Takuma Takasao

Incorporation of the effect of concentration of flow into the kinematic wave equations and its applications to runoff system lumping

Abstract The usual kinematic wave equations are revised to consider the interaction between surface flow and subsurface flow in mountainous watersheds having curved surfaces covered with A-layers of uniform thickness. A function which represents watershed surface geometry, which is called the geometric pattern function in the paper, is incorporated into the basic equations of the kinematic wave model, and the depth-flow relation equation for the surface-subsurface flow system is derived. When the watershed surface is linearly converging or diverging, its geometric pattern function has a linear form, and numerical simulations for such cases are given here. If the geometric pattern function is regarded as a new parameter of the kinematic wave model, then the kinematic wave flow model becomes very flexible. In fact, when the lateral inflow is spatially uniform, the model may be used as a simple model of a stream network system. Using this simplified model, a method of transformation of the kinematic wave flow model of a stream system into a lumped runoff model, which we call a “reservoir cascade” model, is developed.

Short‐term rainfall prediction method using a volume scanning radar and grid point value data from numerical weather prediction

A physically based short-term rainfall prediction method, which uses a volume scanning radar, is extended so that it utilizes grid point values from a numerical weather prediction model as supplementary information. The original short-term prediction method mainly consists of a conceptual rainfall model that can predict rainfall distribution, particularly over mountainous regions, in a qualitative sense. On the other hand, the grid point values from the numerical weather prediction model, the Japan Spectral Model developed by the Japan Meteorological Agency, are operationally distributed as the grid point value (GPV) data. In the original short-term prediction method the three-dimensional wind field as well as initial distributions of the air temperature and water vapor were identified using topography and upper air observations. In the extended method, however, in identifying those initial values, the information from the GPV data is used instead of the upper air observations in order to accommodate large differences in temporal and spatial resolution between radar information and upper air observations. It is noted that this extended method does not use predicted GPV rainfall data. The conceptual rainfall model plays the role of bridging the gap between radar information and numerical weather prediction scales. This extended method is applied to a rainfall event which occurred in the bai-u season (one of the rainy seasons of Japan) in July 1994. Results show that for the extended lead time of three and four hours, prediction of the expanding rainfall area was improved.

A Real-Time Estimation of the Accuracy of Short-Term Rainfall Prediction Using Radar

The short-term rainfall prediction method proposed by the authors is extended to a stochastic method so that the method could provide in real-time the accuracy of the areal rainfall predicted using radar information. The extension was carried out based on the investigation on the stochastic properties of model parameters of the basic prediction model. First, we present results of theoretical analyses on the features of analytically predicted accuracy relating to the patterns of the movement of rainfall distribution. Next, we present a case study using actual rainfall distribution observed by radar during a Japanese typhoon. The results show that we can predict the mean square error of predicted areal rainfall at least within 1 hour ahead by the method proposed here.

Advanced use into rainfall prediction of three-dimensionally scanning radar

A computational method for the determination of rainfall distribution for applications in short term rainfall prediction is presented here. The method is strongly influenced by the experience gained from the observation and analysis of data gathered on a heavy rainfall event in 1986 that occurred during the Baiu Season in Japan. The method is based on the concept that rainfall occurs as an interaction between an instability field, appropriately modeled, and a field of water vapor under the influence of topography. The results from this computational method showed good agreement with the temporal variation in the rainband that moved across the observation field in 1986. Towards determination of the parameters in the computational model, another method for the determination of the rainfield is also developed. This second method determines the rainfall distribution from estimation of the conversion rate of water vapor to liquid water through use of data from a three dimensional scanning radar. The results are consistent with those obtained from the first method.

Evaluation of rainfall-runoff models from the stochastic viewpoint

Abstract Through laborious investigations, we hydrologists have stored up many experiences, much information, and various techniques and criteria for modeling river basins. There are many rainfall-runoff models available; hence we must determine which adequately fits a specific hydrological problem. Nowadays, evaluation of rainfall-runoff models has become one of the most significant themes in hydrology. This paper discusses a basic idea of model evaluation from the stochastic viewpoint. The authors present an evaluation framework as the first realization of this idea; the framework uses a Monte Carlo simulation to compare simplified models, considering whether or not they preserve the stochastic transformation properties of the runoff system represented by an ideal model. It is applied to two types of hillslope runoff processes: (1) a simple surface runoff (overland flow) process; and (2) a surface-subsurface runoff process. In the former, the kinematic wave model is regarded as an ideal one and three storage function models are compared. In the latter, the kinematic source area variation (KSAV) model proposed by Ishihara and Takasao (1962) is regarded as an ideal one, and the kinematic wave model and the lumped hillslope (LH) model proposed here are regarded as simplified ones and compared.

Design of an Inference Engine for Dam Operational Rules for Synthesized Reservoir Control Support System

A generalized inference engine which takes charge of application of dam operational regulations is designed for a synthesized reservoir control support system. First, the common expression and logical mechanics of operational regulations established for dam reservoirs in Japan are extracted. Second, an inference engine which can treat these logics such as the mutual reference among the provisions of regulations is explored. The system designed here enables us to accommodate each provision of dam operational regulations into a rule as it is, which will be a useful tool not only for inference itself but for providing support information.

Scale-up of a Distributed Runoff Model

To consider the interaction between runoff process and meteorological one, a methodology to scale up a distributed runoff model is developed.In our methodology, the stream network is regarded as a set of sub-networks devided by grids whose size is so large as to be compatible with meteorological scale. The model of flow within these sub-networks and its numerical solution algorithm are given.

A New Framework for Building Runoff Simulation Models

A basin is composed of a number of basin elements. Corresponding to this, we thought up a new system of building a runnoff model by combining element models. To realize the system, we separate following two problems. One is to build element models and the other is to build runnoff models combining element models. To solve these problems, we extract and pack common functions among element models. This new system will make it easy to correct existing runoff models and to build new one.