1520 JOURNAL OF HYDROMETEOROLOGY VOLUME 13 Understanding the Changing Characteristics of Droughts in Sudan and the Corresponding Components of the Hydrologic Cycle ZENGXIN ZHANG Jiangsu Key Laboratory of Forestry Ecological Engineering, Nanjing Forestry University, Nanjing, China, and Department of Geosciences, University of Oslo, Oslo, Norway CHONG-YU XU Department of Geosciences, University of Oslo, Oslo, Norway BIN YONG State Key Laboratory of Hydrology-Water Resources and Hydraulics Engineering, Hohai University, Nanjing, China JUNJUN HU School of Computer Science, University of Oklahoma, Norman, Oklahoma ZHONGHUA SUN Network and Information Center, Changjiang Water Resources Commission, Wuhan, China (Manuscript received 24 August 2011, in final form 27 April 2012) ABSTRACT Droughts are becoming the most expensive natural disasters in former Sudan and have exerted serious impacts on local economic development and ecological environment. The purpose of this paper is to improve understanding of the temporal and spatial variations of droughts by using the Standard Precipitation Index (SPI) and to discuss their relevance to the changes of hydrological variables in Sudan. The analysis results show that 1) droughts start in the late 1960s in Sudan and severe droughts occur during the 1980s in different regions of Sudan—the annual precipitation and soil moisture also reveal the evidence that the droughts prevail since the late 1960s; 2) the greater negative soil moistures anomalies are found in central and southern Sudan during the rainy seasons while greater negative anomalies of precipitation occur only in central Sudan compared between 1969–2009 and 1948–68; 3) the precipitation recycling ratio averaged over 1948–2009 decreases from south to north and the percentage of local actual evapotranspiration to local precipitation in dry conditions is greater than that in wet conditions; and 4) the highest (second highest) correlations appear between soil moisture and precipitation (actual evapotranspiration) and the significant decreases in annual soil moisture are associated with the decrease of annual precipitation and the increase of annual temperature. This suggests that continuous droughts in Sudan are caused jointly by the decrease of precipitation and the increase of temperature in the region. 1. Introduction Droughts may be one of the world’s most costly natural disasters, occurring frequently in many countries. A drought is an extended period of months or years Corresponding author address: Zengxin Zhang, Ph.D., Associate Professor, Jiangsu Key Laboratory of Forestry Ecological Engineering, Nanjing Forestry University, Nanjing 210037, China. E-mail: zhangzengxin77@yahoo.com.cn DOI: 10.1175/JHM-D-11-0109.1 Ó 2012 American Meteorological Society when a region notes a deficiency in its water supply. The severity of the drought depends upon the degree of moisture deficiency, duration, and size of the affected area. It can have a substantial impact on the ecosystem and agriculture of the affected region, which can cause significant damage and harm to the local economy. During the drought, sparse vegetation and dry soil limit evapotranspiration. Less than usual amounts of water vapor in the atmospheric boundary layer reduce the availability of water vapor and potential energy, though OCTOBER 2012 ZHANG ET AL. not sufficient ingredients, for the generation of convective rainfall (Shukla and Mintz 1982; Koster et al. 2004). Being often cumulated slowly over a considerable period of time, it is difficult to precisely determine the onset and end of a drought event. To monitor droughts and wet spells and study their variability, it is necessary to devise numerous specialized indices that combine available data such as precipitation and temperature (Heim 2000; Trenberth et al. 2004; Su and Wang 2007; Kalamaras et al. 2010; Yang et al. 2012). In recent years various indices have been proposed to detect and monitor droughts and have been used in modeling droughts as well as stochastic and water-balance simulations (Palmer 1965; Lana et al. 1998; Mishra et al. 2007). The standardized precipitation index (SPI) is one of the indices commonly used in recent decades. The SPI can simulate climatic conditions over a wide spectrum of time scales. Moreover, it is based on precipitation changes alone. Further, Hayes et al. (1999) argued that the SPI detects moisture deficits more rapidly than the Palmer drought severity index (PDSI; Bonaccorso et al. 2003). The SPI attempts to determine the rarity of a drought or an anomalously wet event on a particular time scale for any location that has a precipitation record. A drought event can be decided at a time interval when the SPI value is persistently negative, and vice versa. An accurate quantitative knowledge of the hydrological components of the earth–atmosphere system, on a regional and global basis, is of basic importance in many branches of geophysics (Rasmusson 1968). The locally supplied moisture or upward flux of water vapor can be from evaporation of in situ open water or soil moisture, or from plant transpiration. To maintain rainfall, water vapor must be supplied through the divergence of water vapor from its source to sink regions. Thus, the atmospheric branch of the hydrological cycle constitutes a vital component for understanding the changing features of water resources. However, because of the lack of homogeneous data for hydrological variables (e.g., water vapor, precipitation, and actual evapotranspiration), the major objectives of numerous previous studies were mostly aimed at documenting the timemean atmospheric hydrological cycle and its seasonal variation (Peixoto and Oort 1983, 1992; Chen et al. 1995). For example, Zangvil and Karas (2001) investigated the time-scale relationships among the large-scale atmospheric moisture budget components over the Midwestern United States [35% of the Global Energy and Water Cycle Experiment (GEWEX) Continental-Scale International Project (GCIP) domain] in relation to summer precipitation. Both the measurements and numerical experiments in hydroclimatology have confirmed positive and negative land surface–climate feedbacks, of 1521 which moisture recycling is a prominent phenomenon at continental scales. Raddatz (2005) investigated the contribution of land surface evapotranspiration to the atmospheric water balance for the agricultural region of the Canadian prairies by estimating the recycling ratios, including the moistening and precipitation efficiencies, for drought areas for the summers of 1997–2003. The former Republic of