Drought is a costly natural hazard affecting socio-economic activity and agricultural livelihoods, as well as adversely impacting public health and threatening the sustainability of many natural environments. This study was carried out to characterize the temporal and spatial characteristics of meteorological drought in the upper Blue Nile basin to provide a framework for sustainable water resources management. Analysis of historical droughts was undertaken by converting observed monthly precipitation records (1960–2008), for 22 meteorological stations, to the standardized precipitation index (SPI). The SPI was computed at multiple time steps and the Mann–Kendall test was applied on monthly SPI time series for trend detection, and finally severity areal extent frequency (SAF) curves were developed to assess the recurrence pattern of drought severity. Several drought events were observed during the long rainy season and also the short rainy season, and the drought extent and influence were very severe in 1965 and the 1980s. Trend analysis showed statistically insignificant trends in SPI time series, and SAF curves indicated that droughts with a short return period and high degree will cover only small areas of the basin, while only a near-normal drought with a long return period may spread over the whole region.
INTRODUCTION
Drought is one of the most damaging climate-related hazards, and has been a major concern of mankind for centuries (Sivakumar et al. 2005; IPCC 2012). It can affect large areas and may have serious impacts which depend on the severity, duration, and spatial extent of the precipitation deficit (European Commission 2007). Drought impacts are non-structural and spread over a larger geographical area than the damages that result from other natural hazards (Belayneh 2012). Quantifying the impacts and providing disaster relief are far more difficult tasks for drought than for other natural hazards since these impacts can filter through environmentally, socially and economically for months, years and even decades (Wilhite 2005; Bordi & Sutera 2007; Edossa et al. 2010). It is difficult to determine the onset and end of drought, therefore scientists and policy makers often disagree on the basis for declaring an end to drought (Tiwari et al. 2007; Wilhite et al. 2014). Different types of drought are identified in a practical sense, as drought differs between regions and its impacts vary significantly because of differences in economic, social, and environmental characteristics (Svoboda 2000; WMO 2006). All types of drought originate from a deficiency of precipitation, although other factors such as high winds, high temperatures, and low relative humidity may exacerbate the drought's severity (Mishra & Singh 2010). The main purpose of any drought monitoring system is to identify various drought indices to provide information to resources managers and system operators (Keyantash & Dracup 2002; Valipour 2012). Several drought indices are used in drought assessment and monitoring based on rainfall data (NDMC 2006; Hayes 2013; Wilhite et al. 2014). The most commonly used drought indices as reported in Wilhite (2005) and Valipour (2013a, 2013b) are the Palmer drought severity index (PDSI) (Palmer 1965), crop moisture index, standardized precipitation index (SPI) (McKee et al. 1993) and surface water supply index (Shafer & Dezman 1982). In this study, drought vulnerability in the upper Blue Nile river basin was investigated using the SPI, a meteorological index developed by McKee et al. (1993). The advantage of this index is that it identifies emerging drought months sooner than the PDSI, and can be computed at various timescales (NDMC 2006). This study aimed to address multi-nature aspects of drought with its several features; the frequency of drought occurrence and its spatial distribution to identify drought-prone areas; and drought vulnerability at multiple time steps, such as 3, 6, 9, 12, and 24 months. The severity and the frequency of the monthly drought periods were analyzed next and finally, the areal extent, the cumulative severity and the duration of common drought periods were examined. To the best of our knowledge, the issue of water resources management in the context of drought so far has not been addressed in the eastern Nile basin in general as well as in Egypt in particular. Results presented in this paper are part of an ongoing effort to develop a new integrated framework for water resources management in the context of drought. Our approach consists of the integration of qualitative modeling for drought monitoring, drought forecasting (Khadr 2015), a drought management plan, and a simulation model for the operation of the Aswan High dam in Egypt (Khadr & Schlenkhoff 2014). It is hoped that the proposed approach and our findings obtained in this study are useful for further research in the area of water resources management.
METHOD AND MATERIALS
Study region and data set
Methodology
The methodology presented in this paper and applied to UBNRB consisted of calculation of the SPI values for multiple timescales, analysis of the temporal characteristics, the spatial interpolation of SPI gauge data, the generation of gridded SPI data, and trend analysis of SPI time series, followed by the analysis of spatial characteristics of droughts in the UBNRB for monthly and mean annual droughts and multiple timescales.
