1 Introduction

It is hard to talk about twenty-first century water issues without mentioning climate change. Abundant evidence shows that the earth has been warming by several degrees centigrade for the last few decades. It is highly likely that this trend is related to increasing concentrations of carbon dioxide and other greenhouse gasses in the atmosphere (IPCC 2014). Computer models that simulate the earth’s climate tend to agree that the warming trend is likely to continue throughout the rest of this century (Hayhoe et al. 2017). The models show less agreement about projected changes in precipitation but they do tend to agree that some places (primarily the higher latitudes) will be getting wetter and others (primarily the lower latitudes) will be getting drier (Collins et al. 2013). The models also tend to agree that there is likely to be more variation in precipitation. Both floods and droughts are likely to recur with greater frequency, duration and intensity (Parry et al. 2007; Trenberth 2008).

One of the most concerning aspects of climate change is the changes in water cycle. These projected changes can perturb the hydrological cycle; higher temperatures are likely to speed up evaporation from water and land surfaces and speed up transpiration from plants. No matter what happens to precipitation, increased evapotranspiration will dry out soils and leave less water available to flow to streams as well as less water available to infiltrate into the ground and recharge aquifers (Stagl et al. 2014). Warmer temperatures will also cause a shift in precipitation toward less snow and more rain and will lead to earlier and more rapid melting of both the seasonal snowpack and glaciers that have persisted for many years (Hoegh-Guldberg et al. 2018). Melting glaciers might cause increased runoff in glacier-fed streams in the short run, but in the long run as glaciers shrink, the melt water runoff will diminish. Along the coasts, rising sea levels are likely to inundate more low-lying land (Stagl et al. 2014).

In areas with increased precipitation, such as the northeast and certain coastal areas, more severe storms, and flooding are expected; however, in the interior of the country, more drought, less stream flow, and less groundwater recharge will eventuate (Wong et al. 2014). Water quality is likely to be affected too (Xia et al. 2015), with warmer temperatures, muddier water from increased erosion, more pollutants from storm water runoff during heavy rains, and in some places, higher contaminant levels due to less stream flow available due to diluting waste water discharges.

Considering the impacts of both climate change and population growth, water scarcity is projected to rise significantly, According to population action international based on the United Nations Medium Population Projections of 1998, more than 2.8 billion people in 48 countries will face water stress or scarcity conditions by the year 2015, of these countries, 40 are in Western Asia, Africa, or Sub-Saharan Africa. By 2050, the number of countries facing water stress or scarcity could rise to 54 with combined population of 4 billion people, about 40% of the projected global population of 9.4 billion (Gardner-Outlaw and Engleman 1997).

Climate change effects will vary significantly, severely affecting infrastructure in different regions of the world (IPCC 2014). As climate change progresses, an important question must be asked—how to modify urban infrastructure dimensions to adapt to these expected impacts (Carter et al. 2015). Usually, urban infrastructure is constructed, assuming a stable climate and until recently climate change has not been taken into account by engineers during the design stage. Now, changes in the frequency and intensity of hydroclimatic averages and extremes will probably change the critical design parameters (Westra et al. 2014). For this reason, a simple risk formulation, which takes into consideration the impact of climate change, must be developed to adapt for these vulnerabilities (Şen 2014; Şen et al. 2017).

The objective of this study is to generate a relationship between the climate trends and the corresponding risk levels for temperature and precipitation data in Northeast region of Algeria. This relationship is required to establish precautionary boundaries to account for climate change impacts.

2 Study Area and Data

Algeria is a semi-arid country located in North Africa and has experienced a period of very severe droughts with annual average temperature increases ranging from 0.65 to 1.45 °C between 1970 and 2004. An average temperature increase comparable with the global average is observed over the period 1906–2005 (Mahmood et al. 2019). Precipitation in Northern Algeria has shown a succession of excessive episodes and deficiency periods compared to historic norms with a large variability (Meddi 2009).