Sudan was Africa’s largest country with over 90% of its people living below the poverty line. Southern Sudan was split from the north and created the world’s newest nation in July 2011. This study was completed before the separation of the Sudan and our study area covers the Sudan and southern Sudan; in the rest of the paper we call the study area Sudan for short. Frequent droughts and environmental degradation are the major obstacles to livelihood security and food self-reliance in Sudan. Over 80% of Sudan’s population lives in rural areas, depending on agriculture and livestock to make a living. It is becoming a phenomenon in Sudan that 1 in every 5 years is dry. When the droughts come, agriculture collapses, people migrate, and those who stay face conflict over food and water supplies. Since the infamous famine of 1984/85, Sudan has suffered severe droughts in 1989, 1990, 1997, and 2000. Each drought brought crop failure, loss of livestock, and loss of pastureland. In 1984, a crop failure and spread of waterborne diseases caused by drought in Sudan took the lives of 55 000 people, which weakened the socioeconomic capabilities of the nomadic tribes (Osman and Shamseldin 2002). The droughts and famine might be one of the most serious threats to Sudan. Comprehensive analysis and reviews of rainfall trends and variability in Africa, including the Sahel region and Sudan, had been carried out by many researchers and most reputable works include those of Hulme and his coauthors (e.g., Trilsbach and Hulme 1984; Hulme 1987; Hulme and Tosdevin 1989; Hulme 1990; Walsh et al. 1988). Walsh et al. (1988) reported that declining rainfall in semiarid Sudan since 1965 has continued and intensified in the 1980s. Hulme (1990) pointed out that rainfall depletion has been most severe in semiarid central Sudan between 1921–50 and 1956–85. The length of the wet season has contracted, and rainfall zones have migrated southward (Zhang et al. 2011). The temperature is rising and rainfall is declining for the past several decades, which might be the main cause of the drought in Sudan (e.g., Alvi 1994; Janowiak 1988; Nicholson et al. 2000). The decreasing precipitation might be related to the atmospheric moisture transport. Much research work has been performed regarding the moisture variabilities over Africa (e.g., Cadet and Nnoli 1987; Fontaine et al. 1522 JOURNAL OF HYDROMETEOROLOGY 2003; Osman and Hastenrath 1969). They pointed out that at more local scales moisture advections and convergences are also significantly associated with the observed Sudan–Sahel rainfall and in wet (dry) situations, with a clear dominance of westerly (easterly) anomalies in the moisture flux south of 158N. Zhang et al. (2011) revealed that the precipitation of the main rain season (i.e., July, August, and September) and annual total precipitation in the central part of Sudan decreased significantly during 1948–2005 and the decreasing precipitation in Sudan was associated with the weakening African summer monsoon. The summer moisture flux over Sudan tended to be decreasing after the late 1960s, which decreased the northward propagation of moisture flux in North Africa. The atmospheric branch of the hydrological cycle reflects the natural variability of weather and climate at the regional and global scales. However, it is not obvious how these changes will be reflected in terms of droughts. Precipitation recycling plays a key role in the hydrological process and the precipitation recycling ratio is a diagnostic measure for interactions between land surface hydrology and regional climate. The analysis of atmospheric hydrology recycling usually relies heavily on the National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) or European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis data; however, the precipitation, actual evapotranspiration, and soil moisture data are derived from unconstrained reanalysis systems. In other words, observations of these quantities are not assimilated into the reanalysis system, so the assimilating model is free to produce them, typically through parameterizations. So the reanalysis precipitation and actual evapotranspiration are highly model dependent. Roads et al. (2002) pointed out that maintaining the NCEP/ Department of Energy Global Reanalysis 2 (NCEP-2) close to observations requires some nudging to the shortrange model forecast, and this nudging is an important component of analysis budgets to assess global and regional water and energy budgets. Trenberth et al. (2011) analyzed the water and energy cycles in the last version of the NCAR climate model [Community Climate System Model, version 4 (CCSM4)] and found that the moisture transport from ocean to land should all be identical but are not close in most reanalyses, and they thought that major improvements are needed in model treatment and assimilation of moisture, and surface fluxes from reanalyses should only be used with great caution. Therefore, it is not unexpected that the reanalysis precipitation and actual evapotranspiration exhibit some deficiencies and that this field does not compare as well with observations as other reanalysis VOLUME 13 fields such as heights, winds, and temperatures that are assimilated directly into the model. This poses serious risks to conclusions drawn from analyzing this type of data (Trenberth and Guillemot 1998). Even in more modern reanalysis systems designed specifically for hydroclimate research (i.e., North American Regional Reanalysis) the terrestrial water budgets are problematic (Nigam and Ruiz-Barradas 2006; Weaver et al. 2009). In this research, we only chose the wind fields, height fields, and humidity fields to compute the atmospheric moisture content and precipitation recycling ratio, while other variables—such as actual evapotranspiration, soil moisture, and temperature data—are taken from the Climate Prediction Center (CPC) and the Climatic Research Unit (CRU). To understand, and hopefully to be able to predict, the impact of these changes on the droughts in Sudan, we need to understand the relationship between the droughts and the hydrologic variables in Sudan. The specific objectives of this study are 1) to analyze the changing feature of the drought and its involvement in the atmospheric branch of the water cycle in Sudan, 2) to improve our understanding on the hydrologic processes in the land surface branch of the water cycle in Sudan, and 3) to explore the relationship between the droughts and their possible cause based on the atmospheric branch of the hydrologic cycle. 