Current Status of Time Series Analysis in Hydrological Sciences - Hydrologic Time Series Analysis: Theory and Practice - page 115

suggested that the series may be considered as realization of autoregressive
processes of the first order. Superposed to the red-noise spectra there were
waves with periods of 2.4 and 8 years in the case of Danube stations and with
periods of 5.7 and 6.7 years in the case of southern rivers.
Lye and Lin (1994) analyzed the peak flow series from 90 Canadian
rivers to examine stationarity. The results suggested that although short-term
dependence is practically absent for most peak flow series, significant long-
term dependence is present for a large number of peak flow series tested. It
was demonstrated that the most statistical tests of independence or stationarity
are designed to detect only short-term serial correlation. They were found
insensitive to the long-term serial correlation structure of flood records, which
can be far more important.
Lin and Lye (1994) investigated the suitability of Sen's method (Gilbert,
1987) for modelling hydrologic time series, especially the generation of
synthetic flow series. It was found that several problems exist with the proposed
method. They are: too many parameters in the model, difficulties in modelling
skewed series, and finding a suitable stochastic model for the residuals between
the original and fitted cumulative departure curves. On the other hand, it was
found that the Sen's method is effective in preserving the Hurst phenomenon
and is especially suited for modelling time series with a relatively high Hurst
coefficient but a low lag-one serial correlation coefficient. It was also
demonstrated that after some modifications in the Sen's method, some of the
problems could be overcome to a certain extent.
Knapp (1994) analyzed the long-term streamflow records of Upper
Mississippi River Basin to determine trends in streamflows and flooding.
Trends in average flow and flooding were found strongly correlated to the
coincident increases in average annual precipitation. For many portions of the
watershed, the precipitation and streamflows over the last three decades were
found higher than any earlier period on record. Outside of the dominant
influence of climate variation, only one major change in Mississippi River
flood discharges was observed. Further, the flood control reservoirs in the
Missouri River watershed appeared to produce a 10% reduction in the average
flood peak and flood volume for the Mississippi River at St. Louis, Missouri.
Rao (1995) analyzed the rainfall (1901-1990) and streamflow series (1926-
1980) of Mahanadi River, India in relation to the climatic change in the river
basin. The analysis of trends in the runoff from the upper catchment suggested
a steady decrease in the river flows at Hirakud and Naraj gauging stations
during the 55-year period of the study. In order to increase confidence in this
result, the moisture indices for the catchment were computed and examined,
which also indicated a clear declining trend during the period 1901-80. It was
concluded that the climate warming that occurred over the basin resulted in a
gradual decrease of river flows of the upper catchment as well as of the entire
basin during the period 1926 to 1980.