Development of a model-based flood emergency management system in Yujiang River Basin, South China

Abstract

Flooding is the most frequent disaster in China. It affects people’s lives and properties, causing considerable economic loss. Flood forecast and operation of reservoirs are important in flood emergency management. Although great progress has been achieved in flood forecast and reservoir operation through using computer, network technology, and geographic information system technology in China, the prediction accuracy of models are not satisfactory due to the unavailability of real-time monitoring data. Also, real-time flood control scenario analysis is not effective in many regions and can seldom provide online decision support function. In this research, a decision support system for real-time flood forecasting in Yujiang River Basin, South China (DSS-YRB) is introduced in this paper. This system is based on hydrological and hydraulic mathematical models. The conceptual framework and detailed components of the proposed DSS-YRB is illustrated, which employs real-time rainfall data conversion, model-driven hydrologic forecasting, model calibration, data assimilation methods, and reservoir operational scenario analysis. Multi-tiered architecture offers great flexibility, portability, reusability, and reliability. The applied case study results show the development and application of a decision support system for real-time flood forecasting and operation is beneficial for flood control.

Keywords

flood / decision support system / numerical modeling / scenarios analysis / Yujiang River Basin

Introduction

Flooding is an excess of water which is higher than the normal water level in a channel and thus extends out of the channel, pouring into floodplains or lands with intensive human activities (Easterling et al., 2000; Tan et al., 2010a, b). Recently, great floods have affected many regions, causing great economic and property loss and leading to major concerns across the world. As stated by Loucks et al. (2005), flooding has affected approximately 520 million people and their livelihoods, and is responsible for over 25,000 mortalities worldwide every year. At the same time, annual economic losses caused by flooding were estimated at 8.5 to 24.5 billion US dollars (World Meteorological Organization, 2004). Alleviating flooding impacts and effectively planning of emergency management practices are necessary for many countries including China which is highly populated and vulnerable to flooding (Ahmad and Simonovic, 2006). However, there are many complex processes that should be considered, such as intensity of precipitation, probability of flood occurrence, intervention of human activities, as well as the related economic implications. Moreover, many factors and parameters of this management process may appear dynamic and uncertain, which can hardly be tackled by conventional tools. Such complexities and dynamics would greatly affect relevant decision-making processes, calling for efficient and effective approaches for supporting emergency management of flooding. Therefore, development of computer-aided tools for identifying decision alternatives for managing emergencies associated with various flooding events is desired.

