Introduction

Floods are natural phenomena that cause property loss and casualties over the last decades (Yalcin 2020). It is probably the most widespread, devastating, and prevalent disaster threatening human communities (Teng et al. 2017). Recently, destructive floods have become more recurrent and intense. This tendency is probably to be increased in the future owing to climate warming and ever-increasing anthropic activities on river floodplains (Yalcin 2018).

The urban commune of Oued Laou is becoming more and more vulnerable in terms of flash flood inundation as a result of the continuous evolution of urbanization. To face and mitigate the human and socio-economic impacts that have serious consequences (Estupina Borrell 2004), reliable modeling of flood extents, water depths, flow velocity, and discharge distribution in urbanized areas prone to floods is required (Mignot et al. 2019). Many researchers since 1970 have considerably ameliorated the performance of flood modeling. The models are developed to simulate flooded areas (Baky et al. 2012), to generate flood risk mapping, and have been used to assess flood damage (Merz et al. 2010), as well as having served as a relevant tool for the calculation of flash flood parameters. For this reason, a one-dimensional hydrodynamic model was proposed for this research owing to its computational efficiency and its extensive application for flood simulation (Army Corps of Engineers 2016; Finaud-Guyot et al. 2011; Mashriqui et al. 2014).

Moreover, many scientists have recently mentioned different usabilities of the 1D model. For instance, it can be employed to simulate open surface floodplain runoff (Teng et al. 2017) and to regenerate the magnitudes of paleofloods (Carling et al. 2010). Besides interpolating river stages generates a flood inundation map by mean of GIS-based techniques (Werner 2001), helps hydraulic engineers in streamflow analyses and floodplain modeling (Logah et al. 2017), and calculates water surface elevation along a river in an unsteady and steady flow routine inclusive of sediment transport modeling (Zainalfikry et al. 2019). This latter permits the reckoning of the fluid dynamic field parameters of a current (Doronzo and Dellino 2010).

The arrival of Geomatic tools and typically Geographic Information Systems (GIS) recently has highly enhanced the capability of flood models. In this research, the association of HEC-GEORAS, HEC RAS, and GIS information is employed for a preliminary analysis of the spatiotemporal data already available in the catchment area, and then delineate the areas susceptible to be flooded, to proceed with a simulation for three return period events.

Study area

This research focuses on the river reach that covers the areas downstream of the Oued Laou catchment. The entire basin covers approximately 928 km2 and its perimeter (P) = 182 km. It is a coastal basin located in the north-west of Morocco, at 45 km NR16 of Tetouan city, on the borders of Tetouan-Chefchaouen Provinces. It was formed following the union between three main tributaries, including Oued Farda, Oued Haram, and Oued Dardara, with a well-defined drainage network and deeply ravenous soils. According to Strahler’s classification, the catchment is of order 6, and the length of its mainstream, is approximately 58 km (Fig. 1).

Fig. 1
figure 1

Map layout of the study area and the hydrographic network of the river basin of Oued Laou

Full size image

Morphological context

To assess the effects of morphologic details and the geometric structure on flood inundation maps (Lamichhane and Sharma 2018), it is necessary to define the topographic predisposition of the watershed, insofar as the morphologic data influence flow conditions (Martí et al. 2019) and hydraulic simulations through geometric discretization used in a model (Cook and Merwade 2009). Oued Laou catchment has a Gravelius compactness index KG = 1.67. It is characterized by its abrupt slopes and steep reliefs, where altitudes vary between 0 and 2122 m (culminating point). The average altitude point is of 920 m. As Fig. 2 shows, the most represented altitude range (450–650 m) is covering an area of 179 km2. Steep slopes are predominant and mainly distributed in the middle, south-east, and north-west of the basin river following the high altitudes, while the extreme north and south (outlet) are characterized by gentle slopes and flat landscapes (Fig. 3).

Fig. 2
figure 2

Hypsometric map of the Oued Laou catchment

Full size image
Fig. 3
figure 3

Slope map of the Oued Laou catchment

Full size image

Geological context

The geological history of the Rif chain has been a determining factor in the definition of the actual study site reliefs. In a north-southwesterly oriented section, different structural assemblages (Fig. 4) were identified by (Kornprobst 1974) and which constitute the internal Rif (Rkiouak et al. 1997). From the north to south, these complexes are respectively Ghomarids and Sebtides constituted by a Paleozoic metamorphic material (Benmakhlouf and Chalouan 1992), the Limestone Dorsal whose age ranges from Triassic to Oligocene (Maate et al. 1993; Hlila and Sanz De Galdeano 1994) and finally, the series of flysch which are mainly Cretaceous.