2. Study area and data Former Sudan is a vast country with an area of about 2.5 million km2 and hosts an estimated population of about 41.1 million people. The location of Sudan and South Sudan can be seen in Fig. 1a. Stretching over 188 of latitude and 168 of longitude, the climate ranges from arid in the north to tropical wet and dry in the far southwest. About two-thirds of Sudan lies in dry and semidry regions. The most significant climatic variables are rainfall and the length of the rainy season (Xu et al. 2010). Monthly precipitation data have been selected from the global precipitation reconstruction data (PREC) estimates on a 0.58 3 0.58 latitude–longitude grid over the period 1948–2009 in Sudan. The PREC analyses are derived from gauge observations from over 17 000 stations collected in the Global Historical Climatology Network (GHCN), version 2, and the Climate Anomaly Monitoring System (CAMS) datasets (Chen et al. 2002). The areal mean PREC data and the Sahel precipitation index during 1948–2009 are compared and the results show that the PREC data has a good agreement with the Sahel precipitation index in the long term in the Sahel region (108–208N, 208W–108E). Spatial distributions of mean annual precipitation revealed by the PREC data OCTOBER 2012 1523 ZHANG ET AL. FIG. 1. (a) The location of Sudan and southern Sudan and (b) the distribution of the annual average precipitation based on PREC data during 1948–2009 in Sudan. are then compared with the interpolated observed data of 39 stations for the period 1961–90 and the comparison shows the spatial patterns of PREC data are similar to that of the observed annual mean precipitation (Zhang et al. 2011). Atmospheric data was provided by the NCEP–NCAR reanalysis (R-1) over the period 1948–2009. Wind, temperature, atmospheric pressure, and specific humidity are available on a 2.58 3 2.58 latitude–longitude grid. The soil moisture is selected from a CPC global monthly soil moisture dataset at 0.58 resolution produced by a one-layer ‘‘bucket’’ waterbalance model (Fan and van den Dool 2004). The driving input fields are global monthly precipitation (PREC) and global monthly temperature. The potential evapotranspiration and actual evapotranspiration data are computed by the Thornthwaite monthly water-balance model driven by global monthly precipitation (PREC) and global temperature from the CRU TEM3v dataset at 0.58 resolution. The CRU temperature data has been proven by many researchers, which has a high credibility in many areas (Simmons et al. 2004). Wind components, specific humidity, and covariance, which are needed for atmospheric content and precipitation recycling ratio computations, are provided at eight standard pressure levels (1000, 925, 850, 700, 600, 500, 400, and 300 hPa). Although there are a large number of variables that can be examined to understand the characteristics of the droughts in Sudan, this study focuses on the examination of atmospheric hydrological components (e.g., precipitation, actual evapotranspiration, soil moisture, and precipitation recycling ratio) and their relation with the droughts in Sudan. Better understanding of the relation between the atmospheric hydrological components and the droughts may lead to additional confidence in our ability to predict the droughts. For better understanding the relationship between the features of droughts and the associated hydrological variables in Sudan, we will analyze the variations of precipitation, atmospheric moisture content, potential evapotranspiration, actual evapotranspiration, temperature, soil moisture, and the precipitation recycling ratio. 3. Methods In the actual atmosphere, the atmospheric moisture is very low over 300 hPa, so p 5 300 hPa will be used in the calculation. The moisture content (Q) was calculated based on the following equations (Zhou et al. 1998): Q52 1 g ðp q(P) dP, (1) ps where q is the specific humidity, ps is surface pressure, p is atmospheric pressure at 300 hPa, and g is acceleration of the gravity. The precipitation recycling ratio was computed approximately following the approach of Eltahir and Bras (1994, 1996). The recycling formula is based on the 1524 JOURNAL OF HYDROMETEOROLOGY principle of mass conservation. Two species of water vapor molecules are considered: molecules that are in the atmosphere because of evaporation from within the region considered and molecules that are in the atmosphere as a result of atmospheric transport across the boundary of the region (outside the region). For a finite control volume of the atmosphere located at any point within the region, conservation of mass of the two species requires the following. According to the principle of water balance, water vapor content changing temporally is expressed by the following equations: ›Ww 5 Iw 1 E 2 Ow 2 Pw ›t and ›Wo 5 Io 2 Oo 2 Po , ›t (2a) (2b) where P, W, and E are the regional average precipitation, water vapor content, and actual evapotranspiration, respectively; Iw and Ow are water vapor inflow and outflow fluxes supplied by evapotranspiration within the region; and Io and Oo are water vapor inflow and outflow fluxes supplied by evaporation from outside the region. In deriving the general recycling formula, we make two assumptions. The first assumption states that water vapor is well mixed in the planetary boundary layer (PBL) of the earth’s atmosphere. The PBL is of the order of 1 km deep and contains most of the water vapor in the atmosphere. Observations of the vertical distribution of water vapor and other conserved tracers show a practically uniform distribution through the PBL up to the level where the air from the PBL mixes with the upper air. Based on the above-mentioned assumption, the precipitation recycling ratio, r, can be defined as r5 Pw Ww Ow 5 5 . Pw 1 Po Ww 1 Wo Ow 1 Oo (3) At any location within the region, r estimates the ratio of recycled precipitation to the total precipitation falling at that location. For a large-scale region Iw is very small compared with water vapor fluxes Ow at a long time scale. That is to say we can make the assumption that Iw is zero in large spatial and temporal scale. Equation (2) can be rearranged as follows: Iw 1 E 5 Ow 1 Pw Io 5 Oo 1 Po . and (4a) (4b) Substituting for Ow, Pw, Oo, and Pw from (3) into (4) results in VOLUME 13 Iw 1 E 5 r(Ow 1 Oo ) 1 r(Pw 1 Po ) and Io 5 (1 2 r)(Ow 1 Oo ) 1 (1 2 r)(Pw 1 Po ) . (5a) (5b) Combining Eqs. (5a) and (5b), the average precipitation recycling ratio is deduced as follows: r5 Iw 1 E . Iw 1 E 1 Io (6) Here, r is the average recycling ratio over a region. The U.S. Geological Survey (USGS) Thornthwaite