Previously, a huge number of studies were conducted on flood management and associated impacts. For example, Simonovic and Ahmad (2005) divided a flood management process into three stages, i.e., pre-flooding planning, flooding emergency management, and post-flooding recovery. Among them, the process of flooding emergency management involves the forecast and update of floods. Frequent assessment of current flood situations and operations of flood-controlling infrastructures are important during this stage. Based on statistical methods and monitoring data, Plate (2007) predicted the occurrence of floods downstream of a river. At the same time, the methods based on linear stochastic auto-regressive moving-average models (ARMA), artificial neural network (ANN), and cellular automata approach (CAA) were adopted for the prediction of rainfall-runoff, river inflow, and the corresponding flooding process (Toth et al., 2000; Shim et al., 2002; Jarosław and Tomasz, 2004; Chidthong et al., 2009; Liu et al., 2009; Chen et al., 2010; Feng and Lu, 2010; Dottori and Todini, 2011). More recently, a number of physical approaches were developed for forecasting flooding events which were mainly based on hydrological and hydraulic models (Bocchiola and Rosso, 2009; Grimaldi et al., 2013). The advantage of physical models is their ability to predict the effect of changes in the river system (Braud et al., 2010). Moreover, along with the rapid development of information technologies, it is necessary for flooding management to integrate internet, GIS, hydrological modeling, and database management technologies into a general computer-aided framework to form a decision support system (i.e., DSS; Martin et al., 2005). A DSS is a computerized management advisory system that could utilize databases, models, and user-dialog systems to provide decision makers with timely management information and strategies (Grigg, 1996). At the same time, GIS is a computer-based tool that can efficiently generate, store, analyze, retrieve, manipulate, manage, and graphically display complex spatial information (Ramlal and Baban, 2008; Dong et al., 2011). In order to strengthen advantages of these tools, their combination has attracted much attention (Plate, 2007; Cai et al., 2011; Zeng et al., 2011, 2012). For instance, Zhao et al. (2005) presented the application of an inflow forecasting system based on MIKE 11 and GIS in the world’s largest dam (i.e., the Three Gorge Dam in China). They provided inflow forecasts to facilitate operations of the dam (e.g., maximizing power production, retaining minimum downstream river flows, and protecting the public against flooding). Todini (1999) developed a flood operational decision support system, which could be used for supporting hydrological and hydraulic analyses and real-time flood-forecasting with the possibility of analyzing the advantages of different human-intervention scenarios. Krzhizhanovskaya et al. (2011) proposed an early warning system (EWS) to estimate dike failure probability, and simulate possible dike breech scenarios and flood propagation. de Kort and Booij (2007) developed a pilot-scale decision support system for flood control in the Red River basin of Vietnam and China. Levy (2005) believed that the adoption of DSS could improve effectiveness of flood-risk planning and management under uncertainty through analyzing, summarizing, and displaying critical flood information. Qi and Altinakar (2011) advanced a decision support system for integrated flood management within the framework of ArcGIS based on a two dimensional flood simulation modeling system. More recently, external data sources such as radar systems were used to complement conventional management tools (Jasper et al., 2002). Moore et al. (2005) utilized remote sensing data and numerical weather prediction models to improve flood forecasting performances. Cloke and Pappenberger (2009) reviewed the improvements of flood forecasting based on ensembles of numerical weather predictions (NWP).

Flooding is particularly frequent in China. There have been over 1,000 great floods in China in the last 500 years, including the disastrous ones on the Yangtze River, Nenjiang River, and Songhuajiang River in 1998 (The Ministry of Water Resources of the People’s Republic of China, 1999). Over 1.10 hundred million people were affected by floods in 2010 in China (Su et al., 2011). Conventional policies of flood forecasting and reservoir operation in China were mostly based on statistical simulation by hydrologists. There were a few shortcomings in these methods, such as long forecasting time, an individual’s knowledge limit, lack of communication, and absence of expertise (Li et al., 2006). Comparatively, a DSS for flood control can bring significant convenience to decision makers who are engaged in flood forecasting and control. It can also allow real-time contribution of many experts at different locations. Due to this capability, practical applications of DSS have been reported in China since the early 1980s. In the mid-1980s, many Chinese scholars attempted to adopt DSS to facilitate water resource planning and management (Tian et al., 2007). Such applications mostly cover three aspects: (i) the development of communication networks and data processing systems to collect water related data (Sheng et al., 2004; Li et al., 2006; Guo et al., 2010), (ii) the development of automated forecasting systems for flood prevention (Xi, 1990; Zhao et al., 2005; Lin et al., 2010), and (iii) the undertaking of water resource planning and management (Feng et al., 2007; Yang et al., 2008). Although great progress has been achieved in flood forecast through using computers, network technology, and geographic information system technology, the prediction accuracy of models are not satisfactory due to the unavailability of real-time monitoring data. According to Achleitner et al. (2012), the mean relative bias between the forecasted and the observed discharges vary between 0.2 and 0.25 in different catchments. The complexities of flood generation processes and their dependency on different factors related to watershed properties and rainfall characteristics make flood-forecasting a difficult task (Rozalis et al., 2010). In order to develop an online flood forecasting and control system, database technology, GIS, and hydrological and hydraulic models should be integrated as into a general framework to form a DSS. This is of particular significance in many regions in China due to many challenges facing by various water managers and decision makers, such as lack of resources and real-time monitoring systems.