Numerous studies have been conducted on the northern regions of Algeria regarding precipitation and temperature, but they are directed more towards the west than the east (Mohammed and Al-Amin 2018). Bessaklia et al. (2018) studied 23 rainfall stations in the extreme northeast of Algeria over the period 1970–2010 and concluded that the results of Mann–Kendall test show that areas of high precipitation concentration tend to increase. Merabti et al. (2017) examined five locations spread over the Northeast part of the country and inferred that arid and semi-arid zones have experienced a larger number of drought events while the humid and sub-humid locations received greater precipitation events.

The region Northeast of Algeria (Fig. 1) represents all of the territory from the small Kabyle to the Tunisian border, occupying an area of 31,555 km2 and more than 5 million inhabitants.

Fig. 1
figure 1

Study area and observation station locations

Full size image

Any scientific research needs reliable data in as much detail as possible and, therefore, daily temperature and precipitation data are considered from the National Agency for Hydraulic Resources (ANRH) as well as from Algerian Meteorological Organization since 1982 until 2017 for 35 years. Monthly time series are obtained using daily data. In this paper, monthly average temperature and precipitation records are considered from six stations, which are well distributed within five different watersheds as presented in Table 1.

Table 1 Observation stations and related information
Full size table

3 Methodology

Risk can mean different things to different people and in different contexts, but in very simple terms, it is how likely is it that someone will be exposed to a bad consequence:

Risk = likelihood * consequence.

Likelihood == hazard.

Consequence == how bad the outcome will be.

Since the consequences of climate impacts can be so catastrophic for those exposed to the risk, it is important to urgently commit resources to develop a greater understanding of the risk as far as possible, to fill data gaps and overcome key weaknesses in our understanding and act in a precautionary manner rather than postpone action until better data are collected.

Risk assessment and risk models cannot make decisions but they can inform policymakers. Encouraging and developing a culture of risk identification, risk understanding, risk assessment and risk modeling ultimately benefits society by making it more resilient and saving lives, livelihoods and property.

In this paper, a general risk assessment procedure is implemented from a climate change perspective. Climate change is expected to affect all hydroclimatic variables at many different levels. Theoretically several convenient formulations have been generated for precipitation design values. Climate change effects are ignored in most calculation procedures. In water resource projects, risk levels are determined based on the available methods which currently ignore climate change influence. In the classical approach, risk level is defined based on the frequency of damage occurrence. The simple definition of classical risk (r) suggested by (Kupper 2012) is as follows:

$$ r = \frac{1}{R}, $$
(1)

where R is observed only once during the whole planned future life of any water structure. This simple risk approach is modified by Şen et al. (2017) by taking into consideration the climate change effect as follows:

$$ C_{r} = \frac{1 + \alpha }{R}, $$
(2)

where Cr is the risk under the climate change impact and α is the slope of trend, and α = 0 implies neither increasing or decreasing climate change occurrence. In other cases, climate change can be considered directly both as positive (increasing) and negative (decreasing) contributions to risk. In this methodology, α is calculated as the slope of the trend within the hydrometeorological time series after normalization procedures are applied to the available data. The Matlab software was used to determine risk values based on different risk levels and cumulative distribution function (CDF curves) for each station.

4 Application

An extreme event is defined as the occurrence of a variable value above (or below) a threshold near the upper (or lower) ends of its observed range in a specific region.

The time series for each station with the trend component and corresponding CDF curves for exceedance probability calculations are presented in a set of figures for precipitation and temperature records. For instance, Annaba station trend and risk graphs are presented in Fig. 2.

Fig. 2
figure 2

Trends and CDF curves for Annaba Station. a Temperature, b precipitation, c temperature CDF, and d precipitation CDF

Full size image

The figures show a negative trend for temperature and positive trend for precipitation, but neither trend is significant compared to the variability in the series. There is a noticeable difference between the temperature and precipitation results. For example, the temperature trend graph has a shift each 10-year interval from below (above) average while the precipitation trend graph remains unstable throughout the time period, especially, during the period between 1995 and 2000 which shows dramatic variation between successive years.

The straight line plot from minimum to maximum of the temperature CDF curve reveals a high kurtosis, with higher probability of extreme events occurring at the right-hand tail representing high temperatures. The area below the line is greater than above indicating a negative skew (median > average), with more than the half of daily temperatures (86% of the time series) exceeding the average. The precipitation CDF curve does not show high kurtosis, the area below the straight line from maximum to minimum is greater than above, indicating negative skew (median > average), 60% of the time series exceeds the average with 10% of the series reaching very extreme levels between 850 mm to more than 1100 mm.

