Home > News

GRAPES_GFSV2.0 development and application

Source:China Meteorological News Press21-10-2016

On June 1, GRAPES_GFS V2.0 (hereinafter referred to as GRAPES), the new generation of Global/Regional Assimilation and Prediction Enhanced System which is independently developped by China Meteorological Administration (CMA), has been put into formal operation. Since that day, all provincial, regional and municipal meteorological services can receive products of GRAPES_GFS V2.0 through CMACast system and provide utilization for users by CIMISS.

Researchers engaged in this work have been dedicated to independent research and development for one decade. They have made consistent breakthroughsand promoted the sustained development of GRAPES.

GRAPES Photo by Guan Chenggong

Research and Development

Independent research and development is not “work behind closed doors”. We need to improve our model and adopt new thinking and approaches in global modelling cloud physics program, and satellite data assimilation via communication with other parties.

Liu Qijun, researcher of Numerical Weather Forecast Office, responsible for the research and development of cloud physics program, said, “A few year ago, GRAPES used foreignmesoscalemodels cloud plan like WRF. It found that the forecast oflow-layer clouds, especially in tropical low-latitude regions and gridded precipitation is less obviously and huge error.”

He said the research team found out the defects of GRAPES cloud scheme in theoretical design after analyzing theoretical framework and computing method of advanced global modelling cloud scheme in ECMWF and UK Met Office. Because the rough resolution of the global pattern, the computing method of water vapor condensation into cloud directly using mesoscale model cloud scheme will not factor in the cloud. We do not consider that rough resolution global model is dynamic and convective in tropical regions and the direct effects of subgrid convective rollout process on gridded scale cloud.

Therefore, the research team conducted research and development of new mixed-phased cloud scheme and activated high computing accuracy and complex double parameter cloud physics plan. Right now, this cloud physics plan has been coupled with the newest GRAPES version and becomes the operational modelling plan, which markedly increased cloud forecast performance, address the issue of water content of cloud, cloud content and precipitation forecast in tropical low-latitude regions, prominently improve low cloud prediction, and rectify the phenomenon that large tropical cloud prediction deviation resulting in huge temperature deviation. At the same time, it also enhanced the prediction performance of GRAPES in situation fields.

Progress of GRAPES  

There is no previous experience to draw on in terms of independent research and development as well as GRAPES data assimilation. After nearly 10-year efforts, the information application level of GRAPES is much higher than that of T639,with satellite data accounting for nearly 80% for the first time. This hasprovidedeffective supplement for prediction in ocean and desert lacking in regular data.

In order to apply satellite data more rationally and effectively, Han Wei, deputy director of modelling research and development office, developed satellite data deviation correction method, which markedly improved the application level of satellite data. Liu Yongzhu, Doctor of modelling research and development office, took one year to recompile tangent model, accompanying model, and several ten thousand lines of source code needed in GRAPESfour-dimensional variation, which improves the calculation precision and efficiency, and lays a solid foundation for the future development of the four-dimensional variation.

Application

Application products possess rich and high refined features which could provide better reference for forecasters. Meanwhile, GRAPES plays good role in high temperature forecast during flood season, more instructive than EC and T639.
Chen Jing, director of an operational office of Numerical Forecast Center said GRAPES products are distributed to meteorological data sharing servers of every province through CMACAST. They provided reference for the national forecasters.

GRAPES_Meso application

The National Meteorological Center, the Observatories of Liaoning, Fujian, Jiangsu, Sichuan and Gansu conduct GRAPES forecast products testing every day. Li Deqin, a vice director of Shenyang Observatory told that they added comparison and test analysis contents in daily products including products of precipitation test, upper-air system, surface temperature, vertical section and upper-air physical quantity.

“National Meteorological Center and provincial meteorological observatories have successfully made GRAPES products data access to platforms of MICAPS3 and MICAPS4, produced complex chart and configure quick menu to help forecasters to obtain products as soon as possible”Dai Kan, a director of ensemble forecast application team of National meteorological Center said.

Fujian meteorological observatory utilize EWIPS platform to realize rapid display about upper-air situation fields, temperature fields and water vapor fields of MICAPS format.

The best GRAPES application is high temperature forecast. It’s better and more accurate than EC and T639 model, especially in southern parts of China. According to inspection results of Fujian meteorological observatory in June and July, GRAPES model has excellent highest temperature station forecast capacity, higher forecast accuracy than ECMWF, T639 and NCEP-GFS model. In addition, comparing to ECMWF model, it’s better to use GRAPES to forecast the scale 8 gale of typhoon Nepartak.

GRAPES can meet the needs of the mesoscale weather forecast and grasp the middle and long-term changes of atmospheric long-wave adjustment. The 500hPa geopotential height field weather forecast system is stable, such as subtropical high, Northeast vortex, Wu Mountain ridge and other large-scale system. The weather system forecast is better in short term and can reflect the actual physical process more accurately. The forecasting stability of rainfall, temperature and wind speed is more obvious than T639 and other models.

The real-time system of GRAPES_MESO 3km for tornado of Funing, Jiangsu province.

Sun Jian, a director of Model Research Office of Numerical Forecast Center said the research and development of GRAPES, in a way, has promoted the clarification of observation data process, replacement of high-performance computer and communication with other departments. There are obvious improvement in quantitative application of FY satellite data, wind profile radar data and surface GPS data. Meanwhile, high-resolution model of GRAPES also plays important role in improving forecast level of severe convective weather forecast.  

Chinese version:http://www.cma.gov.cn/2011xzt/2016zt/20160506/201609/t20160907_321324.html

 

Editor Hao Jing and Liu Shuqiao