DETERMINATION FOR RISKS IN HYDROLOGICAL AND WATER RESOURCES PROJECTS.

Statistical analysis is used in hydrology where variables and processes
are observed and stated in terms of numbers. The majority of hydrologic
phenomena in nature, such as precipitation, runoff, evaporation, sediment
transport, water quality properties, elc., are stochastic processes. However,
in considering events on an annual basis, the occurence of the events may be
assumed independant and the hydrologic system may be assumed to be time invariant.
Magnitude-frequency relationships are used frequently in water resources
design of engineering projects. The empirical simple frequency distributions,
as the estimate of the population probability distributions, are
often the only available information. The practical plotting position formulas
are summarized in this paper.
The return period established by frequency analysis such as just described
indicates only the average interval between events equal to or greater
than a given magnitude, or the probability that such an event will occur in
anyone year. If it is desired to select a design flow which is most likely
to occur during the life of the structure, it is necessary to use a return
period greater than the estimated useful life.
Often the design return period of a water resources project can be
treated as a function of the useful life of the project and the permissible
risk of failure. The risk in this subject is identified with probabilities
of value greater, or smaller than, a given value. However, there are addi
tional risks involved which should be considered in water resources projects,
such those due to random measurements and computation, errors of observation,
systematic errors in data, non homogenity, loss of information, sampling
errors, use of inefficient statistical methods in extracting the information
from a pool of data and other errors.
The paper shows the development of cumulative frequency curve and its
statistical interpretation. The solution techniques of certain problems based