Analysis of the Climatological Drought Trend Variations Using Mann-Kendall, Sen and Pettitt Tests in Isfahan Province

Abstract

 

Global climate has changed significantly in the last hundred years. The climate change already having significant and costly effects on temperature and precipitation. Precipitation is particularly important, because changes in precipitation patterns may lead to floods or droughts in different areas. Also, climate change and agriculture are interrelated processes. Since lack of enough rainfall is the main limiting factor in agriculture in most tropical areas, monitoring precipitation variability for periods of months to detect possible trends is of great interest.

Furthermore, trend analysis of precipitation and drought plays a significant role in the future development and management of water resources. It is the main purpose of this paper to analyse meteorological droughts and trends of monthly rainfall of 10 synoptic stations located in Isfahan province, namely Ardestan, East Isfahan, Golpaygan, Isfahan, Kabootar Abad, Kashan, Khorobianbanak, Naein, Natanz, Shahreza, during 1990-2017. In order to investigate the variability of the Standardized Precipitation Index (SPI) for different time intervals, the SPI values of all these stations were calculated, at multiple 1, 3, 6, 9, 12, 24 and 48 monthly scales. Then, trends of these monthly SPI results were analysed by applying the non-parametric Mann–Kendall at the 5% significance level for all stations. The magnitudes of trends were computed and tested using the Sen’s slope estimator method.

    At this point, it is worth mentioning that the purpose of the Mann-Kendall test is   statistically assessing the existence of monotonic upward or downward trend of the variable of interest over time. A monotonic upward (downward) trend means that the variable consistently increases (decreases) through time, but this trend may or may not be linear. For detecting linear trend, the slope is usually estimated by computing the least squares estimate using linear regression. However, this is only valid when there is no serial correlation. Moreover, the least squares method is very sensitive to outliers. Therefore, for analyzing SPI Indices for detecting possible trend the non-parametric Sen’s slope estimator was used, which is a robust method against these two least square assumptions.

Further trend analysis was   applied to detect possible single change-point using Pettitt’s test.  Using this test, we were able to test the significant abrupt change in the level of time series, at

let say the 5% significance level, for all stations and different SPI monthly scales. All statistical analysis has been done using R statistical software and the facilities of its packages.

The results of applying Mann–Kendall and Sen’s slope tests based on SPI Index for 9, 12, 18, 24 and 48-month time periods, shows drought trend is significantly increasing for all stations out of Ardestan, Esfahan and Shahreza stations. In Ardestan station, the drought trend is significantly decreasing for 9, 12, 18, 24- and 48-month time periods and in Isahan station, the drought trend is significantly decreasing for only 48-month time period, and in Shahreza statition, the drought trend is significantly increasingonly for only 18-month time period.

Despite all stations, the drought trend for one-month time period, is significantly increasing just for Naein station.

In addition, applying Mann–Kendall test on monthly rainfall for all station shows downward but not significant trend.

Finally, applying Pettitt’s change point test based on SPI Index for 9, 12, 18, 24 and 48-month time periods indicates the existence of a significant change point. For same periods we observe no change point for the monthly rainfall  in all stations.

In summation, considering the SPI drought index, about 59% of  all stations show significant downward trend bases on Mann-Kendall test and 60% of  all stations show significant slope  based on Sen's slope test and 75% of  all stations show significant change point based on Pettitt's test.In general, for drought analysis using different time periods for the SPI index, in a short time period. (such as 6 months) drought is more frequent but shorter, and as the period increases the duration of drought also increases but frequency decreases. All together, we are facing a water crisis in Isfahan province and we must manage water demand very urgently.

 

Key words: Drought, Isfahan Province, Mann-Kendall Test, Pettitt’s Change point Test, Sen’s Slope Test, Standard Precipitation Index, Trend Analysis.