Bejaia station analysis results are presented in Fig. 3. As for temperature, there is a positive trend in precipitation but this is of less significance. Additionally, one can also observe the variability in precipitation records alone. The temperature records are notably more regular except for 2006 where there is large variation. The temperature CDF curve reveals very high kurtosis implying a very high probability of extreme events in the tails of the distribution. As previously, the area below the straight line from minimum to maximum is greater than above, indicating negative skew (median > average) with a very large proportion of the time series (96%) exceeding the average. Finally, the precipitation CDF curve shows a higher standard deviation (implying a high level of dispersion and risk). The area below the max–min straight line is greater than above, indicating negative skew (median > average), 75% of days received greater precipitation than average, with only 10% of the results regarded as extremes between 1000 and 1500 mm.

Fig. 3
figure 3

Trends and CDF curves for Bejaia Station. a Temperature, b precipitation, c temperature CDF, and d precipitation CDF Ab

Full size image

Bordj-Bouarriridj station analysis results are presented in Fig. 4 and show a negative trend for temperature and a positive trend for precipitation. Variability exists in both parameters but in different ways. Temperatures are increased at the same rate until 2013 but then increased much more rapidly, whereas precipitation varied significantly in the early years but then stabilized with time. For temperature, the CDF curve has a high kurtosis with high probability of extreme events occurring in the tail of the distribution. In both cases, the right tail represents very high temperatures with the area below the line exceeding that above, indicating negative skew (median > average); 82% of each day’s temperatures are above the average, 5% exceed 18 °C to a maximum of 20 °C. For precipitation, the CDF plot is flat with higher standard deviation (dispersion and risk). Again, the area below the line is greater than that above, indicating a negative skew (median > average), on 76% of days the average was exceeded, with 10% of results reaching the extremes of between 1000 and 1500 mm.

Fig. 4
figure 4

Trends and CDF curves for Bordj-Bouarriridj Station. a Temperature, b precipitation, and c temperature CDF

Full size image

Figure 5 shows the results for Constantin station. There is a significant positive trend for temperature with the large variability stabilizing with time. However, precipitation trends negatively with minor significance but high variability remains for the full time series. While CDF curves reveals that almost three quarters of days were hotter and wetter than average and 20% were considered as extremely rainy days.

Fig. 5
figure 5

Trends and CDF curves for Constantine Station. a Temperature, b precipitation, and c temperature

Full size image

Figure 6 depicts the Setif station analysis results. Both temperature and precipitation have a positive trend at a slope equal to 0.03 and variability increasing from 1990 to 2005. Both CDF curves reveal normal distributions since there is no kurtosis—the median equals the average. Less than 5% of this time series showed extreme results.

Fig. 6
figure 6

Trends and CDF curves for Setif Station. a Temperature, b precipitation, and c temperature CDF

Full size image

Tebessa station represents a significant positive trend in temperature with moderate variability; on the other hand, precipitation illustrates a minor positive trend with major variability. CDF curves reveal that 75% of time series of temperature, and 87% of received precipitation were above the average, where 10% of those days reached extreme levels (Fig. 7).

Fig. 7
figure 7

Trends and CDF curves for Tebessa Station. a Temperature, b precipitation, c temperature CDF, and d precipitation CDF

Full size image

5 Results and Discussion

Before any infrastructure construction, risk assessment and return period (or average frequency of occurrence) should be assessed and taken into consideration. The classical risk is the probability of dangerous event occurrence only once during the entire life of the infrastructure. In this paper, it is proposed that the risk calculations should additionally be based on the slope of possible trends in temperature and precipitation records over the life span.

To calculate the risk under climate change effects, a climate factor α (the slope of the trend component within the hydrometeorological time series) needs to be added to the numerator of the classical risk formula. If α = 0, there is no climate change variation trend, otherwise α takes positive (increasing) and negative (decreasing) values.