Therefore, the objective of this research is to develop an integrated decision support system through the integration of multiple modules. Through the developed DSS, real-time water information and the associated flood-occurrence probabilities can be predicted, providing a series of follow-up strategies and policies in dealing with floods. At the same time, the developed DSS is to be applied in Yujiang River Basin to demonstrate its applicability, leading to the formation of the decision support system for flood emergency management in the basin. Such a system will provide functions of online query, forecast, and decision support as a sharing system on the web when a river flood emergency occurs. The hydrological and hydraulic models combined with the data assimilation method will highly improve the accuracy of flood prediction. Real time operations of reservoirs, as an important online flood control method, are also realized by the developed DSS. Users can select best operation rules of reservoir according to different simulating results in the system, which will be created by set different operation rulers of reservoirs.

Overview of Yujiang River Basin

The Yujiang River Basin is located in Guangxi of southwestern China. It is one of the major branches of the Pearl River Basin, which is one of the seven major river basins of China. The Yujiang River Basin has two main branches, i.e., Youjiang River and Zuojiang River. The Youjiang River flows roughly south-east from its source close to Baise City in Guangxi, and the Zuojiang River flows north-east from its sources in Vietnam. The two rivers meet in the west of Nanning City and merge as Yujiang River. The Yujiang River Basin includes a population of 62 million people and a total catchment area of 9.2 × 10⁴ km², in which the Youjiang River sub-basin accounts for 4.1 × 10⁴ km² and 3.2 × 10⁴ km² of the Zuojiang River sub-basin and 1.9 × 10⁴ km² of the downstream of main Yujiang River Basin. The main river length is approximately 1,182 km with average river slope of 0.33%.

The average annual rainfall in the Yujiang River Basin varies from 1,200 to 2,200 mm. Most rainfall is from June to September, and average rainfall in a wet year may be three times than that of a dry year. The Yujiang River Basin covers around nine cities. Among them, Nanning City, as the capital of Guangxi, is the most important protected target in the basin flood control.

Many blocks in Nanning City have low elevation, which are prone to flood disaster. According to statistical data, it had 12 flooding years from 1950 to 1975. In 1968, the affected population was 410,000, 4,000 hm² of cultivated land was flooded, and property damage was 32.1 million CNY at current prices. In order to prevent flooding, the local government built a number of engineering infrastructures to protect local socio and economic resources against flooding, such as Baise, Chengbihe, and Naban reservoirs. However, the flood control design standards of those reservoirs were established 20 years ago. According to statistical analysis, Nanning economic loss would be 10.1 billion CNY (price of 1994 level) if on 50 year flood happened. The flood events with flow of more than 11,100 m³/s are listed in Table 1.

In this situation, for the effective forecasting of the characteristics of flood flow and utilizing the flood control system that is existing in the basin, a flood control decision support system, which integrated rainfall monitoring technology, data transfer technology, database storage technology, geographic information technology, hydrological and hydraulic models, and web technology should be established for evaluating, planning, and managing floods.

System structure

Through analyzing the characteristics of flood occurrence and prevention, the structure of the decision support system can be proposed through the combination of three components, i.e., user interfaces, services, and database subsystems, which are shown in Fig. 1. Among them, the database subsystem stores all the data types required by the DSS, such as images, spatial data, and attribute information. Images are interpreted photos of satellite remote sensing and other sources. Spatial data include the basic mapping data (e.g., topographical maps, land use data, and other spatial features, as well as output maps (e.g., characteristics of floods).The attribute data include time-series hydrological data, hydraulic information, social and economic data, and output date. The service subsystem, which serves the DSS-YRB by receiving and processing basic data and putting results to the user interface or to the database management subsystem, is comprised of two modules: (i) a modeling module, include the hydrological and hydraulic mathematical models, and data assimilation methods, such as NAM, MIKE 11, MIKE FLOOD, MIKE11 DA module (DHI, 2009), and (ii) database management module and GIS. The system can provide information inquiring functions, such as precipitation, flood flow, hydraulic architecture, and also provide scenario analysis functions, which are the basis for desired operational strategies of reservoirs during a flood. The user interface can be regarded as a platform through which users could identify real-time flood situations presented by the DSS-YRB and facilitate the adoption of data and models. In order to develop the system, Microsoft Visual Studio 2005 is used as the programming environment and C is selected as the coding language.

