HYDROASIA 2008 FLOOD ANALYSIS STUDY AT INCHEON GYO CATCHMENT

1. HYDROASIA 2008 FLOOD ANALYSIS STUDY AT INCHEON GYO CATCHMENT TEAM GREEN NGUYEN HOANG HUY SUN YABIN GWON YONGHYEON SUZUKI ATSUNORI LI WENTAO LEE CHANJONG ADVISERS: Prof. LIONG SHIE YUI Prof. TANAKA KENJI

2. OUTLINE • BACKGROUND OF CATCHMENT • MODELING TOOLS • - SOBEK • - MOUSE • SIMULATION RESULTS • FORECASTING: NEURAL NETWORKS • FORECAST RESULTS • CONCLUSION • Q & A

3. INCHEON-GYO WATERSHED

4. Incheon • Located in the mid-west Korea peninsula near Yellow Sea • With both international port and international airport • The third biggest city in Korea • Population : 2,730 thousand

5. Gaja WWTP Pump Station Incheon Gyo Coastline before 1984 Yellow Sea Juan station Gansuk station Study area City Hall Reclamation Area Pump station Study Area Incheon Gyo Incheon-gyo Catchment • Total area : 34 km2 Length :8 km • Tidal difference : 9 m • Avg. of Rainfall : 1,702.3 mm/year • Most of present Incheon Gyo watershed was sea before completed to reclamation in 1985 • Reclamation area used for industry & residence • Culvert slope is very mild(Avg. of Slope : 0.01 %) • Flooding in 1997 to 2001 (except 2000)

6. MODELING TOOLS

7. MOUSE SETUP • Import from the excel file “Imported data to Mouse.xls” to Mouse • Setting up Urban Drainage model with MOUSE • Validation

8. Input Rainfall Data • 4/8/1997 1AM ~ 4/8/1997 4PM (15 hrs) • Maximum rainfall : 19mm/10min 100%

9. Flood(100_100)

10. WATER ON STREET AT NODES (MANHOLES) MANHOLES AT FLOOD AREA

11. SIDE VIEW OF SIMULATION RESULTS

12. SIDE VIEW OF SIMULATION RESULTS

13. SOBEK SET UP

14. WATER ON STREET AT NODES (MANHOLES) NODES NOT AT FLOOD AREA

15. WATER ON STREET AT NODES (MANHOLES) NODES NOT AT FLOOD AREA

16. SIDE VIEW OF SIMULATION RESULTS

17. WATER ON STREET AT NODES (MANHOLES) NODES AT FLOOD AREA

18. WATER ON STREET AT NODES (MANHOLES) NODES AT FLOOD AREA

19. SIDE VIEW OF SIMULATION RESULTS

20. WATER ON STREET AT NODES (MANHOLES) NODES AT FLOOD AREA

21. SIDE VIEW OF SIMULATION RESULTS

22. USING NEURAL NETWORK AS A FORECAST SYSTEM

23. Definition An artificial neural network (ANN) is a mathematic model or computational model based on biological neural networks. ANN consists of an interconnected group of nodes, akin to the vast network of neurons in the human brain.

24. Application • Function approximation • Regression analysis • Pattern recognition • Time series prediction

25. Schematic Diagram

26. Reference Haykin, S. (1999) Neural Networks: A Comprehensive Foundation, Prentice Hall, ISBN 0-13-273350-1

27. THE RESULT OF NEURAL NETWORK

28. WHY A FORECAST SYSTEM IS NEEDED?

29. The Multilayer Perceptron Neural Network is then used to forecast the total discharge at the reservoir. The data series are splitted into 2 portions, one for training while the other for validation Neural Network setup for input and output Dt=30 minutes

31. DISCHARGE S AT RECERVOIR OF THREE MAIN METWORKS (4 August 1997)

32. Correlation coefficient R squared

33. SOBEK SIMULATED VS ANN FORECAST 30 minutes leadtime

34. SOBEK SIMULATED VS ANN FORECAST 60 minutes leadtime

35. Rainfall & Wind Forecasting Catchment Runoff & Sea Level Forecasting Optimal Reservoir Operation SUGGESTIONS Online forecast system

36. Conclusion • MOUSE and SOBEK have been used to study Incheon catchment for the event in 1997. • Several scenarios have been successfully generated by both MOUSE and SOBEK. • Present an idea of using neural network at a forecast system for reservoir operation • An Artificial Neural Network model has been trained by the scenarios generated with sense. • Discharge at the next time step has been reasonably predicted by ANN. • Suggest some solutions to improve the forecast system

37. THANK YOU Q & A

38. Our team movie