1 Introduction

Drought is the scarcity of water over a long period and is one of the impacts of climate change that manifests in several manners: hydrological, meteorological, agriculture and edaphic (related to soil conditions). For example, meteorological or atmospheric droughts ruin a region's agricultural production, reduce food availability, reduce hydropower production, cause grassland or forest fires and, most importantly, cause shortages of drinking water; thus, they may lead to the emergence of epidemics and diseases in the long term.

For meteorological drought assessment, numerous indices have been suggested and proposed, e.g., the standardized streamflow index (SSI) (Shukla and Wood 2008; McKee et al. 1993), Palmer drought severity index (PDSI) (Palmer 1965), standardized precipitation evapotranspiration index (SPEI) (Vicente-Serrano et al. 2010), reconnaissance drought index (RDI) (Tsakiris and Vangelis 2005), streamflow drought index (SDI) (Nalbantis and Tsakiris 2009), regional drought area index (RDAI) (Fleig et al. 2010, 2011), dimensionless precipitation anomaly index (DPAI) (Eris et al. 2020), Percent of Normal Precipitation Index (PNPI) (Willeke et al., 1994), and rainfall anomaly index (RAI) (Van Rooy 1965). Each index has its own strengths and weaknesses.

In the framework of climate change and drought analysis, several studies in recent years have been conducted around the world (e.g., Brasil Neto et al. 2021; Silva et al. 2020; Şen et al. 2020; Jehanzaib et al. 2020; Malik et al. 2020; Junqueira et al. 2020; Cavus and Aksoy 2020; Santos et al. 2019, 2018; Vergni et al. 2016; Murthy et al. 2016). Monitoring agricultural drought using land surface temperatures from Australian retrievals indicates that MTSAT-2 offers earlier drought control than do the greenness indices (Hu et al. 2020). In the Poyang Lake basin, Zhou et al. (2019) proposed an approach for meteorological drought migration monitoring based on topological spatial relations. During the study period, the proposed approach successfully identified a total of 454 drought clusters and 77 drought events. Nikbakht et al. (2013) analyzed hydrological drought in Urmia Lake (Northwest Iran) by PNPI and the Mann–Kendall (MK) test during the 1975 − 2009 period based on 14 hydrometric stations, and the analysis showed that the worst droughts indicated in the discharge were observed at all stations between 1999 and 2001. Santos et al. (2020) used PERSIANN-CDR-estimated rainfall data for geospatial drought severity analysis at Odisha state in India (1983–2018). The results indicated extreme drought events with increase in the drought severity timescale.

In the Upper Paraíba River basin, Brazil, the spatiotemporal variability in vegetation related to drought dynamics was analyzed using SPI, RAI and Moderate-Resolution Imaging Spectroradiometer (MODIS) products during the 2012–2017 period. Water stress was observed to be related to severe and erratic drought behavior in 2013, 2016 and 2017 (Ferreira da Silva et al. 2020). In the Johor River Basin, in Malaysia, Tan et al. (2019) analyzed spatiotemporal hydrometeorological drought and reported that the standardized precipitation index (SPI) indicated dry conditions occurring in the basin, whereas the SSI showed a strong significant decreasing trend due to drought, reservoir construction and the use of water for agriculture. Cavus and Aksoy (2019) indicated quite different severities simultaneously of droughts that has been experienced the Mediterranean region of Turkey (Seyhan River basin), with increasing drought duration in return periods. The rainfall trends and drought were analyzed over the Oum Er-Rbia River basin, Morocco, during the 1970–2010 period by SPI and MK tests. The results show that a tendency toward drier conditions mostly occurred between the 1980s and 1990s, with a deficit exceeding 50% (Ouatiki et al. 2019). Across KwaZulu-Natal Province, South Africa, Ndlovu and Demlie (2020) assessed wet conditions and meteorological drought using two drought indices. The PNPI and RAI analyses indicated that the most extreme dry years occurred in 1992 and 2015, and the latter drier year was associated with El Niño conditions in the studied period.

