Home > News

An introduction of ensemble forecast of CMA

Source:China Meteorological News Press14-09-2018

One of vital revolutions in modern weather operation is its transformation from deterministic to probability forecast. Ensemble forecast is confirmed as an important way to solve the uncertain problems of single and deterministic forecast. It has been served as a key technology to improve levels of disastrous weather forecast and medium and short-term forecast in advanced centers of countries. In order to make the ensemble forecast play a better role in operation, China meteorological Administration (CMA) established a special team in July, 2012, which has improved the modernization of national operation.


Operational System


The application system of ensemble forecast operation which is independently developed by this team is the first real-time and operation-toward ensemble forecast application platform system. It gathered a few forecast data from both home and abroad operational centers and could help forecasters to acquire forecast information rapidly and effectively through visualized interaction platform. It has been applied in nation-wide provincial and city-level meteorological observatories and received good feedback.

500hPa height field image

200hPa height spread image

Boxplots on temperature of 2m from the ground in the single observatory

Rose diagram on 10m wind field in the single obsevatory.


The application of quantitative precipitation forecast

The ensemble forecast can provide quantitative precipitation forecast (QPF) information with much higher accuracy rate than single and deterministic forecast.


This Sheet shows the grade of QPF is higher than single and deterministic forecast.


The application in extremely disastrous weather forecast

Ensemble forecast could provide  next 7-day extreme weather index forecasts including temperature, precipitation and wind speed supporting forecasters to make early warning for disastrous weather.


Extreme weather index forecasted the extreme warm weather in Northeast China on April 26, 2015.

According to the real weather conditions, the temperature of Harbin, Changchun and Shenyang reached or exceeded 30℃ which all broke history record.


The application in medium and long term forecast

The ensemble forecast gets good performance in medium and long range weather forecast as well. It can extend traditional 1-10 day forecast, providing  forecasters with daily forecast in coming 4 to 15 days and weekly forecast in next 30 days.f For example, in early January of this year, Yunnan Province saw severe precipitation. The ensemble forecast system forecasted the probability of unusually high precipitation anomalies in Yunnan was more than 70%. With the forecast period approaching, the probability reached 95% and the actual precipitation and the forecast result are consistent.


The probability products of weekly cumulative unusually high (low) precipitation anomalies.


The application in severe convective weather probability forecast

CMA developed ensemble probability products including water vapor, instability, dynamic lifting and other physical quantities, to promote the development of severe convective weather probability forecast.


The relative humidity vertical cross-sectional view (left) and single station stratification curve produced.


The application in typhoon forecast

CMA uses typhoon path revised technology of multi-center ensemble model. It effectively reduces track forecast errors, and strongly support typhoon operational forecast. Picture 6 is the path revised forecast map  of the No. 10 typhoon Matmo in 2014. The revised typhoon path is closely to the real path.


The No. 10 typhoon Matmo path revised forecast made by multi-center ensemble model.


The application in marine forecast

It can provide the probability and warning products of offshore winds and sea fog forecasting, to promote the development of refined marine forecast.(May 5, 2015)


The probability forecast of cumulative wind with force scale 7.

925hPa relative humidity and wind ensemble average.


Reporter Dai Kan and Cao Yong

Editor Hao Jing and Shi Long