rossby waves as a major component (Miura et al. 2007).
In addition to the MJO, the geneses and evolution of the tropical cyclones associated with the MJO were successfully simulated over the Indian Ocean (Fudeyasu et al., 2008). The tropical cyclogenesis was simulated at almost the same location and the timing as the observation even the experiment was initiated at two weeks before the cyclogenesis. Remarkably, the internal and the vertical cloud structures of the tropical cyclones were simulated similar to those observed by the TRMM (Tropical Rainfall Measuring Mission). The key factors in the success are suitable representations of large-scale circulation associated with the MJO (the westerly wind burst) and of interactions between the large scale flows and mesoscale ensembles of convection. The result suggests that the cloud-resolving framework provides a promising tool for understanding and forecasting tropical cyclogenesis and its association with MJO.
The convective systems accompanied with MJO are in various forms and characteristics. Nasuno et al. (2009) discussed an interesting multi-scale hierarchical structure of convections associated with the MJO discussed by Miura et al. (2007). The hierarchy includes synoptic-scale, convectively-coupled eastward-propagating and westward-propagating disturbances as frequently detected in observations (Dunkerton and Crum, 1995). Some westward-propagating signals show convective structure and the accompanied circulation fields resembling those in the westward-propagating disturbances of tropical-depression-type and Mixed-Rossby-Wave type (Takayabu and Nitta, 1993). .
To understand the mechanisms of the hierarchical structure is relevant to identify the origin of stochastic or systematic nature of the multi-scale tropical convective systems (Wheeler et al., 2000) and their roles in the development of MJO. Combined with the high-resolution TRMM and other satellite data, use of the NICAM outputs for investigating these convective systems will help solve the long-standing problems of the mechanism of the maintenance of tropical convective disturbances including MJO and their relationships with moist equatorial waves (Yamasaki, 1969; Hayashi, 1971).
NICAM successfully simulates the propagation of diurnal cycle of precipitation associated with land-sea breeze in both 14-km and 7-km runs (Sato et al. 2009). Over continental area, the 3.5-km run produced the peak time and amplitude very similar to those in TRMM PR (Precipitation Radar) observations. The peak time of the cycle in the 14-km run is 3 hours later than the 7-km run. This sensitivity of the phase cycle to the horizontal resolution was attributed to the differences in structure and life cycle of convective systems. The improvement of the bias in higher resolution suggests that suitable representation of the interactions between mesoscale convection and environmental circulations including gust front is important to simulate the evolution of diurnal precipitation.
A series of case study approaches taken in the APE and the 2006-MJO runs showed encouraging performance of NICAM in producing tropical cloud systems and moist atmosphere. The next plan was to perform longer-time integration. In particular, seasonal-long simulation was essential to expand the scope of study and consolidate the basis of seasonal-range forecast and study cloud systems using GCRM. The first seasonal-long run was performed with NICAM targeting at studying Indo-Asia monsoon, MJO cycles, and boreal summer mean atmospheric states. From a numerical weather prediction viewpoint, boreal summer is a challenging season to forecast. MJOs are clearer in boreal winters and boreal winter MJOs could be simulated by NICAM as shown above. Therefore, this experiment helps identify strengths and weaknesses of the current skill of NICAM when put together with the boreal winter MJO simulation. The time integration was performed for June to August (October) of 2004 for 7-km mesh (14-km mesh), with the prescribed SST from NOAA Reynolds weekly SST dataset.
The seasonal run simulated successfully the multi-scale precipitation features and accompanied circulation in the mature Asian monsoon season. Comparison of the Indian monsoon index (Wang et al. 2004) between the model-derived one and the observation revealed a fair level of skill of the 7-km mesh NICAM in predicting the activity of the monsoon trough up to 30-40 days after the time integration starts (Oouchi et al. 2009b). This exceeds the standard predictability range of atmospheric disturbances, and therefore suggests an encouraging skill of NICAM for monsoon research.