Among the areas most affected by periods of drought are the tropics and subtropics located in the tropics of Cancer and Capricorn. These are in fact the areas with the hottest and driest climates. The most significant droughts in recent years have been located in the Sahel, the Horn of Africa, Mexico, northeastern Brazil and parts of China, India and Russia, as well as in southern Europe. In addition to the arid zones, the most vulnerable countries are found in Africa and West Asia (Cieau 2020). Algeria, which is mostly located in arid and semiarid zones in the Mediterranean basin, is considered one of the regions that can be influenced by periodic and persistent droughts. Since the drought struck the Mediterranean basin, especially Algeria, beginning at the end of 1970, a number of research papers have been written in this context to identify, assess and quantify the established drought in the region (Hallouz et al. 2020; Derdous et al. 2021; Mellak et al. 2020; Bendjema et al. 2019; Habibi et al. 2018; Achite et al. 2017; Khezazna et al. 2017; Meddi et al. 2014, 2010). In the whole of northeast Algeria, Merabti et al. (2018) used the SPI and RDI to analyze the spatiotemporal variability of drought between 1979 and 2014. Severe and moderate droughts were found to have occurred in the southern and more arid subregions, which can be associated with global warming. Using the standardized relative difference of precipitation (SRDP) over 40 years in the Macta basin (Algeria), a notable dry trend was observed in the northern polygon, while a dominant temporal instability was observed at higher elevation polygons (Achite et al. 2017). In the karst area of northwestern Algeria, Bouabdelli et al. (2020) applied bivariate frequency analysis and GR2M model on the representative concentration pathway emission scenario database (RCP4.5 and RCP8.5) for an assessment of hydrological drought risk. All simulations indicated an increase in future hydrological drought risk explained by a decrease in the return periods of droughts. Zerouali et al. (2019), based on cross-wavelet analysis and wavelet coherence, observed a significant and negative impact on the flow decrease, which was explained by drought impacts and by the smuggling and trafficking of gravels and rocks from the Wadi and the commissioning of the Taksebt dam beginning in the year 1998. In this study, spatiotemporal variation in meteorological drought was analyzed and classified using the SPI and geographic information systems (GIS) at monthly, seasonal and annual scales in the Oued Sebaou basin based on 23 rain gauges.

2 Study area and databases

The study area is the Oued Sebaou basin of northern Algeria, which is located between 36°30′N and 37°00′N and between 03°30′E and 04°30′E. This area is part of the coastal Algiers watershed, according to the Algerian National Agency of Water Resources (ANRH) (Fig. 1), and it extends over 2500 km2. It is bordered by the Mediterranean coastal area (in the north), the Djurdjura mountain chain (in the south), the massif Akfadou and Beni Ghobri (in the east) and the Sidi Ali and Jebel Bounab Bouberak mountains (in the west). The total population of the study area is 1,127,607 inhabitants, and 77.7% of the inhabitants live in rural areas.

Fig. 1
figure 1

Land cover map of northern Algeria. (Benslimane et al, 2008), geographic location and digital elevation model of the study basin

Full size image

The Taksebt dam is located in this region, which is an important dam that collects water from the Aïssi valley that is effluent from the Sebaou River. It is mainly intended to be a drinking water supply to the cities of Tizi Ouzou and Algiers. Vegetation cover occupies the largest percentage of the area, amounting to 71.5% of the total area. In general, the climate is affected by various factors, such as geography, topography, distance from the sea and atmospheric circulation. The climate of the study area is characterized by a dry season that begins at the end of May and ends in September and a wet winter that starts in October and ends at the end of May.

Dry, northwest winds, meeting Iberian mountain chains (Spain) and the Moroccan Atlas Mountains, arrive in northwestern Algeria and then recharge on the Mediterranean to provide heavy rainfall on the northern Tell Atlas Mountains. The rainfall amount declines from north to south, superimposed on an increase in the rainfall amount from west to east with longitude; this trend is distinct in Algeria, which is why the Oued Sebaou basin is one of most-watered areas of northern Algeria.