monthly water-balance model is used to compute the potential evapotranspiration and actual evapotranspiration (http://wwwbrr.cr.usgs.gov/projects/SW_MoWS/ software/thorn_s/thorn.shtml). The water-balance model is based on the methodology originally developed by Thornthwaite (Thornthwaite 1948; Mather 1978, 1979; McCabe and Wolock 1999; Wolock and McCabe 1999) and the basic procedure of the model used in this study is similar to that used by Xu and Chen (2005). Inputs to the model are monthly mean temperature, monthly total precipitation, and the latitude of the location of interest. Outputs include monthly potential and actual evapotranspiration, soil moisture storage, snow storage, and runoff. El Haj El Tahir et al. (2012) compared the actual evapotranspiration in Sudan estimated using the remote sensing method [Surface Energy Balance Algorithm for Land (SEBAL)], the modified Thornthwaite water-balance method (WB), and the complementary relationship method [Granger and Gray model (GG); Granger and Gray (1989)] in the Blue Nile, eastern Sudan. The soil water holding capacities in the Thornthwaite model are approximately 205, 108, and 154 mm in 1-mdeep soil for stations Abu Naama, Damazine, and Gedarif, respectively. The results show that the three methods give comparable results, and the agreement between SEBAL and WB is closer than the agreement between SEBAL and GG method during the wet season (July–September). The Mann–Kendall (MK) trend test (Mann 1945; WMO 1966; Kendall 1975; Sneyers 1990) is widely used in the literature to analyze trends in the climate data. In contrast to the traditional MK test, which calculates the statistic variables only once for the whole sample, the MK method can also be used to test an assumption regarding the beginning of the development of a trend within a sample—that is, a changing point in the time series (Zhang et al. 2010). Following the procedure as shown by Gerstengarbe and Werner (1999), who used the method to test an assumption about the beginning of the development of trend within a sample (x1, x2, . . . , xn) of the random variable X, the corresponding rank series for the so-called retrograde rows are similarly obtained OCTOBER 2012 1525 ZHANG ET AL. TABLE 1. SPI categories based on the initial classification of SPI values. Probability of occurrence (%) Category SPI Region I Region II Region III Region IV Sudan Extremely wet Very wet Moderately wet Near normal Moderately dry Severely dry Extremely dry 2.00 and above 1.50 to 1.99 1.00 to 1.49 20.99 to 0.99 21.00 to 21.49 21.50 to 21.99 22.00 and less 3.30 2.16 7.90 68.68 11.49 4.74 1.72 3.02 2.30 10.92 71.84 5.17 2.01 4.74 1.87 3.45 10.06 71.12 6.90 3.45 3.16 0.57 3.02 14.94 67.53 5.75 4.89 3.30 1.44 3.88 10.92 70.11 5.17 5.17 3.30 for the retrograde sample (xn, xn21, . . . , x1). Based on the rank series r of the progressive and retrograde rows of this sample, the statistic variables Z1 and Z2 are calculated for the progressive and retrograde samples, respectively. The Z1 and Z2 values calculated with progressive and retrograde series are named UF and UB, respectively, in this paper. The intersection point of the two lines, UF and UB give the point in time of the beginning of a developing trend within the time series. The SPI is defined to describe the periods of dryness and wetness. It is based on the long-term precipitation data for a desired period. This long-term record is fitted to a probability distribution, which is then transformed into a normal distribution so that the mean SPI for the location and desired period is zero (Edwards and McKee 1997). In this paper gamma probability distribution was selected for SPI calculation at time scales of 3, 6, 12, and 24 months (Bordi et al. 2003). The gamma probability density function is expressed as g(x) 5 1 x ba G(a) a21 x/b e for x . 0, (7) where a . 0 is a shape parameter, b . 0 is a scale parameter, and x . 0 is the amount of precipitation; G(a) defines the gamma function (Thom 1958). Then an equal probability transformation from a gamma to a normal distribution is applied (Guttman 1999): SPI 5 xi 2 xi . s (8) The SPI is a dimensionless index where negative (positive) values indicate drought (wet) conditions. McKee et al. (1993) defined the criteria for a ‘‘drought event’’ for any time steps and classified the SPI to define various drought intensities. Fitting the distribution function to data requires an estimation of a and b values. Edwards and McKee (1997) suggested that these two parameters can be estimated using the maximum likelihood approximation by Thom (1958) for 1 11 a ^5 4A rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ! 4A and 11 3 ^ 5 x, b a ^ (9) (10) where A 5 ln(x) 2 å ln(x) (11) n and N 5 number of precipitation observations. Integrating the probability density function with respect to x and the estimates of a and b yields an expression of the cumulative probability G(x) of precipitation for a given time step (here the time step is one month): G(x) 5 ðx 0 g(x) dx 5 1 ba G(a)0 ðx xa21 ex/b dx . (12) 0 Since the gamma distribution is undefined for x 5 0 and q 5 P(x 5 0) . 0, where q is the probability of zero precipitation, an adapted statistic H(x) can be calculated using the following formula: H(x) 5 q 1 (1 2 q)G(x) . (13) The cumulative probability distribution is then transformed into the standard normal distribution to yield the SPI. Since the above approach is not practical for computing the SPI for large numbers of data points, such as in our case, we used the approximate conversion suggested by Abramowitz and Stegun (1965). Detailed procedures of the calculation of the SPI can be found in Guttman (1999) and Lloyd-Hughes and Saunders (2002) (cited in Kemal et al. 2005). The aim here was to identify areas vulnerable to dryness and wetness at comparable time steps based on their occurrence frequencies (Livada and Assimakopoulos 2007). An SPI classification scale is used to identify drought conditions according to the SPI values (Table 1). 1526 JOURNAL OF HYDROMETEOROLOGY VOLUME 13 FIG. 2. (a)–(d) Distributions of annual mean precipitation, actual evapotranspiration, potential evapotranspiration, and soil moisture averaged on 1948–2009 in Sudan. 