Challenges in operational nowcasting of Beijing
---- from forecaster’s perspective
Jianjie Wang
Beijing Meteorological Bureau
Aug. 2011


There are eager demands for fine 0-6h services on severe convective weathers (torrential rain, thunderstorm, hail, gust wind etc) from the public and governmental agencies in Beijing, because of the high impacts of these weather events to the city. However, precise severe weather prediction in 0-6h is one of the most challenging issues today in weather service of Beijing Meteorological Bureau (BMB here after).

In this paper, the operational use and potential of existing objective techniques (i.e., nowcasting system BJ-ANC, and mesoscale rapid updating cycle model system BJ-RUC with horizontal resolution of 3km and 3h in cycle) for nowcast and very-short-term forecast on the torrential rainin Beijing area are investigated, and the limitations of the two objective systems and challenges in nowcasting operation are discussed as well, by use of a number of real time cases.

The preliminary results show that, the performance of the BJ-RUC system on very-short-range (0-12h) quantitative precipitation forecast (QPF) is influenced by the strength of the large scale weather forcing, i.e., BJRUC performed better in the strong than in the weak large scale weather forcing conditions in general:

  • Under the strong large scale weather forcing (e.g., westerly trough or low circulation system in 500hPa and 700hPa moving toward Beijing from Hetao or middle inner Mongolia regions), the meso-b spatial and temporal features of 3h accumulated precipitation over Beijing area could be captured reasonably well in very-short-term, and QPF products valid at the same clock from cycled runs within 24h period provide generally consistent information with small changes on the forecast of raining time, rainfall location, strength and movement etc;
  • Under the weak large scale weather forcing (e.g., Beijing located in the foreside of the ridge with westerly or north-westerly wind in 500hPa and 700hPa levels), in which short-lived torrential rainfall events in Beijing area occur mostly, there are visible bias generally on BJRUC’s very-short-term QPF products. And the information of QPF products from cycled runs valid at the same clock tends to diverse run to run. Encouragingly, BJRUC could perform quite well in certain runs in predicting torrential rain induced by short-lived convections within very-short-term, when signals of the local convective potential have occurred in model initial time. These imply that BJRUC QPF is more sensitive to model initial conditions under the weak large scale weather forcing, and that BJRUC system has the potential in predicting the short-lived convective torrential rain of Beijing area in very short range by the proper ingestion of local convective potential signals into the model initial conditions.

It was revealed that BJANC 0-1h nowcasting products could provide quite precise and useful information in term of the track, trend, movement and quantitative precipitation of convective cells for BMB’s nowcasting operation, especially for those short-lived severe convections which are more challenging to forecasters in reality. In addition, the nature of quick update (every 6 min) on nowcast products of BJANC makes a fast adjustment to its nowcast bias, therefore, reduces the negative impact on operational use of the products. However, as many other nowcast systems, BJANC’s nowcasting products beyond 2h (up to 6h) have less skill, specifically under the weak large scale weather forcing. And the limitations of the BJANC system could be seen clearly in prediction and identification of the convective cell initiation, and in cell-evolution due to local effects (terrain or land use of Beijing area).

The main gaps in operational nowcasting identified from the investigation are, (1) how to meet the fine requirements to the severe weather warning (in terms of the specification on the severe weather type, strength, location and the time of day) and to the early issuing of the warning (several hours ahead the weather events) with limited capabilities on severe weather nowcasting and very-short-term forecasting techniques; (2) how to take advantages of existing objective forecast systems, such as BJANC and BJRUC, in operational forecasting under different weather conditions by identifying the uncertainties and limitations of their guidance products of 0-12h especially 0-6h ranges, and by development of new guidance products about weather type identification (e.g, hail and lightning ect); (3) how to develop the “hybrid approach” to improve the accuracy of 0-6h severe weather forecasting in operation, namely, to combine forecasters’ knowledge and experiences on local effects into objective forecasting techniques (or with guidance products) to mend some of limitations in the nowcasting system (or to enlarge the usability of meso scale RUC products); (4) how to build up the capabilities of forecasters in using new techniques (e.g., VDRAS and observations from different remote sensing equipments) for nowcasting.

Thanks to my colleagues, Mingxuan Chen, Chenyun Sun, and Jisong Sun etc., for their support to this investigation.