The rainfall of the study area has had a significant relationship with the North Atlantic Oscillation (NAO) for approximately 1 year, 1–3 years and 3–5 years beginning in the early 1980s, corresponding to the dry period. The Southern Oscillation (SOI), Mediterranean Oscillation (MO) and Western Mediterranean Oscillation (WeMO) indicate local influences on the rainfall, with significant values more or less localized in the time–frequency space. The 5–10-year and greater-than-10-year components are the most effective for representing significant climate index-rainfall relationships (Zerouali et al. 2018). The rainfall series database used in this study includes data collected at 23 stations (Fig. 1) at the monthly scale in the period from 1972 to 2010. These rainfall data were taken from the ANRH.

3 Material and methods

3.1 Precipitation concentration index and modified Fourier index

Among the most important aspects of climate studies are monthly precipitation concentrations. An uneven distribution of precipitation reflects periods of excesses and deficits in rainfall, which affect crop growth and cause difficulties in harvests. In addition, precipitation concentrations should be considered when assessing and forecasting soil loss due to water erosion (Michiels et al. 1992). Oliver (1980) introduced the precipitation concentration index (PCI) to study the concentration of rainfall during the hydrologic year by a classification presented in Table 1. The PCI index is given by the following expression:

$${\text{PCI}} = \frac{{\sum\nolimits_{{i = 1}}^{{12}} {P_{{mi}}^{2} } }}{{\left( {\sum\nolimits_{{i = 1}}^{{12}} {P_{m} } } \right)^{2} }} \times 100$$
(1)

where Pm is the monthly rainfall in the month i.

Table 1 PCI, MFI and PCI classifications
Full size table

Fournier et al. (1960) first proposed the index of aggressiveness, named modified Fournier index (MFI); afterwards, this index was modified by Arnoldus (1980). Fournier’s classification is presented in Table 1, and, according to Fournier et al. (1960) and Arnoldus (1980), the MFI is given by the following equation:

$${\text{MFI}} = \sum\nolimits_{i = 1}^{12} {\frac{{P_{mi}^{2} }}{{P_{a} }}}$$
(2)

where Pmi is the monthly rainfall in month i and Pa is the annual rainfall amount for a given year. The PCI and MFI were applied to the data from 23 rainfall stations at the monthly scale. For more detail about PCI and MFI methodology and implantation, the lecture is cordially invited to read the paper of Yeşilırmak and Atatanır (2016), De Luis et al. (2010)

3.2 Standardized precipitation index

McKee et al. (1993) developed the standardized precipitation index (SPI), an index based on probability theory for monitoring meteorological drought. The following SPI classification proposed in Table 1 defines the classification of dry and wet events according to the index values. According to McKee et al. (1993), the SPI approach first fits the hydrometeorological series to a probability distribution and then the series into a normal distribution N (0,1). Thom (1958) found that the gamma distribution fits the hydroclimatological series well. For this reason, the gamma probability density function was used to calculate the distribution of rainfall, which is described in Eq. 3:

$$g(x) = \frac{1}{{\beta^{\alpha } \;\Gamma (\alpha )}}x^{\alpha - 1} e^{ - x/\beta } \;\;{\text{for}}\;x > 0$$
(3)

where α and β are the shape and scale parameters, respectively; Γ(α) denotes the gamma function; and x is the rainfall value. The maximum likelihood method (Eq. 4) was applied for each monitoring rain gauge and each scale observation of each month to calculate the parameters α and β:

$$\alpha = \frac{1}{4A}\left( {1 + \sqrt {1 + \frac{4A}{3}} } \right)\;\;{\text{and}}\;\;\beta = \frac{{\overline{x}}}{\alpha }\;{\text{where}}\;\;A = \ln (\overline{x}) - \frac{{\sum\nolimits_{i = 1}^{n} {\ln {\kern 1pt} \;(x_{i} )} }}{n}$$
(4)

where n and \(\overline{x}\) are the number of observations and their average, respectively. The obtained parameters above were subsequently used to determine the cumulative probability of the observed rainfall episode for the given month- and timescale. The gamma function was not defined for dry months (i.e., x = 0), and the time series distribution may thus have comprised zeros (widely observed in arid and semiarid areas), the new cumulative probability was calculated using Eqs. 5 and 6:

