Journal of Hydrologic Engineering

This paper develops an algorithm for real-time correction of stream flow concentration based on a Kalman filter to improve the performance of real-time forecasting of river discharge under circumstances in which the nonlinearity of stream flow concentration is significant. The Muskingum matrix equation expresses the system of stream flow concentration as a time-varying linear system and satisfies the state-space expression of the Kalman filter. Updating of the parameter matrices of the system impair the influence of the nonlinearity of stream flow concentration on the linear filtering. The advantage of the algorithm is that predictions of every subbasin can be corrected twice by getting “remote” and “local” correction values and can achieve rational updating. Furthermore, to prevent the occurrence of filter divergence and to reach better filtering accuracy, a new real-time statistical method is proposed to estimate the process noise covariance matrix and measurement noise covariance matrix. The algorithm proves reasonable and effective by its application in the example of the Three Gorges Basin.

Transient bed sediment waves in alluvial rivers have been described using a multitude of hydraulic formulations. These formulations are based on some form of the St. Venant equations and conservation of mass of sediment in suspension and in bed. Depending on the assumptions employed, a hierarchy of formulations is expressed. These formulations in the literature employ uncoupled, semicoupled, or fully coupled transport models treating the sediment waves as either hyperbolic (dynamic wave) or parabolic (diffusion wave). It is, however, hypothesized that the movement of bed sediment waves in alluvial rivers can be described as a kinematic wave. Kinematic wave theory employs a functional relation between sediment transport rate and concentration and a relation between flow velocity and depth. This study summarizes the hierarchy of the formulations while emphasizing the kinematic wave theory for describing transient bed sediment waves. The applicability of the theory is shown for laboratory flume data and hypothetical cases.

In this study, a physically based two-source potential evapotranspiration (PE) model was developed to calculate the spatio-temporal variation of PE over the Hanzhong watershed in China. The calculated PE was then integrated with the Xin’anjiang hydrological model to perform streamflow simulation. A formula similar to the Penman–Monteith equation was used in the PE model to explicitly calculate the PE components of potential canopy transpiration, interception evaporation, and potential soil evaporation. Based on the $1\mathrm{km}$ global land cover data, the related vegetation parameters were derived from the Land Data Assimilation System. The monthly maximum composite of NOAA-AVHRR NDVI data with a $1\mathrm{km}$ resolution was used to retrieve the spatio-temporal variation of leaf area index. Meteorological data from the China Meteorological Administration were employed for calculations. A single grid cell test shows that land cover exercises a significant impact on the PE amount, its temporal variation, and associated components. The replacement of different vegetation types at the same grid cell may lead to different PE results. The spatial distribution of mean annual PE over the watershed is remarkably uneven, ranging from $268.6\text{to}970.2\mathrm{mm}$ . The simulated daily streamflow by the Xin’anjiang model using the estimated PE matches well the observed daily streamflow at the watershed outlet during the period of 1980–1986.

On the basis of geographic information systems, indices of basin terrains and some other characteristics were extracted. Drainage networks were created by different mesh sizes of digital elevation model (DEM) in the basin with different kinds of terrain. These drainage networks were analyzed and the indices of the length and slope of the flow routine were derived. According to the basic principle and the mechanism of rainfall-runoff production, the definition and physical meaning of the parameters of the Xinanjiang hydrological model, relevant relationships between several sensitive parameters, and the indices were studied. The relationships were adopted to determine the parameters and the influence was studied with different kinds of mesh sizes of DEM in the basins with different terrains. In this way, the model parameters studied derived from the relationships can be used in an ungauged basin. The results extracted in this paper can then be adopted as the theoretical foundation for the study of the digital hydrological model.

A regionalization procedure is presented and used to predict probable flooding in four ungauged coastal river basins of El Salvador: Paz, Jiboa, Grande de San Miguel, and Goascoran. The flood-prediction problem is sequentially solved for two regions: upstream mountains and downstream alluvial plains. In the upstream mountains, a set of rainfall-runoff parameter values and recurrent peak-flow discharge hydrographs are simultaneously estimated for 20 tributary-basin models. Application of dissimilarity equations among tributary basins (soft prior information) permitted development of a parsimonious parameter structure subject to information content in the recurrent peak-flow discharge values derived using regression equations based on measurements recorded outside the ungauged study basins. The estimated joint set of parameter values formed the basis from which probable minimum and maximum peak-flow discharge limits were then estimated revealing that prediction uncertainty increases with basin size. In the downstream alluvial plain, model application of the estimated minimum and maximum peak-flow hydrographs facilitated simulation of probable 100-year flood-flow depths in confined canyons and across unconfined coastal alluvial plains. The regionalization procedure provides a tool for hydrologic risk assessment and flood protection planning that is not restricted to the case presented herein.