The C/S and B/S methods are combined in the development and integration of the system. Flood forecasting and scheduling are based on the professional platform software of DHI. This part adopts the C/S mode for development. Other related information and results are used frequently for all levels of users. This part adopts B/S model for development. The flow charts of the system are shown in Fig. 2.

The operating process of DSS-YRB is as follows: meteorological data are transferred automatically into the system as input time-series data. The rainfall-runoff information of each basin is forecasted by the RR module (NAM) of Mike11, which has been integrated into the system. Water levels and discharge amount at each river section are calculated by the HD module of Mike11. The predicted hydrological results of HD are updated based on the monitoring data of hydrological stations through using DA module. The analysis results can be displayed by maps and/or tables based on the web GIS. The connection between the DSS and the modeling software are as follows: The flood forecasting model is constructed off-line. Then, the proposed DSS will automatically identify the time series input data of the modeling boundary information (e.g., rainfall, evaporation and upstream inflow) through rules defined in advance. The module in DSS updates the data and consequently initiates the forecast model. The input data and the generated results are transported to the database. Then, the results and the corresponding input date and relevant information can be queried, displayed, and extracted in the front desk of DSS.

The modeling system

Rainfall-runoff modeling

A so-called continuous conceptual mode NAM model which is fully and automatically coupled to the Mike11 package was adopted for rainfall-runoff modeling of the Yujiang River Basin, due to its characteristics of continuous process and much higher accuracy compared to event-based models. The NAM model simulates the rainfall-runoff by continuously accounting the moisture content in four different interrelated storage locations representing the physical elements of the catchment, snow layer, surface zone (e.g., vegetation, small channel and lakes), root zone (the depth from which plant roots draw water), and ground water (DHI, 2009).

The meteorological input data to the NAM model are precipitation, potential evapotranspiration, and temperature (if snow modeling is included). On this basis, it produces, as its main results, runoff and groundwater level values as well as information about other elements, such as soil moisture content and groundwater recharge. The runoff of catchments is classified conceptually into overland, inter, and base flows. An automatic calibration module is available in NAM, which allows calibration of the nine most important model parameters. It is based on a simultaneous optimization of up to four different objectives, including water balance, overall hydrograph shape, peak flows and low flows. For a model calibration that includes all nine parameters, a maximum number of model evaluations in the range 1,000 to 2,000 normally ensure an efficient calibration, and this is typically done in 30 to 60 CPU seconds.

Hydrodynamic modeling

The MIKE11 hydrodynamic model is used to simulate flows and water levels in the river channel, which forms the framework for real-time flood forecasting. MIKE11 HD is a one-dimensional hydrodynamic model related to flood forecasting and reservoir operation, simulation of flood control measures, tidal and storm surge studies in rivers and estuaries. The model is based on the vertically integrated equations of conservation of continuity and momentum, i.e., the Saint Venant equations. These equations are solved using an implicit finite difference scheme by applying the double sweep algorithm. The solution can apply to single branched, as well as to looped and branched river systems. The computational grid comprises alternating Q (discharge) and H (water level) points. Cross-sectional data are given at H-points whereas Q-points are automatically placed midway between neighboring H-points and at hydraulic structures.

Data assimilation

The simulation results may have errors comparing with observed data. Many factors (such as quality of data, uncertainty of model structure, uncertainty of model parameters) impact accuracy of the model. This can be reduced by data assimilation methods (Blöschl et al., 2008). The data assimilation method used in DSS is the MIKE11 DA module. It includes two different melding schemes based on the Kalman filter: the Ensemble Kalman filter and a constant weighting function. The main principles of this module are that it can update the initial state of a system by using errors, which exist between the latest prediction and observed data, before the next prediction. The accuracy of estimate will be improved by providing improved initial conditions.

