Premium
On the impact of gaps on trend detection in extreme streamflow time series
Author(s) -
Slater Louise,
Villarini Gabriele
Publication year - 2017
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/joc.4954
Subject(s) - series (stratigraphy) , allowance (engineering) , streamflow , maxima , climatology , time series , trend analysis , environmental science , magnitude (astronomy) , statistics , mathematics , geography , geology , drainage basin , physics , paleontology , art , cartography , astronomy , performance art , art history , mechanical engineering , engineering
Streamflow time series often contain gaps of varying length and location. However, the influence of these gaps on trend detection is poorly understood and cannot be estimated a priori in trend detection studies. We simulated the effects of varying gap size (1, 2, 5, and 10 years) and location (one quarter, one third, and half of the way) on the detection rate of significant monotonic trends in annual maxima and peaks‐over‐threshold, based on the most commonly‐used trend tests in time series of varying length (from 15 to 150 years) and trend magnitude ( β 1 ). Results show that, in comparison with the complete time series, the loss in trend detection rate tends to grow with (1) increasing gap size, (2) increasing gap distance from the middle of the time series, (3) decreasing β 1 slope, and (4) decreasing time series length. Based on these findings, we provide objective recommendations and cautionary remarks for maximal gap allowance in trend detection in extreme streamflow time series.