4. Results a. Characteristics of droughts in Sudan From the distribution of annual mean precipitation, actual evapotranspiration, potential evapotranspiration, and soil moisture (Fig. 2), it can be found that the average annual actual evapotranspiration and soil moisture vary greatly in Sudan and decrease from south to north, which is very similar to the pattern of annual mean precipitation. However, the potential evapotranspiration is different; the higher value occurs in northern Sudan and lower value appears in southern Sudan. Four climate regions were divided based on the annual mean precipitation features in Sudan (Fig. 1b). Figure 3 shows the SPI series on the 24 months based on monthly precipitation (PREC) data in the four regions of Sudan for 1948–2009. The SPI at 24 months is considered as a hydrological drought index that can be used to monitor surface water resources [e.g., river flows (Hayes et al. 1999)]. At this time scale, droughts lasted OCTOBER 2012 ZHANG ET AL. 1527 FIG. 4. The MK Z values of the hydrological variables in the hydrological cycle. (a) UF and UB represent the Z values for progressive and retrograde precipitation series, respectively, and (b) Z values for progressive series of six hydrological variables are shown. FIG. 3. The SPI series for 24 months in different regions of Sudan and southern Sudan (the different regions are based on Fig. 1). longer, but were less frequent with few dryness or wetness periods (Livada and Assimakopoulos 2007). As a whole, the dryness and wetness variabilities show similar patterns in different regions in Sudan and wet conditions prevailed in the whole of Sudan during 1948–68, while drier conditions were experienced during 1969– 2009. The extremely wet and dry events were recorded in 1945/55 and 1984/85, respectively. From this figure, it is clear that the transition from wet period to dry period occurs in the late 1960s and more dryness and less wetness are found since then. Although the dry and wet conditions look similar in different regions in Sudan (Fig. 3), a close look reveals different features of the dryness and wetness variations. For example, the wet condition seems to be ending earlier in north Sudan than in south Sudan during the 1960s (Figs. 3a,d), and the amplitude of drought in central and south Sudan seems higher than in north Sudan. From Table 1 we also find that fewer extremely wet and very wet events are found in south Sudan (the probability of occurrence for regions III and IV are 1.87% and 0.57%, respectively) than in central and north Sudan with the probability of occurrence over 3%. However, more extreme dry events can be found in central and south Sudan than in north Sudan. But for the whole country, we can find that the frequency of extreme dry events is more than that of extreme wet events. The annual precipitation and soil moisture are also used to monitor the droughts in Sudan. The areal average annual precipitation over the whole country decreased during 1948–2009 and the abrupt change point can be found in 1968 by using the Mann–Kendall method (Fig. 4a), and the soil moisture also shows a significantly decreasing trend during 1948–2009 (Fig. 4b). As stated previously, more droughts can be found from the late 1960s and the droughts have lasted over more than 40 years in Sudan. From the spatial aspects, more droughts can be found in central Sudan. Similar results were obtained by Hulme (1990) when he reported that the depletion has been most severe in semiarid central Sudan. Other researchers (e.g., Osman and Shamseldin 2002) also found that the areal annual 1528 JOURNAL OF HYDROMETEOROLOGY averaged rainfall values decreased markedly since the 1960s, and the drought in the 1970s produces a large number of impacts that affects Sudan’s social, environmental, and economical standard of living with reduced crop, reduced water levels, increased livestock, and wildlife death rates and damage to wildlife and fish habitat (Zhang et al. 2011). b. The atmospheric hydrological variables Owing to the influence by the tropical and continental climate, the distribution of rainfall in Sudan is very asymmetric. The average annual rainfall shows a descending trend from south to north. More recent analysis indicates that the precipitation of Sudan has a close relation to the amount of moisture transport during the rainy season (Zhang et al. 2011), and the continuous serious droughts of Sudan might be affected by the atmospheric hydrological cycle. Zhang et al. (2011) found that the whole-layer moisture flux in summer [June– August (JJA)] during 1948–2005 decreased significantly in Sudan, which is in good line with the changes of precipitation in Sudan. To better understand the changing characteristics of droughts in Sudan and the corresponding hydrological variables of the hydrologic cycle, the trends of the areal annual mean hydrological variables were analyzed by using the MK method (Fig. 4b). The annual mean atmospheric moisture content, precipitation, actual evapotranspiration, and soil moisture decrease significantly during 1948–2009, while the temperature and potential evapotranspiration show significant increasing trends over the whole country. To further analyze the spatial and temporal variation of hydrological variables, the time–latitude cross section averaged over 22.58–37.58E is shown in Fig. 5. From this figure, obvious positive atmospheric moisture content, precipitation, actual evapotranspiration, and soil moisture anomalies can be found in the 1950s and 1960s, while negative anomalies occur in the 1970s, 1980s, and 1990s. However, the potential evapotranspiration and temperature anomalies are opposite to that of soil moisture and precipitation in which the negative anomalies present in 1950s and 1960s and positive anomalies occur since the late of 1970s. Then we can find that the pattern of potential evapotranspiration and temperature anomalies are opposite to the patterns of actual evapotranspiration and precipitation, which are very similar to soil moisture. c. The relationship between the droughts and the hydrological recycle in Sudan As shown above, the changes of actual evapotranspiration and soil moisture are in good line with that of atmospheric moisture content and precipitation in a long time. But what will happen if they are under dry conditions? VOLUME 13 The meridional cross section of the mean hydrological variables’ differences between the dry period (1970– 2005) and wet period (1948–69) averaged over 22.58– 37.58E are shown in Fig. 6. For atmospheric moisture content, precipitation, actual evapotranspiration, and soil moisture, obvious negative anomalies can be found. The negative anomalies values become greater from January to August and decrease afterward, and the location of maximum negative anomalies varies greatly in different months. The negative precipitation anomalies can be found in the central Sudan in the rainy season with the precipitation anomalies larger than 300 mm yr21. Similar results