Lanzante, J. R., 2016:
Distributional testing employing the Kuiper’s and Kolmogorov-Smirnov tests
accounting for temporal coherence.
To be submitted .
It is shown that the Kuiper’s, and the closely related Kolmogorov-Smirnov tests of
distributional goodness of fit can result in a serious overestimate of the
significance of the results if temporal coherence, commonly found in daily weather
and climate time series, is ignored. Alternately, employing the commonly used
“effective sample size”, based on the lag-1 autocorrelation, can result in a serious
underestimation of the significance. A practical method to overcome this quandary