Evolution and mean precipitation features are also reproduced well in the 7-km mesh run. For example, the simulated precipitation systems over the mountain ranges in Myanmar exhibit interesting hierarchy consisting of diurnal, and synoptic scale elements that propagate in the direction controlled by intra-seasonal synoptic-scale circulation features (Oouchi et al. 2009b). In some cases, synoptic-scale feature is enhanced as activation of the embedded diurnal cycle that is prompted by orographically induced uplift as some observations indicate (Xie et al. 2006).
Notwithstanding such good skill of NICAM in producing multi-scale monsoon features, there remains excessive precipitation bias over the Indian Ocean compared to observed mean boreal summer dataset (Oouchi et al. 2009b). The bias may come from the prescribed way of sea surface temperature as the bottom boundary instead of setting interactive ocean-atmosphere interface. It has been disputable whether monsoon circulation and precipitation is driven by ocean-atmosphere coupling processes or essentially of atmosphere origin. The current results suggest the possibility of the former, although it remains to be understood if the bias might be improved by tuning or changing physical schemes in the model.
One of the important upgrades in physical schemes from the 2006-MJO run (ref. 3-1) is an improvement of subgrid-scale condensation process in a turbulent closure model (Noda et al. 2009) that was based on level 2 of the Nakanishi and Niino (2006) scheme. The upgrade led to an improvement in the spatial distribution of low-level clouds, including the clouds in the offshore region, which were comparable to observations in the 14-km mesh run (Noda et al. 2009). The mean precipitation amounts of low-level clouds were also close to observations. Quantitatively, a systematic bias was found such that low-cloud amounts are excessive, and mid-cloud amounts are low compared to the ISCCP data. The bias may be associated with over-prediction of moisture in the lower atmosphere that can be ascribed partly to lack of turbulent transport process in the model. The process should be prompted by shallow cumulus convections in the real atmosphere. Noda et al. (2009) covers these topics in considerable details.
Relationship between MJO and tropical cyclogenesis is an important subject in the boreal summer run. This is a subject conventional climate models have not adequately addressed because they have generally failed to reproduce MJO of enough strength. MJO is a good measure for evaluating a performance of clouds representation of model. The NICAM run simulated successfully a genesis of TC0407 that developed in the northwestern Pacific about 20 days after the start of the time integration, and revealed that the genesis originates from a seeding disturbance emanating from the simulated planetary-scale eastward-propagating signal resembling MJO (Oouchi et al. 2009a).
In association with the description in Section 2, readers might be interested in how much the resolution should be fine for suitably handling tropical cyclogenesis in GCRM. By design, 14-km mesh cannot resolve deep clouds, but ensembles of clouds such as cloud clusters are represented. Tropical cyclones involve wide range of hierarchy of clouds from synoptic-scale circulation down to meso-scale and individual deep convection. As far as we focus on the cloud features and the mechanisms that are associated with MJO-scale effects, the 14-km mesh run is useful because it represents cloud clusters that are the type of convection most controlled by synoptic-scale environmental changes provided by MJO. In fact, key features of the typhoon genesis preconditioned by MJO in a boreal summer period are captured well in the 14-km mesh setup (Oouchi et al. 2009a). The 14-km mesh experiments are therefore a practical choice to discuss genesis process at this scale of tropical cyclone hierarchy. Yet, when we focus on the hierarchy of genesis process further down to the mesoscale level and discuss its upscale organization, we need much higher resolutions (Fudeyasu et al. 2008). Choice of grid spacing in NICAM simulation should therefore depend on what scientific question or target phenomenon one intends to investigate. Even simply dictated as "cloud-resolving model," it can actually imply much diverse meanings depending on the specific aim of the research.
Our research activity is reported in up-to-date and condensed form in our web site at http://nicam.jp . The site started its operations as of Nov. 2008. All the contents on the website are operated by Hiki (http://hikiwiki.org/en/about.html ) which is a powerful and fast wiki clone written by Ruby language. NICAM developers and users (including data users) are able to exchange information on the website; Also available there are the paper lists, seminar informations/presentations for data users (local page), the page for experimental monitoring MJO by MJO index developed and maintained by Dr. Taniguchi (See "About NICAM page" for details) and other materials for research collaborations. The web site will be developed to be in more flexible and user-friendly format for future wide-ranging purposes. For anyone having feedback on the NICAM-hiki, please feel free to contact the web administrator, Dr. Taniguchi, at taniro'at'jamstec.go.jp.