Application

Results of NAM model

A 90 m × 90 m digital topographic map of the river basin has been downloaded from the web site of United States Geological Survey (USGS). The digital elevation model (DEM) of the studied areas is extracted from this by a GIS spatial data processing package. According to the water divide of each river, Yujiang basin was divided into 41 sub-basins, including 19 source sub-basins and 22 interval sub-basins (Fig. 3).

The data of each sub-basin collected are as follows: (i) The continuous daily rainfall recorded from 107 rainfall stations, including a continuous rainfall time series from 2004 to 2008 from 22 stations, and continuous rainfall data from 2004 to 2006 from the rest of the stations, (ii) The daily potential ability to evaporate in the 20 stations, including 14 stations’ data from 2004 to 2006, and the rest of the stations from 1980 to 1984, (iii) The continuous time series flow record from 2004 to 2008 from 39 hydrological stations, including 13 hydrological stations located in the source river basin, and (iv) The inflow, outflow, and operation rules of reservoirs from 2007 to 2008 . Among them, 13 river sub-basins have historical hydrological data. The other sub-basins, which have no hydrological record history, were treated as forecast stations. If the source sub-basin has historical flow data, the parameters of rainfall runoff model can be direct calibrated in the NAM model. The basin intervals were joint calibrated combined with the NAM model and hydrodynamic model. The no-data sub-basin can only be characterized by referring to a similar sub-basin. The calibration results of three calibration stations, Naan, Bada, and Chengbihe reservoir, are shown in Fig. 4, which represent precision of good, moderate, and poor calibration, respectively.

The precision of calibration in NAM model can be assessed by the coefficient of determination, which is as follows:

(1)DC=\mathrm{1}-{\displaystyle \frac{{\displaystyle \sum _{j=\mathrm{1}}^n{\left[Q_c\left(j\right)-Q_t\left(j\right)\right]}^{\mathrm{2}}}}{{\displaystyle \sum _{j=\mathrm{1}}^n{\left[Q_c\left(j\right)-Q_{ta}\right]}^{\mathrm{2}}}}},

where DC is the coefficient of determination, Q_c(j) is the time serious data of prediction, Q_t(j) is the time serious data of observation, and Q_ta is the average value of time serious data of observation. According to the Chinese water conservancy industry standard (hydrology information forecast specification (SL250-2000)), If DC≥0.7, then the model can be used for forecasting.

The coefficient of determination of six source sub-basins can be seen in Table 2. The calibration accuracies of the Ronghua, Naan, and Yingzu sub-basins are ideal, of which the peak value, peak flow process, flow duration curve, and total water balance of predicting results for outlet flow fits observation results well, the deterministic coefficient are higher than 0.8, and the total water volume error is less than 5%. The accuracy of calibration for reservoir sub-basin is not ideal because not enough rainfall data are available.

Results of MIKE11 HD model

The data needed for MIKE11 HD are as follows: river name, river length and connections among rivers, characteristics of hydraulic structures, position of hydrological station, topography of rivers, and time-series of hydrological data. Fifteen river branches have been included in the model, including three mainstreams: Zuojiang, Youjiang, Main Yujiang. The studied areas also involve Baise reservoir, Chengbihe reservoir, Xianhu reservoir, Naban reservoir, Kelan reservoir, etc. According to the collected information, only Baise reservoir is a scheduling reservoir, the other reservoirs are conventional overflowing flood dams, which have no scheduling ability. Scheduling permits discharge overflow if water level higher than the dam elevation. The nine hydrological stations are calibrated in the studied areas, of which the Nanning station is selected as an example, and the calibrated results can be seen in Fig. 5.

The assessment standard of accuracy of hydrodynamic models is as follows: the prediction results of flooding can be seen as qualified if the average relative error of flood volume between prediction and observation is<20%, and that of relative error of flood peak is<20%. The eight floods from 2007 to 2008 in Nanning station are selected as examples to assess the accuracy of model results, among them, the qualified rate of prediction is 87.5%, which infers that hydrodynamic models can be used for prediction.

