СУЩЕСТВУЮЩИЕ ПРОБЛЕМЫ ГИДРОЛОГИЧЕСКИХ РАСЧЁТОВ. ЧАСТЬ 1

The article begins a discussion on the validity of applying methods of probability theory and mathematical statistics in hydrology calculations. The methodology of river flow calculations nowadays is based on the statistical processing of observation datas. These methods are specified in prescriptive documents (such as SR 33-101-2003) and, in fact, are required for using in engineering calculations. Any other alternative methods are not advisable. However, our experience of using probabilistic-statistical methods in engineering-hydrometeorological surveys let us to doubt the propriety of such application.

It is questionable whether the concept of a random variable is applied to hydrological characteristics, in particular, to water consumption. In terms of mathematics, the sample input data subsequently used in solving statistical problem is the result of an experiment conducted under unaltered conditions. Water consumption depends on many natural factors, some of which are constantly changing. In addition, dependence on some factors is almost functional. For example, precipitation causes an increase of river flow, and their absence causes a decrease. Other factors, in contrast, are quite stable. For each catchment area, they are not random and behave predictably.

For example, extrapolation of the maximum water flow rate in the region of rare probability, seems insufficiently reasoned. The error of measurements (definitions) of water flow in the river increases as its value increases. This is facilitated by the almost impossible measurement of river flow when water enters the floodplain, while floating of ice, while timber drifting on rivers, etc. The measurement methods used in these cases give an error of up to 25% by experts assessment. Herewith water flows are differs by 5-10%. Thus, we do not have the ranked values of several of the highest water flow, but some average maximum flow with regular repeatability.

As a result, it was concluded that the use of probabilistic-statistical analysis in hydrological calculations is insufficiently justified.