can be found with other hydrological components, such as atmospheric moisture content, actual evapotranspiration, and soil moisture, for which greater anomalies largely occurred in central Sudan and in the rainy season. The maximum negative values are located in southern Sudan in the dry season and in central and north Sudan in the rainy season. A similar pattern can be found in potential evapotranspiration and temperature except that the anomalies are positive. Figure 7 shows the spatial distribution of the hydrological variables’ anomalies between the dry and wet periods. From this figure, obvious negative anomalies for atmospheric moisture content, precipitation, actual evapotranspiration, and soil moisture can be found in the whole Sudan and the greater negative anomalies are located in central Sudan, while positive anomalies can be found in the whole country for potential evapotranspiration and temperature. To further investigate the differences of hydrological variables between the dry and wet conditions, we analyzed the precipitation recycling ratio over Sudan (Fig. 8). The precipitation recycling ratio, which is defined as the contribution of local evapotranspiration to local precipitation, aims at understanding the hydrological process in the atmospheric branch of the water cycle (Eltahir and Bras 1996). From this figure, we can find that the precipitation recycling ratio averaged over 1948–2009 decreases from south to north and the large value is located in south Sudan. It can be found that the contribution of local evapotranspiration to local precipitation is more than 30%–40% in south Sudan while the contribution is only 10%–20% in central Sudan (Fig. 8a). The precipitation recycling ratios for the African region are presented by Brubaker et al. (1993); two peaks appear in March (r 5 0.41) and in August (r 5 0.48). The February– March peak corresponds to fairly high E and low P in those months, while the July–August peak corresponds to a season of high E and high P. The annual mean precipitation recycling ratio is about 0.3 on the areal average over Africa. The comparison of precipitation recycling ratio in the dry conditions and wet conditions OCTOBER 2012 ZHANG ET AL. 1529 FIG. 5. Time–latitude cross section (averaged over 22.58–37.58E) of hydrological variables anomalies compared with the average on 1948–2009 in Sudan: (a) moisture content, (b) precipitation, (c) temperature, (d) potential evapotranspiration, (e) actual evapotranspiration, and (f) soil moisture. Units are 8C for temperature, and mm for other variables. is shown in Fig. 8b and Fig. 8c, from which we can find that the precipitation recycling ratio in the dry conditions (averaged over 1948–68) is greater than that of wet conditions (averaged over 1969–2009), which indicates the percentage of local evapotranspiration converting into local precipitation in the dry conditions is higher than that in wet conditions. Similar results can be found in the central United States, as Bosilovich and Schubert (2001) pointed out that the 1988 (drought year) summer recycling ratio is larger than that of 1993 (flood year), and that the 1988 recycling ratio is much larger than average. And the diagnosed recycling data show that the recycled precipitation is large when moisture transport is weak and convergence and evaporation are large. 1530 JOURNAL OF HYDROMETEOROLOGY VOLUME 13 FIG. 6. Time–latitude cross section (averaged over 22.58–37.58E) for the hydrological variables anomalies between 1969–2005 and 1948– 68: (a) atmospheric moisture content, (b) precipitation, (c) temperature, (d) potential evapotranspiration, (e) actual evapotranspiration, and (f) soil moisture. Units are the same as in Fig. 5. To better quantitatively estimate the relationship between the hydrological variables, we calculate the correlations between them (Table 2). From this table, we can find good relationships between the hydrological variables. The table reveals that there are high correlations between potential evapotranspiration and temperature with the correlation coefficient of 0.87 and precipitation and actual evapotranspiration with correlation coefficient of 0.86. Significant correlations can also be found between soil moisture and atmospheric moisture content, precipitation, temperature, and actual evapotranspiration; the highest correlation coefficient appears OCTOBER 2012 ZHANG ET AL. FIG. 7. Spatial anomalies’ distribution of the hydrological variables’ anomalies between 1969–2005 and 1948–68: (a) atmospheric moisture content, (b) precipitation, (c) temperature, (d) potential evapotranspiration, (e) actual evapotranspiration, and (f) soil moisture. Units are the same as in Fig. 5. 1531 1532 JOURNAL OF HYDROMETEOROLOGY VOLUME 13 FIG. 8. Spatial distribution of annual mean precipitation recycling ratio over Sudan: (a) averaged for 1948–2009, (b) averaged for 1948–68, and (c) averaged for 1969–2009. between soil moisture and precipitation, and the second largest correlation coefficient is between soil moisture and actual evapotranspiration, which indicates the soil moisture might be more affected by them. 5. Conclusions In this study, we analyzed the characteristics of droughts in Sudan and the corresponding hydrological components during 1948–2009 with the aim of exploring the changing features of droughts and possible relationship between the droughts and hydrologic variables in Sudan. The following conclusions can be drawn from the study. 1) From the estimation of the SPI on a 24-month time scale, we can find wet conditions prevailed during 1948–68 over the whole Sudan while drier conditions OCTOBER 2012 TABLE 2. The correlations between the areal annual mean hydrological variables over the whole Sudan [temperature (TMP), potential evapotranspiration (PET), actual evapotranspiration (AET), and soil moisture (SM); boldface indicates significant at 0.05 level]. Q PREC TMP PET AET SM 2) 3) 4) 5) 1533 ZHANG ET AL. Q PREC TMP PET AET SM 1.0 0.43 1.0 20.53 20.25 1.0 20.33 20.2 0.87 1.0 0.26 0.86 0.02 0.05 1.0 0.56 0.85 20.53 20.42 0.70 1.0 experienced since the late 1960s. More dry events and fewer wet events were found since the late 1960s and the extreme dry and wet events are recorded in 1984/85 and 1954/55, respectively. The changes of annual precipitation and soil moisture also reveal this evidence. Significant decreasing trends can be found in the annual precipitation, atmospheric moisture content, actual evapotranspiration, and soil moisture during 1948–2009, while significant increasing trends occur in the annual temperature and potential