Results of data assimilation

Taking Nanning station as an example, a flood event of Yujiang in September 24, 2008 is simulated by the DA model. The outlet of Baise reservoir is set as the upstream boundary condition. The forecast period is three days, because the average travel time is about 48 hours from the outlet of Baise reservoir to Naning station. The rolling forecast of historical flood events in 2008 can be seen in Fig. 6. The results show the model accuracy gradually improves through the real time calibration. The average relative error of flood volume between prediction and observation for MIKE11 HD model is 14.7%, and that of relative error of the flood peak is 12.1%. After data assimilation, the accuracy of the model improves to 7.2% and 5.6% each for flood volume and flood peak, respectively. The accuracy of the proposed system has been significantly improved.

Flood management system

The function of a system for early warning of flood is realized through contrast between the forecast maximum water level and that of the levee. If height difference is less than one meter, an early warning will be issued. Decision makers can quickly understand the danger of dam breakage through this system. The system also provides the information of water level per longitudinal river section. Dynamic changing with time in the water level along the river will help decision makers find the danger of dam breakage along river. Flood control through the reservoir regulation system is realized by using the interface of the reservoir regulation module. Decision makers can control the downstream flood by changing reservoir outflow through this module. Different predicted downstream flooding will be a product of the above process. And decision makers can select the optimal scenario according to the flood control targets. The stage hydrograph at Nanning station under condition of four alternatives of water release for Baise reservoir are compared in Fig. 7. The water level averages 63.0 m at the default release-way of the reservoir, which exceeds the 20 cm of river bank at this station. The water level can be decreased about 30 to 50 cm under different reduced reservoir outflows. Decision makers can identify the optimal release-way of the reservoir to avoid a downstream river bank break.

Conclusions

In this study, a decision support system for emergency management of floods was developed in the Yujiang River Basin, China (DSS-YRB). Such a system was based on a typical GIS platform with the capabilities of displaying real-time hydrological and meteorological information in the basin. Also, it could be used for supporting flood forecasting, and analyzing flood management alternatives and thus greatly assist decision makers in understanding the impacts of such alternatives flooding patterns and impacts. A comprehensive real time forecasting system is generated for the Yujiang River Basin. The system comprises several real time telemetry databases and hydrological and hydraulic models. The system includes highly visual tools for examining observed and forecasted hydrological and hydraulic data. In a web environment the system provides users with the capabilities to issue spatial queries for information of hydrology and hydraulic structure. The system can be used to simulate different scenarios of reservoir regulation, and manage flood disasters in the best way possible.

References

[1] AchleitnerS, SchöberJ, RindererM, LeonhardtG, SchöberlF, KirnbauerR, SchönlaubH (2012). Analyzing the operational performance of the hydrological models in an alpine flood forecasting system. J Hydrol (Amst), 412–413: 90–100

[2] AhmadS, SimonovicS P (2006). An intelligent decision support system for management of floods. Water Resour Manage, 20(3): 391–410

[3] BlöschlG, ReszlerC, KommaJ (2008). A spatially distributed flash flood forecasting model. Environ Model Softw, 23(4): 464–478

[4] BocchiolaD, RossoR (2009). Use of a derived distribution approach for flood prediction in poorly gauged basins: a case study in Italy. Adv Water Resour, 32(8): 1284–1296

[5] BraudI, RouxH, AnquetinS, MaubourguetM M, ManusC, VialletP, DartusD (2010). The use of distributed hydrological models for the Gard 2002 flash flood event: analysis of associated hydrological processes. J Hydrol (Amst), 394(1–2): 162–181

[6] CaiY P, HuangG H, TanQ, ChenB (2011). Identification of optimal strategies for improving eco-resilience to floods in ecologically vulnerable regions of a wetland. Ecol Modell, 222(2): 360–369

