Risultato della ricerca: Otherpeer review


In the last decades the growing concerns about the existence of global climatic changes push manyresearchers to use different trend test in order to identify whether monotonic trends exist inhydroclimatological time series such as temperature, precipitation, and streamflow. Unfortunately, these timeseries often suffer from missing data values mainly due to malfunctioning of gauge for specific time periods.Starting from this premise, the main target of our work is to investigate the effect of data gap in a time serieson the results provided by the most used trend test: the nonparametric Mann–Kendall statistical test. Firstly,different synthetic time series characterized by different size, trend and statistical significance of the trend,i.e. p-value, have been generated by superimposition of a stochastic component on a trend component. UsingMonte Carlo simulation, data have been randomly removed from the time series in percentage varying from1% to 20%; then, the Mann-Kendall trend test has been applied to the obtained incomplete time series. Thecomparison between the results of Mann-Kendall trend test of complete and incomplete time series, providesa quantitative assessment of the influence of data gap as function of the percentage of data gap and of thesample size of the complete time series. In particular, these results indicate that when the p-value of completetime series approaches to the test acceptability threshold (significance level of the test), the probability ofidentify the presence of a trend when it does not exist (or vice-versa) increases dramatically. Finally, theinfluence of the shape of the data gap, whether continuous or random, has been investigated as well, findingout that it does not influence in a significant way the results of Mann-Kendall trend test.
Lingua originaleEnglish
Numero di pagine0
Stato di pubblicazionePublished - 2012

Fingerprint Entra nei temi di ricerca di 'EFFECTS OF DATA GAP ON THE CAPABILITY TO DETECT TREND IN HYDROCLIMATOLOGICAL TIME SERIES'. Insieme formano una fingerprint unica.

Cita questo