Hydrological trend analysis with innovative and over-whitening procedures

Different statistical methodologies can be employed to identify possible trend components in any hydro-meteorological time series. A pre-whitening (P-W) procedure has been suggested to reduce the serial correlation effect on Mann-Kendall (M-K) trend analysis. In this paper, instead of P-W, an over-w...

Full description

Saved in:
Bibliographic Details
Published inHydrological sciences journal Vol. 62; no. 2; pp. 294 - 305
Main Author Sen, Zekai
Format Journal Article
LanguageEnglish
French
Published Abingdon Taylor & Francis 25.01.2017
Taylor & Francis Ltd
Subjects
Arid regions
Arid zones
Carbon dioxide
Carbon dioxide concentration
Components
Hydrologic analysis
Hydrology
Hydrometeorology
innovative template
over-whitening
Periodic variations
Periodicity
Procedures
Rain
Rainfall
Records
Relative humidity
serial correlation
simulation
Statistical methods
Stream discharge
Stream flow
Time series
trend
Trend analysis
Trends
Tropical climate
Water runoff
Online AccessGet full text

Cover

More Information
Summary:Different statistical methodologies can be employed to identify possible trend components in any hydro-meteorological time series. A pre-whitening (P-W) procedure has been suggested to reduce the serial correlation effect on Mann-Kendall (M-K) trend analysis. In this paper, instead of P-W, an over-whitening (O-W) procedure is suggested, which generates serially independent series with the same trend slope value. Analytically necessary formulations for O-W are presented with a non-parametric but simple innovative trend assessment procedure, which are supported by extensive simulation studies. The applications of the methodology are presented for eight factual time series records from tropical, temperate and arid regions including temperature, rainfall, streamflow, relative humidity and CO 2 concentrations for different short and long durations. Relative humidity and CO 2 records are monthly time series and, hence, there are trend and periodicity components. It is noticed in all cases that the natural trends remain as they were after the O-W procedure, thus providing an opportunity to determine reliably the trends embedded even in the serially dependent series. The O-W procedure is applicable even in the cases of periodicity in the original records.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0262-6667
2150-3435
DOI:10.1080/02626667.2016.1222533