Although we have a list of future developments and experiments of NICAM described below, GCRM is quite new and we can learn a lot of things from the high-resolution experimental data. The GCRM data can be thought of as a kind of fully observed nature with time sequence of three-dimensional fields. If some atmospheric phenomena are reproduced in the data as in observations, we can analyze detailed structures and obtain quantitative results. From comparison with the GCRM data and corresponding GCM data, we could deepen our understandings of subgrid behaviors in GCMs, such as momentum transports associated with deep convection. In addition, we will understand mechanisms behind those phenomena through sensitivity experiments based on working hypothesis, although it is not always straightforward. The mechanisms of propagation and onset of MJOs are examples partly discovered by the NICAM experiments (Miura et al., 2007, 2009).
If further computational resources are available, we can use NICAM as climate or numerical weather forecasting simulations. Fortunately, we have a plan of the next-generation super computer in Kobe, Japan, which will be in use around 2014 (http://www.nsc.riken.jp/index-eng.html ). Using such super computers, one future direction is to increase resolution of NICAM with mesh size less than 1 km to more realistically resolve cloud systems. As another direction, we might perform much longer integrations for multi-years. The AMIP-type experiments for a few decadal integrations or time-slice runs under future global warming conditions will be possible. Development of both dynamics and physics schemes of NICAM also continues. The list of future development includes implementation of more sophisticated cloud microphysics schemes with aerosols transports, coupling to high-resolution ocean models, and the assimilation system using high-resolution satellite data.
Even without massive super computers, NICAM is flexible and can be used for multi-purposes. NICAM can run as a coarse resolution general circulation model by using subgrid cloud parameterizations. It also can be used as a regional model by transforming grids to focus a specific area (stretch-NICAM; Tomita 2008). These models are especially useful for testing and improving schemes. Some examples are shown at http://www.ccsr.u-tokyo.ac.jp/~satoh/nicam/nwppub/index.html . NICAM is open for people who want to collaborate, and please contact us if you are interested in our activities.
The authors appreciate Taroh Matsuno for motivating and launching the global-cloud-resolving model project, and continued supports during the model development and research activities. Thanks are also extended to Dr. Tomita and NICAM team members for model development and scientific discussions. All the experiments described in this article have been performed on the Earth Simulator, Japan Agency for Marine-Earth Science and Technology (JAMSTEC), under the support by the Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Agency, and the Innovative Program of Climate Change Projection for the 21st century (KAKUSHIN) project funded by Ministry of Education, Culture, Sports, Science and Technology.
- Dunkerton, T.J. and F.X. Crum, 1995: Eastward propagating 2- to 15-day equatorial convection and its relation to the tropical intraseasonal oscillation. J. Geophys. Res., 100, 25,781-25,790.
- Fudeyasu, H., Y. Wang, M. Satoh, T. Nasuno, H. Miura, and W. Yanase, 2008: The global cloud-system-resolving model NICAM successfully simulated the lifecycles of two real tropical cyclones. Geophys. Res. Lett., 35, L22808, doi:10.1029/2008GL0360033.
- Hayashi, Y. 1971: Large-scale equatorial waves destabilized by convective heating in the presence of surface friction. J. Meteor. Soc. Japan, 49, 458-466.
- Iga, S., H. Tomita, Y. Tsushima, M. Satoh, 2007: Climatology of a nonhydrostatic global model with explicit cloud processes. Geophys. Res. Lett., 34, L22814, doi:10.1029/2007GL031048.
- Inoue, T., M. Satoh, H. Miura, B. Mapes, 2008: Characteristics of cloud size of deep convection simulated by a global cloud resolving model. J. Meteor. Soc. Japan, 86A, 1-15.
- Liu, P., Satoh, M., Wang, B., Fudeyasu, H., Nasuno, T., Li, T., Miura, H., Taniguchi, H., Masunaga, H., Fu, X., Annamalai, H., 2009: An MJO Simulated by the NICAM at 14-km and 7-km Resolutions. Mon. Wea. Rev., accepted. DOI: 10.1175/2009MWR2965.1.