Tables 2 and 3 represent the annual precipitation records at each meteorological station and the statistics of both cases with and without climate change effect for different return periods. The relative error, β, is calculated by taking into consideration precipitation (or temperature) levels under the effects of climate change Ccc and with no climate effect, P, according to the following expression:

$$ \beta = 100\left( {\frac{{C_{\text{cc}} {-} P}}{{C_{\text{cc}} }}} \right). $$
(3)
Table 2 Climate impact on precipitation features
Full size table
Table 3 Climate impact on temperature features
Full size table

Concerning the precipitation parameter, Table 2, note that Setif and Tebessa stations present the higher positive levels of relative error percentage mean that the impact of climate change on these two stations is more severe than for the others, followed by Bejaia and Bordj-Bouarriridj station with least affect for Annaba station.

All stations show increasing precipitation. However, Constantine station presents a negative relative error percentage reflecting decreasing precipitation under climate change effect while all other stations show an increasing effect.

Table 3 shows the results appropriate for temperature. The β value for Annaba and Bordj-Bouarriridj stations during all return periods has a negative sign (−), which means that these two stations are subject to lower annual average temperature, whereas other stations, namely Tebessa, Constantine, Setif and Bejaia have positive β, implying that climate change will increase their temperatures with Tebessa increasing most.

After calculating climate change risk level from Eq. (2), one can use Eq. (1) to find the new return period, incorporating the climate change effect. For this case, the relative error percentage (γ) is calculated similar to Eq. (3) as

$$ \gamma = 100\left( {\frac{{C_{R} {-} R}}{{C_{R} }}} \right), $$
(4)

where CR and R are the return periods with and without climate change effect, respectively, leading to results in Tables 4 and 5.

Table 4 Return period (year) comparison for precipitations
Full size table
Table 5 Return period (year) comparison for temperatures
Full size table

Table 4 reveals that only Constantine station has a positive relative error percentage (γ), which means that the return period for extreme levels of precipitation including climate change considerations is longer than without climate change impact. The greater the return period increase, the longer it extends. In contrast, at the other five stations, shorter return periods prevail implying that rare events will occur faster due to climate change.

Table 5 expresses the return periods including climate change impacts. Annaba and Borj-Bouarriridj stations expect longer return periods, while Tebessa, Constantine, Bejaia and Setif anticipate shorter return periods for extreme temperature occurrence.

6 Conclusion

It is well known that precipitation and temperature have been steadily increasing all around the world, and especially, in the Mediterranean basin over the last 3 decades. The most important concept in any water resource system design is the risk associated with hydrometeorological events. The purpose of this study is to analyze trends in temperature and precipitation records and incorporate these into risk assessments by incorporating the changing climate (temperature and precipitation) for the Northeast of Algeria over the 35-year period (1982–2017). For this purpose, a climate change factor is incorporated into the classical formulation, based on the trend slope in historical records. The risks associated with 2-, 5-, 10-, 25-, 50-, 100-, 250-, and 500-year return periods are calculated for each station. The study indicates that climate change impact is expected to be more influential at all locations except Constantine where the risk of extreme precipitation appears likely to diminish. Temperatures in Tebessa, Constantine, Setif and Bejaia are expected to reach more frequent extremes.

Infrastructure that has been planned for construction in Annaba, Tebessa, Bejaia, Borj-Bouarriridj and Setif that use precipitation data should take into consideration the impact of climate change on extreme weather return periods. In the same light, the temperature data considerations should follow the same approach for Tebessa, Constantine, Bejaia and Setif cities.

Positive trends in temperature and precipitation records appear in most stations with varying degrees. For example, in Constantine and Tebessa, a significantly increasing trend for temperature is indicated with a slope 0.05, while Bejaia and Setif have less significant trends with 0.03 slopes. Additionally, Annaba and Bordj-Bouarriridj show a decreasing, but not significant trend. As for the precipitation records, all stations indicate positive, but not significant trends, except Constantine station which tends to decrease with little significance. The cumulative distribution function (CDF) curves reveal that Annaba, Bejaia, Borj-Bouarriridj, Constantine and Tebessa have more than 70% of their days being hotter and wetter than average with only the Setif station having equally likely temperatures and precipitations above and below the average. This study confirms that climate change has and will continue to have impacts on temperature and precipitation and should be taken into consideration in the calculations of all infrastructure constructions.