[7] ChenC S, ChenB P T, ChouF N F, YangC C (2010). Development and application of a decision group Back-Propagation Neural Network for flood forecasting. J Hydrol (Amst), 385(1–4): 173–182

[8] ChidthongY, TanakaH, SupharatidS (2009). Developing a hybrid multi-model for peak flood forecasting. Hydrol Processes, 23(12): 1725–1738

[9] ClokeH L, PappenbergerF (2009). Ensemble flood forecasting: a review. J Hydrol (Amst), 375(3–4): 613–626

[10] de KortI A T, BooijM J (2007). Decision making under uncertainty in a decision support system for the Red River. Environ Model Softw, 22(2): 128–136

[11] DHI (2009). Mike11—A modeling system for rivers and channels. Reference manual, DHI Software 2009, DHI Water & Environment, Horsholm, Denmark

[12] DongC, HuangG H, CaiY P, XuY (2011). An interval-parameter minimax regret programming approach for power management systems planning under uncertainty. Appl Energy, 88(8): 2835–2845

[13] DottoriF, TodiniE (2011). Developments of a flood inundation model based on the cellular automata approach: testing different methods to improve model performance. Phys Chem Earth, 36(7–8): 266–280

[14] EasterlingD R, EvansJ L, GroismanP Y, KarlT R, KunkelK E, AmbenjeP (2000). Observed variability and trends in severe weather climate events: a brief review. Bull Am Meteorol Soc, 81(3): 417–425

[15] FengL H, LuJ (2010). The practical research on flood forecasting based on artificial neural networks. Expert Syst Appl, 37(4): 2974–2977

[16] FengS, LiL X, DuanZ G, ZhangJ L (2007). Assessing the impacts of South-to-North Water Transfer Project with decision support systems. Decis Support Syst, 42(4): 1989–2003

[17] GriggN (1996). Water Resources Management: Principles, Regulations, and Cases. New York: McGraw-Hill

[18] GrimaldiS, PetroselliA, ArcangelettiE, NardiF (2013). Flood mapping in ungauged basins using fully continuous hydrologic-hydraulic modeling. J Hydrol (Amst), 487: 39–47

[19] GuoT, YuH, LuoJ (2010). The development and research of water management information system based on Web GIS. Environmental science and management, 35 (9):173–178(in Chinese)

[20] JarosławJ N, TomaszD (2004). Decision Support System for flood control in trans-boundary Nysa Klodzka catchment, Integrated Water Management of Transboundary Catchment—A Contribution from TRANSCAT, Conference Proceedings, Venice, Italy

[21] JasperK, GurtzJ, LangH (2002). Advanced flood forecasting in Alpine watersheds by coupling meteorological observations and forecasts with a distributed hydrological model. J Hydrol (Amst), 267(1–2): 40–52

[22] KrzhizhanovskayaV V, ShirshovG S, MelnikovaN B (2011). Flood early warning system: design, implementation and computational modules, International Conference on Computational Science, ICCS 2011, Procedia Computer Science, 4:106–115

[23] LevyJ K (2005). Multiple criteria decision making and decision support systems for flood risk management. Stochastic Environ Res Risk Assess, 19(6): 438–447

[24] LiX Y, ChauK W, ChengC T, LiY S (2006). A Web-based flood forecasting system for Shuangpai region. Adv Eng Softw, 37(3): 146–158

[25] LinC A, WenL, LuG, WuZ, ZhangJ, YangY, ZhuY, TongL (2010). Real-time forecast of the 2005 and 2007 summer severe floods in the Huaihe River Basin of China. J Hydrol (Amst), 381(1–2): 33–41

[26] LiuY, HuangG H, CaiY P, ChengG H, NiuY T, AnK (2009). Development of an inexact optimization model for coupled coal and power management in North China. Energy Policy, 37(11): 4345–4363

[27] LoucksD P, van BeekE, StedingerJ R (2005). Water Resources Systems Planning and Management: An Introduction to Methods, Models and Applications, UNESCO, Paris