- Matsuno, T., Satoh, M., Tomita, H., Nasuno, T., Iga, S., Miura, H., Noda, A. T., Oouchi, K., Sato, T., Fudeyasu, H., Yanase, W., 2009: Cloud-cluster-resolving global atmosphere modeling - A challenge for the new age of tropical meteorology. The WMO 4th International Workshop on Monsoons (IWM-IV), Beijing in October 2008, Lecture Note, in press.
- Miura,H., Satoh, M., Nasuno, T., Noda, A.T., Oouchi, K., 2007: A Madden-Julian Oscillation event realistically simulated by a global cloud-resolving model. Science, 318, 1763-1765.
- Miura, H., Satoh, M., Katsumata, M., 2009: Spontaneous onset of a Madden-Julian oscillation event in a cloud-system-resolving simulation. Geophys. Res. Lett., 36, L13802, doi:10.1029/2009GL039056.
- Nasuno,T., H. Miura, H., M. Satoh, A. T. Noda, and K. Oouchi, 2009: Multi-scale organization of convection in a global numerical simulation of the December 2006 MJO event using explicit moist processes. J. Meteor. Soc. Japan, 87, 335-345.
- Noda, A. T., K. Oouchi, M Satoh, H. Tomita, S. Iga, and Y. Tsushima, 2009: Importance of the subgrid-scale turbulent moist process: Cloud distribution in global cloud-resolving simulations. Atmos. Res., in press.
- Oouchi, K., A.T. Noda, M. Satoh, H. Miura, H. Tomita. T. Nasuno, S. Iga, 2009: A simulated preconditioning of typhoon genesis controlled by a boreal summer Madden-Julian Oscillation event in a global cloud-system-resolving model. SOLA, Vol. 5, 065-068, doi:10.2151/sola.2009-017.
- Oouchi, K.,A. T. Noda, M. Satoh, B. Wang, S.-P. Xie, H.G. Takahashi, T. Yasunari, 2009: Asian summer monsoon simulated by a global cloud-system-resolving model: Diurnal to intra-seasonal variability. Geophys. Res. Lett.,36, L11815, doi:10.1029/2009GL038271.
- Sato, T. Miura, H., Satoh, M., Takayabu, Y.N., Wang, Y., 2009: Diurnal cycle of precipitation in the tropics simulated in a global cloud-resolving model. J. Clim., accepted; DOI:10.1175/2009JCLI2890.1.
- Satoh, M., T. Matsuno,T., H. Tomita, H. Miura, T. Nasuno, S. Iga, 2008: Nonhydrostatic Icosahedral Atmospheric Model (NICAM) for global cloud resolving simulations. J. Comp. Phys., 227, 3486-3514.
- Takayabu, Y.N. and Ts. Nitta, 1993; 3-5 day period disturbances coupled with convection over the tropical Pacific Ocean., J. Meteor. Soc.Japan, 71, 221-246.
- Tomita, H., 2008: A stretched grid on a sphere by new grid transformation. J. Meteor. Soc. Japan, 86A, 107-119.
- Tomita, H., Miura, H, Iga, S. Nasuno, T., and Satoh, M., 2005: A global cloud-resolving simulation: preliminary results from an aqua planet experiment. Geophys. Res. Lett., vol.32, L08805, doi:10.1029/2005GL022459.
- Wang, B., I.-S. Kang, and J.-Y. Lee (2004), Ensemble simulations of Asian- Australian monsoon variability by 11 AGCMs, J. Clim., 17, 803- 818.
- Wheeler, M., G. N. Kiladis, and P. J. Webster, 2000: Large-scale dynamical fields associated with convectivelycoupled equatorial waves. J. Atmos. Sci., 57, 613-640.
- Xie, S.-P., H. Xu, N. H. Saji, Y. Wang, and W. T. Liu (2006), Role of narrow mountains in large-scale organization of Asian monsoon convection, J. Clim., 19, 3420- 3429.
- Yamasaki, M., 1969: Large-scale disturbances in a conditionally unstable atmosphere in low latitudes. Pap. Meteor. Geophys., 20, 289-336.