The SPI
Drought intensity classification in various categories with different values of SPI is given in Table 1. Positive SPI values indicate greater than median precipitation, and negative values indicate less than median precipitation. In this study, the SPI program developed by the author (Khadr 2011) is used to compute the time series of drought indices (SPI) for each station in the basin and for each month of the year at different timescales (3, 6, 9, 12 and 24 months).
SPI . | Classification . |
---|---|
>2 | Extremely wet |
1.5 to 1.99 | Very wet |
1 to 1.49 | Moderately wet |
0.99 to –0.99 | Near normal |
–1 to –1.49 | Moderately dry |
1.5 to –1.99 | Severely dry |
–2 and less | Extremely dry |
SPI . | Classification . |
---|---|
>2 | Extremely wet |
1.5 to 1.99 | Very wet |
1 to 1.49 | Moderately wet |
0.99 to –0.99 | Near normal |
–1 to –1.49 | Moderately dry |
1.5 to –1.99 | Severely dry |
–2 and less | Extremely dry |
Mann-Kendall trend test
is the value read from a standard normal distribution table, with α being the significance level of the test. At the 99% significance level, the null hypothesis of no trend is rejected if ; at the 95% significance level, the null hypothesis of no trend is rejected if ; and at 90% significance level, the null hypothesis of no trend is rejected if .
Temporal and spatial analysis of drought in UBNRB
The temporal and spatial characteristics of drought in UBNRB were assessed by analyzing the SPI values. Firstly, the temporal variation of SPI was assessed based on the SPI characteristics presented in Table 1, using the calculated SPI from the time series of the monthly precipitation. The magnitude of a considered drought event (DM), which is the summation of negative SPI values, was also calculated at all stations. Secondly, the computed SPI values based on various timescales within the study period (January 1960–December 2008) were entered into a geographical information system (GIS) database. Then, the spatial characteristics of drought during these severe dry periods were analyzed and visualized within the GIS. The spatial distribution of the SPI was determined through spatial interpolation techniques, using the inverse distance weighting (IDW) method to analyze the meteorological drought with due emphasis to ungauged catchments. The IDW, which is a deterministic method for multivariate interpolation, relies on the theory that the unknown value of a point is more influenced by closer points than by points further away (Khadr & Schlenkhoff 2014a). The SPI characteristics presented in Table 1 were used to generate drought maps for the UBNRB using GIS.
Drought severity areal extent frequency (SAF) curves were developed for the UBNRB using the computed gridded monthly SPI values for various timescales. The probability of annual drought occurrence for each year and in each grid was estimated by dividing the number of months that have a negative SPI value for the particular timescale by 12, then the annual weighted cumulative drought severity in each grid was estimated by multiplying the annual sum of the negative SPI values for a particular timescale by the probability of drought occurrence for each year. The map produced in the GIS was then used to obtain the drought severity associated with the areal extent and to perform frequency analysis for each drought areal extent percentage to associate the drought severity with return periods, considering an adequate probability distribution. In the end, the families of SAF curves corresponding to different return periods were constructed by plotting area (in percent of whole region) versus average SPI in each category.
RESULTS AND DISCUSSION
Temporal characteristics of the drought in UBNRB
As one can see in Figure 4, a drought's characteristics change with time and at longer timescales droughts become less frequent but their duration increases. The time series of 3-month SPI (Figure 4(a)) shows that the station experienced frequent moderate, severe and extreme droughts. Analysis of the relative frequency of occurrence (Table 2) of the annual minimum monthly SPI, during rainy seasons as mentioned earlier, shows that May is the month during which the 3-month SPI most frequently takes the annual minimum value (13.7%), and it is followed by July (8.22%). The annual minimum SPI values for the 6-, 9-, 12-, and 24-month time series for the period of analysis were mainly observed in June, May, August and June, respectively (Table 2).
