The 4th GEOSS AWCI ICG Meeting Kyoto, 6-7 February 29 WEB-DHM and IWRM Lei Wang a, Toshio Koike a, Cho Thanda Nyunt a, Oliver Cristian Saavedra Valeriano a, Tran Van Sap b, Tsugito Nagano c a Department of Civil Engineering, the University of Tokyo, Japan b National Hydro-Meteorological Service, Ministry of Natural Resources and Environment, Vietnam c Earth Observation Research Center, JAXA, Japan wang@hydra.t.u-tokyo.ac.jp 1
Research Strategy LDAS-UT Atmosphere AMSR-E SiB2 Land Surface WEB-DHM Support Water Resources Management River GBHM LDAS-UT: land data assimilation system developed at Univ of Tokyo; GBHM: geomorphology-based hydrological model. 2
WEB-DHM (Water and Energy Budget-based Distributed Hydrological Model) Subgrid Parameterization Wang, Koike, et al, JGR, 28, In press 3
Little Washita Basin, USA Area: 621 km 2 Hourly simulation with 5 m grid size 4
Latent Heat Flux Net Radiation Sensible Heat Flux Ground Heat Flux CO2 Flux 9 7 5 3 1-1 6-3 5 4 3 2 1-1 4 3 2 1-1 -2 3 2 1 NOAA flux site Net Radiation Latent Heat Flux Sensible Heat Flux -1 5-2 Observed Simulated 3 12 24 36 48 1-1 -3 Model Evaluations with SGP97&SGP99 Observations Ground Heat Flux CO2 flux: µ mol m -2 s -1 Obs Hour (Jun27~Jul17) -5 12 24 36 48 Hour (Jun27~Jul17, 1997) Sim Precipitation (mm/hour) Surface Soil Moisture 8 6 4 2.35.25.15.5 12 24 36 48 Hour (Jun27 - Jul17) Surface soil moisture Obs Sim Precipitation Surface Soil Moisture Obs Discharge (m 3 /s) Discharge (m 3 /s) 8 7 6 5 4 3 2 1 Obs Sim 1997.6.27 1997.7.2 1997.7.7 1997.7.12 1997.7.17 1997.7.22 Date 6 Discharge 5 4 3 2 1 Sim Basin-scale Discharge Calibration BIAS = -.6% Nash =.956 Validation Obs Sim 1998.9.1 1998.12.1 1999.3.1 1999.6.1 Date 5 1 2 3 4 5 6 7 8 Precipitation (mm/hour) 4 8 12 16 2 24 Precipitation (mm/day)
The upper Tone River Basin, Japan Simulation with 5 m grid and hourly time step. 6
Calibration and validation with discharges at main stream gauges 7
21 Annual Largest Flood Peaks (1) 21 (2) 22 23 24 (a) Murakami (b) Yakatabara (c) Iwamoto (d) Maebashi 8
Comparison of 8-daily LSTs averaged for the upper Tone River Basin (21 to 24) Daytime (around 1:3, local time); Nighttime (around 22:3, local time) 9
Characteristics of WEB-DHM Continuously simulating the exchanges of fluxes (water, energy and CO2) in the SVAT system at the basin-scale in a spatiallydistributed manner. Physical description of ET process. Satellite data is used to describe the vegetation state and phenology. (LAI & FPAR) Applicability to large river basins. 1
IWRM: Integrated Water Resources Management IWRM is a systematic process for sustainable development, allocation and monitoring of water resource use in the context of social, economic and environmental objectives. Implementation of integrated flood management together with structural and non structural measures, coordination of residents and other stakeholders and capacity development for water related problems are some ways of IWRM. Therefore, the regional IWRM can be actually applied in the real world by demonstrating the improvements of existing water management practices. 11
WEB-DHM coupled with SCE for improved reservoir operation Reservoir routing Flowchart for the integrated system 12
The Red River Basin 13
Model evaluation with the streamflows at the Da sub-basin Discharge (m 3 /s) Discharge (m 3 /s) 2 16 12 8 4 2 16 12 8 4 Inflow to Hoa Binh Reservoir Sim Obs 6.1 6.11 6.21 7.1 7.11 7.21 26 Inflow to the Hoa Binh Reservoir Sim Obs 6.1 6.11 6.21 7.1 7.11 7.21 25 Calibration in 26 Discharge (m 3 /s) 12 8 4 Muong Te Validation in 25 Discharge (m 3 /s) 12 8 4 6.1 6.11 6.21 7.1 7.11 7.21 26 Muong Te 6.1 6.11 6.21 7.1 7.11 7.21 25 14
Evaluation of reservoir routing at the Hoa Binh reservoir (26) (a) The volume and water level (V-H) relation curve of the Hoa Binh reservoir. (b) Simulated water level of the reservoir with the observed release. 15
Multi-objective Optimization Multi-objective optimization refers to problems including several objectives to be fulfilled simultaneously. These objectives have conflicts with each other and they are in different unit measures. (1) Aggregation approach: the priorities of the objectives are established beforehand; It combines the different objectives into one aggregated scalar to be optimized. (2) Pareto domination approaches: no preference information is considered or is available before the search. It is not based on a single comparative value but on whether one solution is dominated by another. 16
Aggregation approach for multiobjective optimization This method transforms the various objective functions into a single scalar objective function. The resulted optimal solution is based on the weights assigned to the objectives. The aggregated objective function is given as follows: MinimiseF Parameter set Weight N ( X ) = w g ( f ( X )) i= 1 i i Individual objective function i Transformation function 17
Transformation function is determined for the same influence on the aggregate objective function by each objective function because they have difference in magnitudes of different measures. In this study, the transformation to a common distance scale is applied so that all the objective functions will get about the same distance to the aggregated objective function near the optimum. The formulation of the transformation function is given by: = g i ( f ( X )) i = f i ( X ) σ i + ε Transformation constant: ε i max min f j σ i ( X ) f ( X ) j, j = σ i 1, 2,... is the standard deviation of the ith objective function. N min i σ i 18
Objective Function for the Hoa Binh reservoir operation This study is to deal with the trade-off between the various objectives of reservoir operation: the reduction of flood peak at downstream is the first priority and hydropower generation is the second priority. MiniseF T 2 1 2 ( H H ) + w ( R R ) T 1 = w1 ds _ sim ds _ opt 2 t= 1 T t= 1 T dam _ sim max H ds _ sim, : simulated, optimal water level at downstream flood control point; H ds _ opt R dam _ sim max, R : simulated, maximum water level of reservoir T : total number of time steps; w 1, w 2 : weight. 19
Optimized reservoir operation (26) Ben Ngoc is in the 1 km downstream of the Hao Binh reservoir. 2
Simulated discharges at Ben Ngoc by using the optimized release for the Hoa Binh reservoir (initial observed reservoir water level) 21
Future directions To optimize the operation of multiple reservoirs simultaneously. (not only the Hao Binh reservoir, but also the other two reservoirs in the Red River Basin). To use LDAS to assimilate AMSR-E surface soil moisture data into WEB-DHM for improving the initial conditions for the flood simulation. 22
Thank you for your kind attention! 23