Logo IMG
HOME > PAST ISSUE > March-April 2009 > Article Detail


Real-time Flood Forecasting

We’ve learned to predict typhoons. What is required to predict the floods they bring?

Chintu Lai, Ting-Kuei Tsay, Chen-Ho Chien, I-Ling Wu

Typhoon Nari

When HIRDT began our work with the 10th River Bureau, the plan was to develop our project systematically, and ultimately to house it in the Tamsui River Basin Flood Control Command Center, supported by that facility’s infrastructure and computing resources. There we would pursue the usual timeline: Test our ideas, develop input and output subprograms, develop our databases of hydrological data, test the integrated system rigorously, find problem areas and make improvements, then test again. But time and typhoons wait for no hydraulician. Typhoon Nari arrived in northeastern Taiwan on September 14, 2001, and she was a wild one. Nari proved to be a most peculiar typhoon, dumping record amounts of rain and refusing to leave. The winds of Nari were not as violent as those of many typhoons Taiwan had seen before, but Nari hovered over Taipei for an astonishing two full days. When record rainfall was met by the incoming tide, the city of Taipei became swamped within mere hours. The previous local record for rainfall in a single day was 358 millimeters. Nari delivered over 400 millimeters on the first day of her assault.

As torrents rushed into the narrow and winding Keelung River, there was no time to lose. Despite our incomplete preparations, the flood forecast model was put into action. The office of the first author, situated in the Hydrotech Research Institute, National Taiwan University, was converted into an impromptu control center. Tasks were assigned. One team member established contact with stations in the field to acquire real-time data. Another entered those data into a computer. Another channeled the data into our new programs and interpreted the results. Simulation output was converted into computer graphics and animations. Key field locations were monitored for auxiliary data. As fresh simulation data accumulated, boundary conditions were updated and future water-surface profiles were projected. The dawn of the second day of Nari’s visit found the entire team hard at work in the “command center.”

Outside, the Taiwan Journal compared the scene to Venice without the gondolas. The greater Taipei area was experiencing the worst flooding in 50 years.

Figure%207.%20The%20authors%20forecast%20hourly%20flood-stage%20profilesClick to Enlarge ImageIn the command center, the experience was grueling and exhilarating. We applied the model to the whole Taipei basin, with close monitoring of the Keelung River reaches until the floods subsided on September 21, 2001. The result was dramatic: a perfect forecast. Figure 7 shows a forecast of water-surface profiles for each hour over five hours for a section of the Keelung River feeding into the Tamsui River. The model predicted accurately the time and place at which the river torrent would overtop the banks near Great Nan-Hu Bridge and the She-Hou Bridge.

Our report of this prediction helped the river authority issue timely warnings and take other appropriate emergency measures that very likely saved lives, as there were no deaths reported as a result of this overtopping event. And with that we achieved a critical goal of our labor: lives saved and suffering averted by a civil authority armed with flood-forecasting powers.

Our work also affirmed the progress of the disciplinary area to which the crew of the HIRDT had devoted themselves—the application of a deterministic model to unsteady-flow hydrodynamics. The event was a victory of theory and technique: The design algorithm held together as we executed the modeling procedures; the simulation activities themselves proceeded smoothly; and the forecasts were more exact than we could reasonably have hoped.

Following the successful though somewhat frenetic forecast of Flood Nari, the HIRDT team was poised to meet and disarm the next in the steady train of flood-bringing typhoons: Haiyan, Lekima, Sinlaku, Nakri, Melor, Conson, Mindulle, Kompasu, Rananim, Aere, Haima, Nocktan .… For each storm, HIRDT crew was on duty at the Flood Forecast Center of the 10th River Management Bureau, vigorously applying the real-time flood forecast model. The simulation results were reported to civil authorities as regularly timed warnings or intermittent emergency messages when dramatic changes in the rate of flow were predicted, fulfilling our technical goal of real-time, basin-wide, tide-affected, unsteady open-channel flow simulation.


The authors acknowledge valuable support from the Water Resources Agency, Ministry of Economic Affairs, Taiwan, and the research facilities and environment provided by the 10th River Management Bureau and the Hydrotech Research Institute, National Taiwan University.


  • Lai, C. 1986. Numerical modeling of unsteady open-channel flow. In Advances in Hydroscience, Vol. 14, V. T. Chow and B. C. Yen, eds. Orlando, FL: Academic Press. 161–333.
  • Lai, C. 1988. Comprehensive method of characteristics model for flow simulation. Journal of Hydraulic Engineering 114:1074–1097.
  • Lai, C. 1988. Two multimode schemes for flow simulation by the method of characteristics. Proceedings of the Third International Symposium on Refined Flow Modeling and Turbulence Measurement, July 26–28, 1988. Tokyo, Japan, 159–166.
  • Lai, C. 1994. Multicomponent-flow analyses by multimode method of characteristics. Journal of Hydraulic Engineering 120:378–395.
  • Lai, C. 1999. Simulation of unsteady flows in a river system (operation manual with attachment: “Quick Reference Guide”). Hydrotech Research Institute, National Taiwan University, Taiwan.
  • Lai, C., T. K. Tsay, C. H. Chien and I. L. Wu. 2002. Perfect flood forecast—From real-time flow simulation of Flood Nari on the Keelung River. Fifth International Conference on Hydrodynamics, Oct. 31–Nov. 2, 2002. Tainan, Taiwan.
  • Lai, C., et al. 2002. A project for establishing a Keelung River flood forecast model. Research Report W10D90C008, Yen Ching-Ling Industrial Research Institute, National Taiwan University, Taipei, Taiwan. (In Chinese.)
  • Lai, C., et al. 2004. Some extension and renewal works to: Establishing a Keelung River flood forecast model. Research Report W10P92011, Yen Ching-Ling Industrial Research Institute, National Taiwan University, Taipei, Taiwan. (In Chinese)
  • Lai, C., et al. 2007. A real-time flood forecast model for a tide-affected river basin. Hydrotech Research Institute, National Taiwan University, Taipei, Taiwan.

comments powered by Disqus


Of Possible Interest

Engineering: The Story of Two Houses

Letters to the Editors: The Truth about Models

Spotlight: Briefings

Subscribe to American Scientist