Vízügyi Közlemények, 1977 (59. évfolyam)

4. füzet - A Vízügyi Konferencián előadások keretében elhangzott magyar tanulmányok - Starosolszky Ödön: Nagy vízfolyások hidrológiai paraméterei becslésének és elemzésének módszerei

Nagy vízfolyások hidrológiai paraméterei Ii25 either individually, or regularly in an established observation network. The relevant quantities are found directly, or indirectly from relationships between cause and effect. Data series are characterized commonly by their inean— and extreme values, further by the standard deviations thereof. The description by means of distribution functions is widely adopted. Difficulties are encountered on account of the fact that the relationship between the values observed and to the computed, e.g. the stage-discharge curve, is not an unambiguous one, but takes often the form of a hysteresis loop (Fig. 1). The characteristic hydrologie quantities are often found by interpolation, as exemplified by the channel dimensions involved in the determination of surface profiles and unsteady water levels ( Fig 2). An adequate number of surveying crosssections at the desired position is hardly conceivable. In other cases the parameter value required can be found in several successive steps only, as for instance the constant describing the velocity profile (Fig. 3). The determination of I lie roughness coefficient is similarly an involved process. The frequency- and distribution functions of the parameters of channel geometry (Fig. 4) provide the information needed for the determination of typical values. The standard deviation of smoothness values obtained in this way is appreciably smaller. The effect of the smoothness coefficient on the surface profile can be examined in terms of the length of sections over which averaging is performed (Fig. 5). The magnitude of К is greatly influenced by the length of the computation (averaging) sections. The correct assumption of the hydraulic parameters is essential to further, more complicated hydraulic computations, such as the computation of surface profiles upstream of barrages affected by the operation ol' hydroelectric stations (Fig. в). Specific problems call for particular applications of the parameters. Flood regimes are characterized with the help of functions having annual periodicities (Fig. 8). The probability of the duration of flood waves surpassing a particular elevation—as one of the important parameters in flood fighting operations—can be estimated by identifying the critical flood waves (Fig. 9). Among the engineering parameters of streams great importance is attributed to those describing the ice regime and the particulars of the ice cover (Fig. 10). The construction of flood levees along the streams calls for analyses concerning the potential impacts of confining flood flows [10], the conventional methods of which involve a static approach and the water level elevations obtained hv them are misleading. The method developed recently [10J is founded on the theory of unsteady, varied flow and yields more accurate results. For some typical cases (Fig. lő) the difference has been computed using a particular kind of flood hydrograph (Fig. 14), the results being presented in Table 11. The water quality parameters apply to individual cross-sections, but can be examined along the stream as well. Three main types of cross-section parameter are distinguished [13] (Fig. Hi). The representation along the stream (Fig. 17) cannot be interpreted, unless il applies to steady conditions and for this purpose the design streamflow rate must be estimated. Variations in the biological parameters are influenced by the phenomenon of sell-purification. The phenomenon of mixing (Fig. 18) may assume paramount significance in the vicinity of effluent discharges and may fundamentally influence sampling procedures and the analysis of results alike. The parameters of stochastic character are described by diverse distribution functions, of the empirical-, theoretical- and fitting distribution functions tlieempirical- and fitting functions being of practical interest. In the case of Ihe three-parameter distribution functions (log-normal and gamma) the parameter Xo — is assumed, the remaining two are estimated therefrom ( Figs. 19, 20) and the fitting function is then constructed ( Fig. 21 ). The problem of generating artificial time series is also frequently encountered. The identification and elimination of the deterministic components is commonly followed by resolution into components (trend-, periodic- and random components). li should be emphasised that the artificial time series contains no more information than the observed record on which il is based, nevertheless it is useful in some problems, such as storage. As will be perceived from tlie foregoing, Ihe estimation of the manifold parameters involved in fluvial hydrology calls for extensive preliminary obser-