Month . | SPI 3 . | SPI 6 . | SPI 9 . | SPI 12 . | SPI 24 . |
---|---|---|---|---|---|
January | 9.59 | 7.35 | 4.84 | 6.67 | 7.50 |
February | 0.00 | 8.82 | 6.45 | 5.00 | 7.50 |
March | 10.96 | 10.29 | 9.68 | 3.33 | 7.50 |
April | 9.59 | 7.35 | 6.45 | 6.67 | 5.00 |
May | 13.70 | 10.29 | 12.90 | 8.33 | 7.50 |
June | 6.85 | 11.76 | 9.68 | 13.33 | 12.50 |
July | 8.22 | 8.82 | 9.68 | 10.00 | 10.00 |
August | 6.85 | 10.29 | 11.29 | 13.33 | 10.00 |
September | 6.85 | 5.88 | 6.45 | 10.00 | 10.00 |
October | 6.85 | 5.88 | 8.06 | 6.67 | 7.50 |
November | 6.85 | 5.88 | 6.45 | 8.33 | 7.50 |
December | 13.70 | 7.35 | 8.06 | 8.33 | 7.50 |
Month . | SPI 3 . | SPI 6 . | SPI 9 . | SPI 12 . | SPI 24 . |
---|---|---|---|---|---|
January | 9.59 | 7.35 | 4.84 | 6.67 | 7.50 |
February | 0.00 | 8.82 | 6.45 | 5.00 | 7.50 |
March | 10.96 | 10.29 | 9.68 | 3.33 | 7.50 |
April | 9.59 | 7.35 | 6.45 | 6.67 | 5.00 |
May | 13.70 | 10.29 | 12.90 | 8.33 | 7.50 |
June | 6.85 | 11.76 | 9.68 | 13.33 | 12.50 |
July | 8.22 | 8.82 | 9.68 | 10.00 | 10.00 |
August | 6.85 | 10.29 | 11.29 | 13.33 | 10.00 |
September | 6.85 | 5.88 | 6.45 | 10.00 | 10.00 |
October | 6.85 | 5.88 | 8.06 | 6.67 | 7.50 |
November | 6.85 | 5.88 | 6.45 | 8.33 | 7.50 |
December | 13.70 | 7.35 | 8.06 | 8.33 | 7.50 |
Trend analysis was performed using the MK test of the null hypothesis of trend absence in the SPI time series, against the alternative of trend. The result of the test is returned in hm = 0, for all SPI series, and indicates a failure to reject the null hypothesis at the 90% significance level. Moreover, Table 3 indicates a failure to reject the null hypothesis at the 90% significance level since for SPI3, SPI6, SPI9, SPI12 and SPI24, which means that all investigated trends are statistically insignificant.
SPI time series . | SPI 3 . | SPI 6 . | SPI 9 . | SPI 12 . | SPI 24 . |
---|---|---|---|---|---|
hm | 0 | 0 | 0 | 0 | 0 |
P value | 0.5954 | 0.9505 | 0.7478 | 0.9246 | 0.1164 |
ZMK | 0.173 | 0.402 | 1.127 | 0.896 | 0.249 |
Type of trend | Insignificant | Insignificant | Insignificant | Insignificant | Insignificant |
SPI time series . | SPI 3 . | SPI 6 . | SPI 9 . | SPI 12 . | SPI 24 . |
---|---|---|---|---|---|
hm | 0 | 0 | 0 | 0 | 0 |
P value | 0.5954 | 0.9505 | 0.7478 | 0.9246 | 0.1164 |
ZMK | 0.173 | 0.402 | 1.127 | 0.896 | 0.249 |
Type of trend | Insignificant | Insignificant | Insignificant | Insignificant | Insignificant |
Spatial characteristics of drought in UBNRB
CONCLUSION
In this study, a framework of methodologies was presented for the analysis of the temporal and spatial characteristics of the meteorological drought in the upper Blue Nile basin. The SPI was computed at various timescales using spatially distributed precipitation records from 22 meteorological stations. The temporal and spatial drought analyses indicate that the upper Blue Nile Basin received quite frequent moderate to severe droughts with an insignificant trend in the presented SPI series. The region has experienced prolonged and severe droughts during the periods of 1961, 1965 and 1980–1987. In particular, the persistent and prolonged droughts of 1965 and 1982 seriously affected urban water supply and agricultural irrigation. For the 1965 drought, the most affected areas were the area in the center of the UBNRB and with a direction of north to east and southeast, whereas the western and northern areas of UBNRB were mostly affected by the prolonged drought during 1984, which was more severe than the 1965 event. The constructed SAF curves indicated that drought by short return period and high degree cover only small areas of the basin, while only a near-normal drought with a long return period may spread over the whole region. It is hoped that this study will provide useful guidance in drought mitigation and adaptation planning in the study area.
ACKNOWLEDGEMENTS
The author thanks the Eastern Nile Technical Regional Office (ENTRO) for kindly providing data which was used for the analysis in this paper.