Estimation of water fluxes, such as groundwater recharge and loss, is vital for water resources assessment and planning. In this study, lysimeter measurements at the Wudaogou Hydrological Experiment Station in the Huaihe River plain region are used for analyzing characteristics of groundwater recharge from precipitation and groundwater loss due to evapotranspiration. A multilayer soil moisture model integrating water fluxes between vegetation, soil, and aquifer is used for simulating soil moisture dynamics and groundwater recharge and loss. The model is calibrated against observed soil moisture content from 1994 to 2000. Model validation is conducted by comparing model simulated groundwater recharge and loss with the corresponding lysimeter-measured values. Statistical results from lysimeter measurements reveal influences of vegetation, depth to groundwater, and soil moisture content on groundwater recharge and loss. Modeling results prove that the model is able to reliably simulate soil moisture content and water fluxes. Water budget analysis shows that the multiyear mean of groundwater recharge is approximately 25% of the mean of potential evapotranspiration and the multiyear mean groundwater evapotranspiration loss is 22% of the precipitation amount.

Contaminated river water irrigation was conducted in two couples of lysimeters with 1 and $3.5\mathrm{m}$ soil depths, respectively, to identify factors affecting transport and transformation of organic matters and nitrogen compounds. During $2\text{months}$ of experimentation, $372\mathrm{kg}∕\mathrm{ha}$ of total nitrogen mineralized into inorganic nitrogen. The amounts of nitrification and denitrification were as high as 387.8 and $321.8\mathrm{kg}∕\mathrm{ha}$ , respectively. $40.5\mathrm{kg}∕\mathrm{ha}$ of $\mathrm{N}{\mathrm{O}}_3^{-}–\mathrm{N}$ leached into groundwater, which amounted to 78% of the initial $\mathrm{N}{\mathrm{O}}_3^{-}–\mathrm{N}$ . Therefore, denitrification and the leaching are the main processes for the loss of nitrogen in soil. Soil moisture was a primary factor affecting the transport and transformation of nitrogen (N) compounds. Lysimeter studies demonstrate that chemical oxygen demand (COD) and $\mathrm{N}{\mathrm{H}}_4^{+}–\mathrm{N}$ concentrations changed with spatial and temporal variations of the water content and the concentration of $\mathrm{N}{\mathrm{H}}_4^{+}–\mathrm{N}$ is highly positively correlated with COD. However, there is no significant correlation between soil moisture and $\mathrm{N}{\mathrm{O}}_3^{-}–\mathrm{N}$ , and the relationship between $\mathrm{N}{\mathrm{H}}_4^{+}–\mathrm{N}$ and $\mathrm{N}{\mathrm{O}}_3^{-}–\mathrm{N}$ is poor.

The head region of the Yellow River is the key area where climate change would affect the hydrological process, since it is in a high elevation and cold area. Temperature and precipitation are two important meteorologic factors influencing the hydrological process. In this study, runoff change was evaluated as a result of the hydrological process change and an evaluation of runoff response to the two factors was made. Data derived from the results of seven general circulation models (GCMs) under two intergovernmental panel on climate change scenarios (A2 and B2) were used as future climate scenarios. First, factual and future climate change is analyzed according to data observed and obtained from GCMs. It was found that temperature has been increasing since 1961 and would continue in the future. The change in future precipitation also showed an increasing tendency. Then, a distributed hydrologic model, taking into account the effect of snow and frozen soil, was developed based on observed data in order to investigate the impact of temperature and precipitation change on runoff. Simulated runoff corresponding to climate scenarios indicates that the runoff amount would change lightly before 2020 and then would decrease approximately 5% per year. Analysis of runoff characteristics showed that runoff would increase in the dry season, resulting in relatively uniform distribution of seasonal runoff. Large variablility in annual runoff in the future implies a high probability and severity of flooding as well as droughts.

The importance of nonirrigation uses of irrigation water to the livelihoods of the rural poor living close to irrigation systems has generally been ignored both in planning and operating these systems. An improved understanding of competition and complementarity of irrigation and nonirrigation uses is essential for increased productivity of irrigation water. This paper selected two typical types of large-scale irrigation systems as case study sites, i.e., the Lower Bhavani Project in Tamil Nadu, India—a canal-based irrigation system and Mahaweli System H, Sri Lanka—a tank-based irrigation system, and explored their potential for and constraints to integrated management for irrigation and aquaculture. The results revealed that for irrigation canals water supply reliability, canal flow velocity, and cage anchorage but for irrigation tanks water, supply reliability may impose constraints to integration of cage-based fish farming. However, the potential can be improved through more responsive operation and management of canal/tank irrigation systems, and enhancement in irrigation water use efficiency. This paper provides a valuable reference for irrigation managers and engineers in integration of cage-based fish farming within large irrigation systems.