[28] MartinP H, LeBoeufE J, DobbinsJ P, DanielE B, AbkowitzM D (2005). Interfacing GIS with water resource models: a state-of-the-art review. J Am Water Resour Assoc, 41(6): 1471–1487

[29] MooreR J, BellV A, JonesD A (2005). Forecasting for flood warning. C R Geosci, 337(1–2): 203–217

[30] PlateE J (2007). Early warning and flood forecasting for large rivers with the lower Mekong as example. J Hydro-environment Res, 1(2): 80–94

[31] QiH, AltinakarM S (2011). A GIS-based decision support system for integrated flood management under uncertainty with two dimensional numerical simulations. Environ Model Softw, 26(6): 817–821

[32] RamlalB, BabanS M J (2008). Developing a GIS based integrated approach to flood management in Trinidad, West Indies. J Environ Manage, 88(4): 1131–1140

[33] RozalisS, MorinE, YairY, PriceC (2010). Flash flood prediction using an uncalibrated hydrological model and radar rainfall data in a Mediterranean watershed under changing hydrological conditions. J Hydrol (Amst), 394(1–2): 245–255

[34] ShengD, HeX, LiuH (2004). Study on water resources management information system of Manasi River Basin. Journal of Water Resources &Water Engineering, 15(1): 8–12 (in Chinese)

[35] ShimK C, FontaneD G, LabadieJ W (2002). Spatial decision support system for integrated river basin flood control. J Water Resour Plan Manage, 128(3): 190–201

[36] SimonovicS P, AhmadS (2005). Computer-based model for flood evacuation emergency planning. Nat Hazards, 34(1): 25–51

[37] SuW, ZhangX, WangZ, SuX, HuangJ, YangS, LiuS (2011). Analyzing disaster-forming environments and the spatial distribution of flood disasters and snow disasters that occurred in China from 1949 to 2000. Math Comput Model, 54(3–4): 1069–1078

[38] TanQ, HuangG H, CaiY P (2010a). A superiority-inferiority-based inexact fuzzy stochastic programming approach for solid waste management under uncertainty. Environ Model Assess, 15(5): 381–396

[39] TanQ, HuangG H, CaiY P (2010b). Radial-interval linear programming for environmental management under varied protection levels. J Air Waste Manag Assoc, 60(9): 1078–1093

[40] The Ministry of Water Resources of the People’s Republic of China (1999). 1998 Flood in China, Beijing: China Water Power Press (in Chinese)

[41] TianJ, WangY, LiH, LiL, WangK (2007). DSS development and applications in China. Decis Support Syst, 42(4): 2060–2077

[42] TodiniE (1999). An operational decision support system for food risk mapping, forecasting and management. Urban Water, 1(2): 131–143

[43] TothE, BrathA, MontanariA (2000). Comparison of short-term rainfall prediction models for real-time flood forecasting. J Hydrol (Amst), 239(1–4): 132–147

[44] World Meteorological Organization (2004). Integrated Flood Management. The Associated Programme on Flood Management, APFM Technical Document No.1

[45] XiY (1990). The architecture of a DSS for Three Gorges Project, Scientific Decision with System Engineering, Beijing: Science and Technology Press of China (in Chinese)

[46] YangW, NanJ, SunD (2008). An online water quality monitoring and management system developed for the Liming River basin in Daqing, China. J Environ Manage, 88(2): 318–325

[47] ZengY, CaiY, HuangG, DaiJ (2011). A review on optimization modeling of energy systems planning and ghg emission mitigation under uncertainty. Energies, 4(12): 1624–1656

[48] ZengY, CaiY, JiaP, JeeH (2012). Development of a web-based decision support system for supporting integrated water resources management in Daegu city, South Korea. Expert Syst Appl, 39(11): 10091–10102

[49] ZhaoY, MinY, PedersonC B (2005). Provision of a real-time inflow forecasting system tailored for the optimization and operation of Three Gorges Dam, China. International Conference on Reservoir Operation and River Management, Three Gorges, China

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg