Ensemble Hydrological Forecasting in China: Status and Prospects

Contributed by Qingyun Duan, Beijing Normal University, China

The hydrometeorological services of the western countries, including the United States, Canada, Australia and many European countries, have already fully adopted ensemble forecasting methods operationally in their meteorological forecasts, and development and implementation of ensembles in their hydrological forecasts is well underway. Asian countries, particularly China, have been slow in taking on ensemble forecasting approach in their operational hydrological forecasts. As we enter the Year of the Sheep, I take this opportunity to discuss the status of research and development of ensemble hydrological forecasting in China. I will touch on the reasons why China has been lagging behind in adopting the ensemble approach in hydrological forecasting and speculate on the prospect of hydrological ensemble forecasting methods being implemented operationally in China in the future.

In China, the meteorological sector has kept abreast of its international peers in terms of developing its own operational ensemble meteorological forecast products. When the TIGGE (i.e., THORPEX Interactive Global Grand Ensemble) database was established at the end of 2006 as part of the THORPEX initiative, China Meteorological Administration (CMA) joined other major meteorological forecast centers in the world by providing its global medium-range ensemble forecasts generated by then operational numerical weather prediction (NWP) model – the T213 model. Last year, CMA completed its operational NWP model upgrade from T213 to T639. The upgraded ensemble forecasts have improved the effective predictive lead times from 10 days to 15 days.

Despite the advancement made by Chinese meteorologists, there is no parallel progress in developing ensemble hydrological forecast products in China. Operational hydrological forecasting in China has not made the transition from the traditional approach to the new ensemble forecasting paradigm. The gap between the Chinese meteorological sector and the hydrological counterpart actually goes beyond just the adoption of ensemble forecasting approach. The Chinese operational hydrologists have not made much use of any meteorological forecasts from NWP models in its current operational forecasts. As a matter of fact, operational hydrological forecasts rely mostly on real-time observed precipitation (i.e., the precipitation already on ground). Future precipitation is generally assumed to be zero. Therefore, any forecasting skill of hydrological forecasts is derived mostly from the basin initial conditions used in hydrological modeling and from river routing from upper stream to downstream. The reasons that meteorological forecasts are not used much in China are several-fold:

1. there is the perception that the skill of meteorological forecasts is still limited, though much improved recently. It is not clear to practicing hydrologists how they can make use of meteorological forecasts whose skill level varies both in space and with lead times;
2. meteorological forecasts contain significant biases, especially at basin scale, which must be removed before they can be fed to hydrological models to generate useful hydrological forecasts. Currently, no well established procedures are in place to help hydrologists remove those biases;
3. the space-time scales of most meteorological forecast products are not compatible with those of hydrological forecasts. Downscaling of meteorological forecast products is still needed before they can be applied to hydrological forecasting. But not a single downscaling method is available today that has been proven to work well consistently in China;
4. there is no database of historical hindcasts for meteorological forecast products in China, which are necessary for hydrologists to understand the statistical properties of the meteorological forecasts; and
5. there is still a communication gap on how to let the end users (i.e., the hydrologists and water resources managers) realize that ensemble meteorological forecast products, which provide predictive uncertainty information in addition to forecast itself, can provide added values to them. Therefore, there is an urgent need to demonstrate how ensemble forecasting information can help them in practice.

Despite the hesitance of operational hydrologists in adopting the ensemble forecasting approach, several signs indicate that things will change in China in the near future.

First, the Chinese operational agencies are increasingly realizing that there has been a tremendous advance in skill level of meteorological forecasts in China over the recent years, especially at short-range (5 days) to medium range (up to 2 weeks). This improved skill has strong implications for improving flood forecasting in China, which is of paramount importance to China’s Ministry of Water Resources. The very short useful lead time of current hydrological forecasts hinders the ability of the emergence managers in dealing with disastrous flood hazards that occur every year in many parts of China. Coupling meteorological forecasts and hydrological forecasts is a sure and natural way to extend forecasting lead times and improving flood forecasting skill.

Second, there is a lot of progress in developing new ensemble forecasting methods both inside and outside China. The fact that those new ensemble methods have been adopted successfully in other countries operationally gives strong incentives to the Chinese operational agencies to adopt them in their own practice as well. Many operational agencies are inquiring about the ensemble forecasting methods. The key is to provide this new approach in a robust system that can meet the vigorous demand of operational practice.

In 2012, Beijing Normal University hosted the International Workshop on Hydrological Ensemble Prediction Experiment , sponsored by the Chinese Natural Science Foundation, the Chinese Meteorological Administration, the Hydrological Bureau of the Ministry of Water Resources, and The Three-Gorges Electrical Power Company. This workshop served a very useful purpose in bringing the awareness of the ensemble hydrological forecasting approach to China’s science funding agencies. As a matter of fact, there is a boom in terms of research on ensemble forecasting in China (see an accompanying post on China’s ensemble forecasting research that will appear on this website on Thursday 19th, Chinese New Year).

Further, there are several government-funded projects in China now that support the development of ensemble hydrological forecasting methodologies. For example, Beijing Normal University developed a Hydrological Ensemble Prediction System (BNU-HEPS) with the support of the Chinese Ministry of Science and Technology Pillar Project – Development of Digital Watershed Technology; and CMA supported ‘GIS_based hydrologic forecasting techniques’, which have been tested in the Yalong River basin and the Huai River basin with very promising results (see Figure 1).

I think more education on ensemble hydrological forecasting in China will bear fruits in the near future as more Chinese hydrologists realize the advantages of the ensemble approach. The incoming book to be published by Springer “Handbook of Hydrometeorological Ensemble Forecasting” will be a valuable tool for education.

In keeping with expectations for the Year of the Sheep, which has qualities of kindness, creativity, grace and good fortune , all of the positive developments on ensemble hydrological forecasting signal that China’s operational hydrological forecasting agencies will take on the new ensemble forecasting approach in the near future.