Abstract
The equation of multivariate maximum entropy distribution is evolved for the hydrologic frequency analysis, and its characteristics, which include the applicability to hydrologic data, are investigated in detail.
It is proved that the maximum entropy distribution is based on the same level as Pearson's system of frequency-curves because it derives some distributions which are wellknown in statistics. The general equation of multivariate maximum entropy distribution with the restriction of arbitrary function gr (·) is obtained by Lagrange's method of multipliers. In the case of the moment problem, the parameters of the distribution are identified by the iteration method. When three-variate normal distribution and hydrologic data are adopted as the populations, the conformity of multivariate maximum entropy distribution is investigated in connection with low-order moments, and it is proved that 3M (2, 2, 2, 2) is in a good agreement with actual data.