$$G(x) = \int\limits_{0}^{x} {g(x) = \frac{1}{{\beta^{\alpha } \;\Gamma (\hat{\alpha })}}\int\limits_{0}^{x} {x^{{\hat{\alpha } - 1}} \;e^{{ - x/\hat{\beta }}} dx} }$$
(5)
$$H(x) = q + (1 + q)\;G(x)$$
(6)

where q indicates that the probability of the rainfall amount is equal to zero. The cumulative probability H(x) is subsequently transformed into a normal distribution of Z random variables. Equations 7, 89, and 10 were used to convert the cumulative probability into a normal distribution of the Z variable:

$$Z = SPI = - \left[ {t - \frac{{c_{0} + c_{1} t + c_{2} t^{2} }}{{1 + d_{1} t + d_{2} t^{2} + d_{3} t^{3} }}} \right]\;\;\;0 < H(x) \le 0.5$$
(7)
$$Z = SPI = + \left[ {t - \frac{{c_{0} + c_{1} t + c_{2} t^{2} }}{{1 + d_{1} t + d_{2} t^{2} + d_{3} t^{3} }}} \right]\;\;\;0.5 < H(x) \le 1$$
(8)

where c0 = 2.515517, c1 = 0.802853, c2 = 0.010328, d1 = 1.432788, d2 = 0.189269, d3 = 0.001308.

$$t = {\text{ }}\sqrt {\ln \left( {\frac{1}{{H(x)^{2} }}} \right)} {\text{ }}0{\text{ }} < {\text{ }}H(x) \le 0.5$$
(9)
$$t = {\text{ }}\sqrt {\ln \left( {\frac{1}{{(1 - H(x)^{2} }}} \right)} \;0.5{\text{ }} < H(x) \le 1$$
(10)

According to Shah et al. (2015), the SPI is well spread on dry events with extreme conditions; it is simple and practical and is one of the preferred tools for agricultural field assessments. In this work, the annual and seasonal patterns of droughts were assessed using the SPI-12 and SPI-3 values for each gauge, respectively. For SPI-12, 897 cases were evaluated (39 years × 23 stations), and for the seasonal timescale, the same number of cases was also evaluated for each season (summer, autumn, winter and springer), totaling more than 3,500 cases (39 years × 23 stations × 4 seasons). For more details about the methodology of SPI index calculation and implantation, see Santos et al. (2017) and Okpara et al. (2017).

3.3 Interpolation technique

Inverse distance weighting (IDW) is a deterministic interpolation method that is suitable as a local method of interpolation (Cavus and Aksoy 2019). In this method, the estimated value is determined using the weighted average of the closest points to the reference point. The points that are closer, known values are more influenced than more distant points are. Equation 11 shows how computing the estimated values in this paper:

$$\hat{Z}(X_{0} ) = \frac{{\sum_{i = 1}^{N} \left(\frac{{Z(X_{i} )}}{{D_{{X_{0} }} }}\right)}}{{\sum_{i = 1}^{N} \left(\frac{1}{{D_{{X_{0} }} }}\right)}}$$
(11)

where \(Z(X_{i} )\) is the value measured at the location, \(D_{{X_{0} }}\) is the distance between the value of the known point and the estimated point, \(\hat{Z}(X_{0} )\) is the estimated value at location X0, and N is the number of measured values.

4 Results and discussion

Rainfall in the Oued Sebaou River basin was characterized by high spatiotemporal irregularity during the study period and was influenced most by longitude. Figure 2a, b shows the spatial interpolation of the intra-annual rainfall amounts and variation coefficients, and Fig. 3 provides scatter plots of the intra-annual rainfall amounts according to altitude, latitude and longitude. Although altitude is an important factor that explains more than 34% of the variability in rainfall over the region, longitude is considered the most important variable affecting the geographical distribution of annual precipitation, responsible for at least 62% of the variance in the region. The average intra-annual precipitation of the 23 stations varies between 724 and 1200 mm per year, with variation coefficients greater than 25% for most stations.