Three geographic information system (GIS)-based distributed hydrological models were developed for flood simulation and forecasting in a subbasin of the Yellow River. A grid-and-topography-based distributed hydrological physical model, i.e., the GTOPMODEL model, is described that includes vegetation and root interception, evapotranspiration, and runoff generation via the saturation excess mechanism, as well as subsurface via the Darcian approach used by TOPMODEL, runoff concentration, and flow routing. The downslope redistribution of soil moisture is explicitly calculated on a grid by grid basis. Using the isochrone curve method based on GIS, another two models are developed on the basis of the Xinanjiang and TOPMODEL models. The three developed models were applied to the Upper Lushi basin in Luohe River, one tributary of the Yellow River, with an area of $\mathrm{4,716}{\mathrm{km}}^2$ for flood simulation. The results show that all of these models perform well in the simulation and can be used for flood forecasting in the Yellow River basin and that GTOPMODEL performed the best among the three models.

Groundwater contamination experiments using irrigation water from the polluted Kuihe River were conducted at the Hanwang hydrological field station, near Xuzhou City. As part of this study, a numerical model was developed on the basis of a mathematical equation with a finite-difference approximation. The model is capable of simulating nitrogen transformation processes during subsurface transport. Numerical simulations were conducted to estimate nitrogen species concentrations along a flowpath during irrigation. Because of the complex field irrigation conditions, uncertainties of the model input parameters were analyzed and probability density functions (PDFs) of the model input parameters were constructed using the available field data or professional judgement. The PDFs and a Monte Carlo simulation technique were used in this study for the evaluation of risk of groundwater contamination. The results of this study demonstrated that the nitrate mass leaching into groundwater with a 25 and 75% probability would be no less than 0.46 and 0.27 times that in irrigation water, respectively, when the Kuihe River irrigation water was used for the first day and clean drinking water was used in the next $30\text{days}$ . Therefore, utilization of the polluted Kuihe River water for irrigation would likely result in groundwater nitrate contamination.

This paper presents an application of the TOPKAPI model, a physically based, fully distributed hydrologic model, to flood simulation at three unique and flood prone sites in China. The TOPKAPI model was applied with a grid size of $1\mathrm{km}\times 1\mathrm{km}$ and a time step varying from 1 to $24\mathrm{h}$ based on the basin conditions and data availability. Model calibration was chiefly based upon moderate variations of parameter values from the ones estimated on physical grounds. The three case studies demonstrated that the model performed well for flood simulation. The TOPKAPI model represents a comprehensive process-based model, which offers a wide range of capability for flood forecasting application, and has potential in prediction of ungauged basins.

The objective of this study is to develop a new method for deriving statistical distribution of maximum discharge from the distribution of annual maximum daily rainfall by using a random cascade model. This model is used to probabilistically downscale daily rainfall into hyetographs in which the daily rainfall is randomly generated according to its probabilistic distribution. Then maximum discharge is calculated from downscaled hyetograph by means of a rainfall-runoff model. Finally, the distribution of maximum discharge is derived numerically. This technique is applied to the Doki River Basin, a small basin in Japan, where the influence of human activities is limited. The observed annual maximum discharge data from 1976 to 1998 are used to verify the calculated distribution of maximum discharge. It is shown that the distribution of maximum discharge was well reproduced.

A physically based distributed hydrological model (THIHMS-SW, TsingHua Integrated Hydrological Modeling System-for Small Watershed) was developed and applied to a $187{\text{km}}^2$ watershed in the severe soil erosion region of the Loess Plateau. In the model, calculations of water and sediment transport were coupled in each grid, and the modeling of water and soil conservation measures, especially the silt-trapping dam, was also included. Continuous simulation for a period of eight years undisturbed by human activities was carried out and the results indicate that the model worked well in terms of estimating water/sediment peak discharge, time to peak, and total volume at different locations. Continuous simulation for three years after the installation of hundreds of silt-trapping dams was also carried out and fairly good results were obtained. The model is physically based and can be used in water resources planning, land use management, flood control, as well as water and soil conservation planning in small watersheds.

Due to the spatial variability of precipitation, topography, soil properties, and antecedent soil moisture, runoff is generated by more than one kind of mechanism. Based upon digital elevation model-derived subcatchments and river network, the Xinanjiang Model (XAJ, saturation excess mechanism of rainfall-runoff partitioning), the Shaanbei Model (SAB, infiltration excess mechanism), and the Hybrid Model (HYB, two mixed runoff mechanisms by combining spatial distribution curves of soil tension water storage capacity and infiltration capacity) were applied for streamflow modeling at four hydrological observational stations in the Laohahe River basin, and compared at different time intervals ranging from 1 to 24 h. The results show that the daily (24 h) HYB Model or XAJ Model performs better than the SAB Model over semihumid and semiarid areas. It is revealed that there is less possibility that rain intensity is larger than infiltration rate in semihumid and semiarid regions within a duration of 24 h. Uniform distribution of precipitation within a duration of 24 h makes the physical concept of infiltration excess fuzzy. In the event-based flood modeling at time intervals of 1, 2, 3, 4, 6, and 12 h, HYB is superior to XAJ and SAB in semiarid areas where the mean annual precipitation is less than 400 mm. It is concluded that it is more meaningful to use infiltration excess mechanisms for shorter-duration flood modeling than for daily streamflow simulation in semihumid and semiarid regions.