evapotranspiration. Precipitation and actual evapotranspiration are the leading terms in the atmospheric water balance over Sudan. The significant decreases in annual soil moisture are associated with the decrease of annual precipitation and the increase of annual temperature. The patterns of soil moisture are most similar to that of atmospheric moisture content and precipitation and the patterns of potential evapotranspiration and temperature anomalies are opposite to that of soil moisture during 1948–2009. Negative anomalies between 1969–2009 and 1948–68 for atmospheric moisture content, precipitation, and actual evapotranspiration can be found in Sudan and the greater negative anomalies are located in central Sudan, while positive anomalies can be found in the whole country for potential evapotranspiration and temperature. So the changes of the hydrological components might lead to more severe droughts in central Sudan. Significant correlations are found between the soil moisture and other hydrologic variables (such as precipitation, atmospheric moisture content, temperature, and actual evapotranspiration); the highest correlation appears between soil moisture and precipitation and the second highest correlation is between the soil moisture and actual evapotranspiration. The precipitation recycling ratio averaged over 1948– 2009 decreases from south to north and the large values are located in south Sudan. It can be found that the percentage of local evapotranspiration to local precipitation is more than 30%;40% in south Sudan while the percentage is only 10%;20% in central Sudan. The precipitation recycling ratio in the dry condition (averaged over 1948–68) is greater than the wet condition (averaged over 1969–2009), which indicates the percentage of local evapotranspiration converting into local precipitation in the dry conditions is greater than that of wet conditions. Acknowledgments. This work is financially supported by the Research Council of Norway with project number 171783 (FRIMUF), ‘‘985 Project’’ (Grant 370003171315), and by the Open Fund of State Key Laboratory of Satellite Ocean Environment Dynamics and the Institute of Desert Meteorology, CMA (Grant Sqj20080011). The second author is also supported by the Program of Introducing Talents of Discipline to Universities—The 111 Project of Hohai University. The authors wish to thank the reviewers for their valuable comments and suggestions, which greatly improved the quality of the paper. REFERENCES Abramowitz, M., and A. Stegun, Eds., 1965: Handbook of Mathematical Formulas, Graphs, and Mathematical Tables. Dover Publications, 1046 pp. Alvi, S. H., 1994: Climate changes, desertification and the Republic of Sudan. GeoJournal, 33, 393–399. Bonaccorso, B., I. Bordi, A. Cancellere, G. Rossi, and A. Sutera, 2003: Spatial variability of drought: An analysis of the SPI in Sicily. Water Resour. Manage., 17, 273–296. Bordi, I., K. Fraedrich, J. Jiang, and A. Sutera, 2003: Dry and wet periods in eastern China watersheds: Patterns and predictability (in Chinese). J. Lake Sci., 15 (Suppl.), 56–67. Bosilovich, M. G., and S. D. Schubert, 2001: Precipitation recycling over the central United States diagnosed from the GEOS-1 data assimilation system. J. Hydrometeor., 2, 26–35. Brubaker, K. L., D. P. S. Entekhabi, and P. Eagleson, 1993: Estimation of continental precipitation recycling. J. Climate, 6, 1077–1089. Cadet, D. L., and N. O. Nnoli, 1987: Water vapor transport over Africa and the Atlantic Ocean during summer 1979. Quart. J. Roy. Meteor. Soc., 113, 581–602. Chen, M. Y., P. Xie, J. E. Janowiak, and P. A. Arkin, 2002: Global land precipitation: A 50-yr monthly analysis based on gauge observations. J. Hydrometeor., 3, 249–266. Chen, T. C., J. M. Chen, and J. Pfaendtner, 1995: Low-frequency variations in the atmospheric branch of the global hydrological cycle. J. Climate, 8, 92–107. Edwards, D. C., and T. B. McKee, 1997: Characteristics of 20th century drought in the United States at multiple time scales. Colorado State University, Fort Collins Climatology Rep. 97-2, 174 pp. El Haj El Tahir, M., W. Wenzhong, C.-Y. Xu, Z. Youjing, and V. P. Singh, 2012: Comparison of methods for estimation of regional actual evapotranspiration in data scarce regions: The Blue Nile region, eastern Sudan. J. Hydrol. Eng., 17, 578–589, doi:10.1061/(ASCE)HE.1943-5584.0000429. 1534 JOURNAL OF HYDROMETEOROLOGY Eltahir, E. A. B., and R. L. Bras, 1994: Precipitation recycling in the Amazon basin. Quart. J. Roy. Meteor. Soc., 120, 861–880. ——, and ——, 1996: Precipitation recycling. Rev. Geophys., 34, 367–378, doi:10.1029/96RG01927. Fan, Y., and H. van den Dool, 2004: Climate Prediction Center global monthly soil moisture data set at 0.58 resolution for 1948 to present. J. Geophys. Res., 109, D10102, doi:10.1029/ 2003JD004345. Fontaine, B., P. Roucou, and S. Trzaska, 2003: Atmospheric water cycle and moisture fluxes in the West African monsoon: Mean annual cycles and relationship using NCEP/NCAR reanalysis. Geophys. Res. Lett., 30, 1117, doi:10.1029/2002GL015834. Gerstengarbe, F. W., and P. C. Werner, 1999: Estimation of the beginning and end of recurrent events within a climate regime. Climate Res., 11, 97–107. Granger, R. J., and D. M. Gray, 1989: Evaporation from natural nonsaturated surfaces. J.Hydrol., 111, 21–29. Guttman, N. B., 1999: Accepting the standardized precipitation index: A calculation algorithm. J. Amer. Water Resour. Assoc., 35, 311–322. Hayes, M. J., M. D. Svoboda, D. A. Wilhite, and Olga V. Vanyarkho, 1999: Monitoring the 1996 drought using the Standardized Precipitation Index. Bull. Amer. Meteor. Soc., 80, 429–438. Heim, J., 2000: Drought indices: A review. Droughts: A Global Assessment, D. A. Wilhite, Ed., Routledge, 159–167. Hulme, M., 1987: Secular changes in wet season structure in central Sudan. J. Arid Environ., 13, 31–46. ——, 1990: The changing rainfall resources of Sudan. Trans. Inst. Br. Geogr., 15, 21–34. ——, and N. Tosdevin, 1989: The tropical easterly jet and Sudan rainfall: A review. Theor. Appl. Climatol., 39, 179–187. Janowiak, J. E., 1988: An investigation of interannual rainfall variability in Africa. J. Climate, 1, 240–255. Kalamaras, N. H. M., H. Michalopoulou, and H. R. Byun, 2010: Detection of drought events in Greece using daily precipitation. Hydrol. Res., 41, 126–133. Kemal, F. S., A. K. Umran, and E. Ayhan, 2005: An analysis of spatial and temporal dimension of drought vulnerability in Turkey using the Standardized Precipitation Index. Nat. Hazards, 35, 243–264. Kendall, M. G., 1975: Rank Correlation Methods. Griffin, 202 pp. Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 1138–1140. Lana, X., C. Serra, and A. Burgueño, 1998: Spatial and temporal characterization of annual extreme droughts in Catalonia (northern Spain). Int. J. Climatol., 18, 93–110. Livada, I., and V. D. Assimakopoulos, 2007: Spatial and temporal analysis of drought in Greece using the Standardized Precipitation Index (SPI). Theor. Appl. Climatol., 89, 143–153. Lloyd-Hughes, B., and M. Saunders, 2002: A drought climatology for Europe. Int. J. Climatol., 22, 1571–1592. Mann, H. B., 1945: Nonparametric tests against trend. Econometrica, 13, 245–259. Mather, J. R., 1978: The Climatic Water Balance in Environmental Analysis. D. C. Heath and Company, 239 pp. ——, 1979: Use of the climatic water budget to estimate streamflow. NTIS Rep. PB 80-180896, 52 pp. McCabe, G. J., and D. M. Wolock, 1999: Future snowpack conditions in the western United States derived from general circulation model climate simulations. J. Amer. Water Resour. Assoc., 35, 1473–1484. McKee, T. B., N. J. Doesken, and J. Kleist, 1993: The relationship of drought frequency and duration to time steps. Preprints, VOLUME 13 Eighth Conf. on Applied Climatology, Anaheim, CA, Amer. Meteor. Soc., 179–184. Mishra, A. K., V. P. Singh, and V. R. Desai, 2007: Drought characterization: A probabilistic approach. Stochastic Environ. Res. Risk Assess., 23, 41–55, doi:10.1007/s00477-0070194-2. Nicholson, S. E., B. Some, and B. Kone, 2000: An analysis of recent rainfall conditions in West Africa, including the rainy seasons of the 1997 El Niño and the 1998 La Niña years. J. Climate, 13, 2628–2640. Nigam, S., and A. Ruiz-Barradas, 2006: Seasonal hydroclimate variability over North America in global and regional reanalyses and AMIP simulations: Varied representation. J. Climate, 19, 815–837. Osman, O. E., and S. L. Hastenrath, 1969: On the synoptic climatology of the summer rainfall over central Sudan. Arch. Meteor. Geophys. Bioklimatol., 17B, 297–324. Osman, Y. Z., and A. Y. Shamseldin, 2002: Qualitative rainfall prediction models for central and southern Sudan using El Niño–southern oscillation and Indian Ocean sea surface temperature indices. Int. J. Climatol., 22, 1861–1878. Palmer, W. C., 1965: Meteorological drought. U.S. Department of Commerce, Weather Bureau Research Paper 45, 58 pp. Peixoto, J. P., and A. H. Oort, 1983: The atmospheric branch of the hydrological cycle and climate. Variation in the Global Water Budget, F. A. Street-Perrott, M. Beran, and R. Ratcliff, Eds., Reidel, 5–65. ——, and ——, 1992: Physics of Climate. American Institute of Physics, 520 pp. Raddatz, R. L., 2005: Moisture recycling on the Canadian Prairies for summer droughts and pluvials from 1997 to 2003. Agric. For. Meteor., 131, 13–26. Rasmusson, E. M., 1968: Atmospheric water vapor transport and the water balance of North America, II. Large-scale water balance investigation. Mon. Wea. Rev., 96, 720–734. Roads, J., M. Kanamitsu, and R. Stewart, 2002: CSE water and energy budgets in the NCEP–DOE Reanalysis II. J. Hydrometeor., 3, 227–248. Shukla, J., and Y. Mintz, 1982: The influence of land-surface evapotranspiration on the earth’s climate. Science, 215, 1498– 1501. Simmons, A. J., and Coauthors, 2004: Comparison of trends and low-frequency variability in CRU, ERA-40, and NCEP/NCAR analyses of surface air temperature. J. Geophys. Res., 109, D24115, doi:10.1029/2004JD005306. Sneyers, R., 1990: On the statistical analysis of series of observations. WMO Tech. Note 143, 192 pp. Su, M. F., and H. J. Wang, 2007: Relationship and its instability of ENSO: Chinese variations in droughts and wet spells. Sci. China, 50D, 145–152. Thom, H. C. S., 1958: A note on the gamma distribution. Mon. Wea. Rev., 86, 117–122. Thornthwaite, C. W., 1948: An approach toward a rational classification of climate. Geogr. Rev., 38, 55–94. Trenberth, K. E., and C. J. Guillemot, 1998: Evaluations of the atmospheric moisture and hydrological cycle in the NCEP/ NCAR reanalyses. Climate Dyn., 14, 213–231. ——, J. T. Overpeck, and S. Solomon, 2004: Exploring drought and its implications for the future. Eos, Trans. Amer. Geophys. Union, 85, 27. Trenberth, K. L., J. T. Fasullo, and J. Mackaro, 2011: Atmospheric moisture transports from ocean to land and global energy flows. J. Climate, 24, 4907–4924. OCTOBER 2012 ZHANG ET AL. Trilsbach, A., and M. Hulme, 1984: Recent rainfall changes in central Sudan and their physical and human implications. Trans. Inst. Br. Geogr., 9, 280–298. Walsh, R. P. D., M. Hulme, and M. D. Campbell, 1988: Recent rainfall changes and their impact on hydrology and water supply in the semi-arid zone of the Sudan. Geogr. J., 154, 181–198. Weaver, S. J., A. Ruiz-Barradas, and S. Nigam, 2009: Pentad evolution of the 1988 drought and 1993 flood over the Great Plains: An NARR perspective on the atmospheric and terrestrial water balance. J. Climate, 22, 5366–5384. WMO, 1966: Climate change: Report of a working group of the Commission for Climatology. WMO Tech. Rep. 79, 79 pp. Wolock, D. M., and G. J. McCabe, 1999: Effects of potential climatic change on annual runoff in the conterminous United States. J. Amer. Water Resour. Assoc., 35, 1341–1350. Xu, C.-Y., and D. Chen, 2005: Comparison of seven models for estimation of evapotranspiration and groundwater recharge using lysimeter measurement data in Germany. Hydrol. Processes, 19, 3717–3734. 1535 ——, Q. Zhang, M. El Haj El Tahir, and Z. Zhang, 2010: Statistical properties of the temperature, relative humidity, and net solar radiation in the Blue Nile-eastern Sudan region. Theor. Appl. Climatol., 101, 397–409, doi:10.1007/s00704-009-0225-7. Yang, C. G., Z. B. Yu, Z. C. Hao, J. Y. Zhang, and J. T. Zhu, 2012: Impact of climate change on flood and drought events in Huaihe River Basin, China. Hydrol. Res., 43 (1–2), 14–22. Zangvil, A., and S. Karas, 2001: Comparative analysis of atmospheric disturbance tracks over the Mediterranean Basin and vicinity. Judea Samaria Res. Stud., 10, 405–420. Zhang, Q., C.-Y. Xu, H. Tao, T. Jiang, and Y. D. Chen, 2010: Climate changes and their impacts on water resources in the arid regions: A case study of the Tarim River basin, China. Stochastic Environ. Res. Risk Assess., 24, 349–358. Zhang, Z. X., C.-Y. Xu, M. El-Haj El-Tahir, J. R. Cao, and V. P. Singh, 2011: Spatial and temporal variation of precipitation in Sudan and their possible causes during 1948–2005. Stochastic Environ. Res. Risk Assess., 26, 429–441, doi:10.1007/s00477-011-0512-6. Zhou, J., Y. F. Xue, and X. F. Liu, 1998: The source/sink distribution of water vapor with its transfer in Asian monsoon region in August, 1994. J. Trop. Meteor., 14, 91–96.