Fig. 2
figure 2

Distribution and spatial interpolation of a intra-annual rainfall (mm) and b variation coefficient, c the PCI index and d MFI index for monthly precipitation (1972–2010) for 23 rain gauges of Oued Sebaou basin

Full size image
Fig. 3
figure 3

Relationship between rainfall, PCI, MFI and geographical factors (latitude, longitude and elevation)

Full size image

Figure 2c shows the spatial distribution of the PCI values in the region, and it is notable that the PCI results show moderate precipitation concentrations for all stations in the basin, varying from 11.0 to 12.6. The MFI index (Fig. 2d) shows that the aggressiveness of the rainfall is proportionally influenced by the same factors that influence the rainfall variability in the watershed. Additionally, the aggressiveness of rainfall events in recent years contributed to the extent of soil erosion, particularly the erosion of bare soil.

Moreover, the stations were divided in relation to their geographical parameters (altitude, longitude and latitude) and their recorded annual precipitation amounts, as shown in Table 2. Zone 1 is the zone with weak aggressiveness and is concentrated over the lower and western parts of the middle Sebaou River (Fig. 2d). Zone 2 is dominant and presents moderate aggressiveness in the eastern and upper Sebaou River, except for the Ait Ouabane station, which is part of Zone 3 (Fig. 2d). Zone 3 is a region located in the mountainous areas (e.g., Aghrib, Ait Ouabane, Tagma and Yakouren) and shows strong aggressiveness (Fig. 2d).

Table 2 PCI, MFI and PCI classifications between the period of 1972 and 2010 for the 23 rainfall stations of the study area
Full size table

The SPI-3 and SPI-12 values were computed for each rain gauge between 1972 and 2010 and were adapted for the analysis and monitoring of drought variability in the study area. The Baghlia (Zone 1), T’N Boubhir (Zone 2) and Yakouren (Zone 3) stations were selected as representative stations of each zone, and their SPI values were computed (Fig. 4). The SPI values oscillate between − 2 and 2, which highlights wet and dry episodes over the region. Moreover, it can highlight the existence of particularly wet or dry years and note that the dry sequences are relatively longer than the wet sequences are.

Fig. 4
figure 4

Annual and seasonal distribution of SPI for Baghlia, T’N Boubhir and Yakouren between 1972 and 2010

Full size image

Overall, it was found that the 1980s and 1990s were more affected by drought events than other decades were, especially between 1986 and 1998, when the frequencies of moderate and severe dry events were higher in the region. This can be explained by the rainfall deficit, which affected the entire African continent during the same period (Nicholson et al. 2018). Figure 5a shows that more than 60% of the studied years were categorized, in general, as near normal to normal. Extreme wet and moderately dry events were observed in a pronounced way over 6 years, but the other drought categories did not exceed 10%. The percentages of stations affected by the different categories are documented in Fig. 5b. Rare, moderate dry events affected less than 18% of stations, especially between 1976 and 1978 and between 1983 and 2001. Moderate and severe dry events prevailed throughout 50% of stations in 1987, 1989, 1993, 1996, 1999 and 2001, which can be related to the drought that started in 1980 and affected the African Sahel region (Ozer et al. 2003; Nicholson et al. 2018).

Fig. 5
figure 5

Classification of drought a average percentage of occurrence of drought and b percentage of stations affected by the different categories of drought according to the SPI index during the period of 1972–2010

Full size image

In contrast, 1972, 1973, 1975, 1980, 1984, 1997, 2002, 2008 and 2010 were characterized by moderate, severe and extreme wet events. Figures 6 and 7 explain, in detail, the frequency of dry and wet events at each rainfall station in the Oued Sebaou basin. The findings in this study are consistent with those of other studies carried out at the scale of the Mediterranean basin, which indicates that the existence of dry periods prevailed more during 1986 and 2001. Over Portugal, for example, Barros et al. (1995) observed that 1993 was the driest year based on a 60-year database, as also found by Bordalo et al. (2006). In southern Spain, Peñuelas et al. (2001) documented that 1988, 1989 and 1994 were the most dangerous, driest years. In Serbia, Gocic and Trajkovic (2014) showed that one of the driest years throughout history was observed in 2000.

