ABSTRACT
One of the hydrologic series that has received considerable attention is precipitation at LaPorte, Indiana. LaPorte is east of the Chicago–Gary complex and the predominant wind direction at Chicago is westerly. This chapter discusses effects of a changing mean on the characteristics of independent random variables and describes the problem of characterizing time series with changes in mean values. Thresholds in hydrologic data may be approximated by jumps or steps in the data sequences. These step changes and times of their occurrence can have considerable impact on the statistics of the data, and on the development of hydrologic models. Single or multiple steps alter the elementary statistics, such as the mean, variance, skewness coefficients, and kurtosis. Many hydrologic series, such as annual rainfall or runoff data, are nonperiodic but correlated, and shift considerably from one level to another, so that testing the significance of the changes is important.
