Climate Variability and Change—Hydrological Impacts (Proceedings of the Fifth FRIEND World Conference held at Havana, Cuba, November 2006), IAHS Publ. 308, 2006, 538–550.

---

Methodologies for trend detection

ZBIGNIEW W. KUNDZEWICZ^1,2 & MACIEJ RADZIEJEWSKI^1,3

1 Research Centre for Agricultural and Forest Environment, Polish Academy of Sciences, Poznań, Poland

zkundze@man.poznan.pl; zbyszek@pik-potsdam.de

2 Potsdam Institute for Climate Impact Research, D-14412 Potsdam, Germany

3 Faculty of Mathematics and Computer Science, Adam Mickiewicz University, Poznań, Poland

Abstract Methodologies for detection of change in river flow data are briefly reviewed. After discussion of the issue of data for change detection, the basics of statistical testing for trend detection are presented, including such concepts as the null hypothesis, the test statistic, the significance level and the trend index. Test assumptions are thoroughly discussed. A review of practical tests (parametric, distribution-free i.e. non-parametric, and resampling: permutation and bootstrap approaches) applicable for different properties of data, is presented. The case where the records are not independent and are not normally distributed is also included. A few special problems, such as extremes and spatial data (with correlation between gauges), and detectability of change are also discussed. Assessment of a few relevant recent publications is offered.

Key words change detection; resampling; river flow data; statistical testing; trend

---