Fig. 6
figure 6

Drought classification (duration or number of years) at a annual, b autumn, c winter, d spring, e summer for each station in the study region during the period of 1972–2010

Full size image
Fig. 7
figure 7

Spatial distribution of percentage of occurrence of each drought category: a extremely wet, b very wet, c moderately wet, d near normal, e moderately dry, f severely dry, g extremely dry during the period of 1972–2010

Full size image

Similarly to the statistical application of Faye et al. (2015), it is assessed a total of 897 events (39 years × 23 stations) using statistical methods. The results indicate that more than 20% of the cases are marked by wet conditions in the 1970s and 2000s, and near-normal to normal events exceed 70% of all the cases (Fig. 8). Specifically, 1.75% of cases were characterized as extreme wet events during the 1970s, and this percentage increased to 7% for the 2000s. Regarding the patterns of the 1980s and 1990s, it can be noted that drought periods far outweighed wet periods, and the summarization of drought events reached 32% and 21%, respectively, which were specifically composed of severe and moderate dry events. Globally, the analysis based on SPI indicates the alternation of three periods: the first is a wet period (1970s), the second is a long-term drought period, established in the basin during the 1980s and 1990s, and the third is considered a wet period (2000s). Figure 9 characterizes the alternation of the three periods.

Fig. 8
figure 8

Frequencies (number of stations multiplied by years in %) of SPI classes during the period of 1972–2010

Full size image
Fig. 9
figure 9

Mapping of average values of SPI index by decade at a annual, b autumn, c winter, d spring and e summer

Full size image

Spatiotemporal drought variability was analyzed (Fig. 9) using a GIS by the comparison and assessment of the mean annual and seasonal SPI magnitudes per decade. As discussed above, Fig. 9 clearly emphasizes excess rainfall amounts during the 1970s and 2000s, which were prominent in all cities, except at the stations of Berber, Baghlia and Ait Ouabane, where extreme dry events were detected. At the scale of the Mediterranean basin, Khomsi et al. (2015) pointed to a nonsignificant upward trend in heavy rainfall events in some Moroccan regions, and Donat et al. (2014) observed a positive trend toward wetter conditions, especially in the western parts of Algeria, Tunisia, Libya, Morocco and Mauritania.

The maps representing two successive decades, the 1980s and 1990s, clearly reveal the deficit of rainfall, which was found particularly in the southern part of the basin, where the SPI values varied between -0.2 and -0.7. The results presented in this paper at the basin scale agree with the observations reported by researchers in northern Algeria and in the Mediterranean region (Gader et al. 2020; Bouabdelli et al. 2020; Achour et al. 2020; Hallouz et al. 2020; Zerouali et al. 2020; Mrad et al. 2020; Caloiero et al. 2019; Spinoni et al. 2018; Jemai et al. 2018; Taibi et al. 2017; Nouaceur and Murărescu 2016; Meddi et al. 2014; Hamlaoui-Moulai et al. 2013). In a south Mediterranean transboundary catchment (Medjerda), drought was assessed by Gader et al. (2020) using multivariate and geostatistical approaches during the period 1973–2012. A significant drought was detected in the 41 stations studied with moderate drought was observed during the decades 1980 and 1990.

Ghenim and Megnounif (2013) revealed a climatic wet period in northwestern Algeria between the 1940s and the mid-1970s and a dry period beginning in the 1980s and continuing to the end of the 1990s. In Serbia, Gocic and Trajkovic (2014) documented severe droughts during the period of 1987–1994, with the worst dry year observed in 2000. Over Morocco, Esper et al. (2007) pointed out dry conditions in the 1980s, and Xoplaki et al. (2004) observed a particularly dry period from 1981 − 1995. In Central Italy, Di Lena et al. (2014) revealed a downward trend between 1970 and 1995, followed by an increasing trend during the subsequent decade. Vicente-Serrano et al. (2006) observed in the Iberian Peninsula that the most severe and intense drought episodes occurred between 1940 and 1960 and from 1980 to 2000. Silva et al. (2015) analyzed the trends in annual precipitations and river flow at a regional scale over 40 years (1960–2000) for Cobres River basin, southern Portugal, using statistical tools to obtain a better understanding of the spatial and temporal variability and trends of rainfall and river flow. In Algeria, a wet period occurred between 1999 and 2009, and similar results were observed in all seasons except in autumn (Khezazna et al. 2017).

