Title: Investigating climate change using AK stationarity test in the Lake Urmia basin

Authors: Mojtaba Moravej

Addresses: Department of Irrigation and Reclamation Engineering, Faculty of Agricultural Engineering and Technology, College of Agriculture and Natural Resources, University of Tehran, Karaj, Tehran, Iran

Abstract: Augmented Dickey-Fuller (ADF) and Kwaitowski-Phillips-Schmidt-Shin (KPSS) test are the most employed tests to test stationarity. Because of the different null hypothesis of these two tests, their power varies in different stationarity cases. The problem arises when the results of ADF and KPSS test are against each other for the same time series. In the current study, the power of these two statistical tests is investigated with a set of artificial hydrological time series. Results indicate that both ADF and KPSS test are unable to detect second order stationarity and KPSS is sensitive to outliers. Then, sequent combination of ADF and KPSS introduced as AK test. The AK test is able to identify first order stationarity, first order non-stationarity (unit root) and long memory. Although some non-stationaries are observed in precipitation, evaporation and stream flow time series in eight main sub-basins of Lake Urmia, but there are no strong evidence to suggest climate change incidence.

Keywords: AK test; ADF test; KPSS test; stationarity testing; long memory; Lake Urmia; climate change; hydrological time series analysis; hydrology; Iran; precipitation; evaporation; stream flow.

DOI: 10.1504/IJHST.2016.079349

International Journal of Hydrology Science and Technology, 2016 Vol.6 No.4, pp.382 - 407

Received: 23 Aug 2015
Accepted: 02 Dec 2015
Published online: 27 Sep 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article