In addition, the analyses of the annual and seasonal rainfall amounts show great variability in the rainfall averages between decades (Fig. 10). At the Ait Aicha station, for example, the precipitation tends toward dry conditions for the first three decades: 1225 mm (1970s), 994 mm (1980s) and 959 mm (1990s). However, in the 2000s, an increase in the average rainfall was observed compared to the 1980s and 1990s, but the value was still below the 1970s average. This reversibility of the trend can be integrated into the decrease in drought in the Sahel, a cyclical periodicity, the fluctuation of rainfall phenomena or even external impacts linked to large-scale teleconnection patterns, such as the NAO, which is responsible for variability in precipitation in northern Algeria and the Mediterranean basin (Zerouali et al. 2018; Zeroual et al. 2017; Vergni et al. 2016; Leila-Hamlaoui et al. 2013; Ferrari et al. 2013).

Fig. 10
figure 10

Spatial interpolation and mapping of rainfall averages variations: a annuals b autumn, c winter, d spring and e summer between the decades 1972–1981, 1982–1991, 1992–2001 and 2002–2010

Full size image

Several research studies have shown that the NAO is a principal cause of the appearance of drought in northern Algeria (Zeroual et al. 2017; Taibi et al. 2017). Using wavelet techniques, Zerouali et al. (2018) showed that the relationship between the NAO and precipitation in northern central Algeria is significant for 1-year and 3–5-year timescales, especially during the period with a dry tendency beginning in 1980. Mathbout et al. (2019) analyzed the spatiotemporal variability in daily precipitation and the relationship of the variability with large-scale circulation phenomena. The decrease in precipitation over the western Mediterranean could be linked to the positive trend of the NAO during the 1975–1995 period. In the Mediterranean region, Montaldo and Sarigu (2017) analyzed the potential links between the NAO and the reduction in precipitation and river flows. The results indicate that a decrease in winter precipitation induces a decrease in river flow, and the NAO was found to be an influencing factor on the appearance of negative trends in winter precipitation. Corona et al. (2019) analyzed the effect of the interannual variability driven by the NAO in the seasonality of precipitation on water resources. The results showed that the decrease in annual runoff was more pronounced than that in annual precipitation, and winter precipitation was significantly negatively correlated with the NAO.

5 Conclusion

This study evaluated the spatiotemporal pattern of meteorological drought in northern central Algeria using SPI and GIS tools. In general, the precipitation over the study area can be described as irregular and nonuniform in spatiotemporal terms, especially considering the intra-annual distribution from 1972 to 2010. The results revealed moderate precipitation concentrations at all stations in the basin, and longitude was found to be one of the most influential factors for this effect, responsible for approximately 60% of the variance in precipitation. The results based on the MFI show three main distributions of aggressiveness over the region: (a) weak, in the case of the lower Sebaou River, (b) moderate, in the case of the middle Sebaou River and (c) strong, in the case of the upper Sebaou River.

In terms of drought, the existence of dry periods that were more pronounced between the 1980s and 1990s is highlighted, which reached rainfall deficits of 30%, and wet periods that occurred frequently in the 1970s and 2000s. On the seasonal scale, the pattern was similar to the results found on the annual scale, except for in spring, which showed a clear downward trend in the 1980s. At the regional level, results presented in this paper corroborate those of other researchers that reported an increase in average precipitation across the Mediterranean basin over the 2000s. The results presented in this paper can be used for understanding past climate and drought variability over the study area and can be useful for decision-making and for the planning and management of water resources.