Fig. 4
figure 4

Map of the structural units in the Oued Laou catchment

Full size image

From a lithological standpoint (Fig. 5), the internal domain was represented by the most resistant facies (limestones, dolomites, and calcaro-dolomitic alternations), the external domain consists mainly of highly erodible marls and clays, and the moderately resistant rocks belong to the flysch nappes. The latter comprises an alternation of more resistant massive sandstone benches and very weathered and deformable clay mudstone benches (Chalouan et al. 2006; In Moratti and Chalouan 2006).

Fig. 5
figure 5

Lithological map of Oued Laou catchment (modified from the map of Talambote, Bab Taza, Bni Hassan 1/50000 and 1/100000 Rif map)

Full size image

Under the same heading (Fig. 6), multiple and very diverse outcrops were identified in the study area. Yet, they have been classified according to their permeabilities into two assortments: the semi-permeable outcrops attributed to the carbonated massive of the limestone ridge thanks to its fracturing network and karstification. The rest of the outcrops are considered impervious due to the fine and compact texture of their materials (clay, marls, sandstones ...). This latter intervenes in the fraction’s distribution of the water flow rate received on the watershed surface.

Fig. 6
figure 6

Permeability map of the Oued Laou catchment

Full size image

Climate context

The climate is a subhumid Mediterranean climate where one can find many types of precipitation such as rainfall, hails, and snows on the tops of high altitudes, which contributes to groundwater recharge and increases deep runoff. The average monthly precipitation (Fig. 7) shows that the wettest month of the year is December with maximum precipitation of 112 mm and the driest month is July with a minimum of 1 mm. Besides, the average annual precipitation shows obvious irregularity regarding the distribution of recorded peaks that has resulted from the alternation of wet and dry years, with a maximum of rainfall recorded in 2009–2010 and a minimum in 2011–2012 (Fig. 8).

Fig. 7
figure 7

Monthly mean precipitation during the period 1969–2017

Full size image
Fig. 8
figure 8

Mean annual precipitation (1969–2017)

Full size image

The average annual inflow including direct and deep runoff is also irregular (Fig. 9) with a maximum discharge of 900 m3/s recorded in December 2009. Such volumes are almost common and result in the lateral transfer of runoff out of the riverbed, namely floods. The overflows normally reach an area of 5.5 km2, which represents the third of the total area of the plain.

Fig. 9
figure 9

Average annual discharge (m3/s)

Full size image

Working methodology (Fig. 10) and development of the hydraulic model

Data requirement

The implementation of the 1D hydraulic model for flood simulation requires the estimation of certain input quantities, such as meteorological data, which is frequently rare (Boulariah et al. 2019); these quantities should be measured in the field. Consequently, all the necessary measures regarding this research were carried out at the level of the Koudiat Kourriren hydraulic station. These measurements are flow rates at a given instant (t) located at the gauging station (x = 520.050, y = 528.950, z = 30.00) during 22 years (1990–2012). The hydrologic data were delivered by the Hydraulic Agency of the Lokkous Basin. Based on the latter datasets, a frequency analysis for flood inundation employing the annual maximum daily flows has been used in order to estimate the period of return of extreme flood events.

Fig. 10
figure 10

The working methodology

Full size image

Topographic datasets are required to produce geospatial data and to implement hydraulic analysis in HEC-RAS (Lamichhane and Sharma 2018). Therefore, accurate datasets were used in this research including the Digital Elevation Model (DTM of 12.5 m) and satellite image (imagery about 30 cm of resolution delivered by ArcGIS). The combination of these two Rasters leads to minimizing the error resulting from DTM while permitting better and realistic discretization of the river channel.

The modeling process

The one-dimensional method of hydraulic modeling is regarded as computationally effective to simulate the floodplain and flood extents, typically for extreme flows, regardless of restricted elevation data (Horritt and Bates 2001). For this reason, a digital elevation model (TIN) and satellite image were first prepared and associated on the Arc-Gis platform. Using the HEC-GeoRAS interface, geometric details of feature classes such as cross-sections, bank stations, and streamlines were created in HEC-GeoRAS and then exported to HEC-RAS. They were created to extract altitude data from the TIN. They can also provide the location of the “bank stations” (Ben Khalfallah and Saidi 2018). Second, the geometric skeleton of the river was exported to HEC-RAS. Meanwhile, the hydrologic datasets, including annual maximum daily flow rate records (Table 1), were selected and adjusted via 6 probabilistic distributions (Gamma, Weibull, Exponential, GEV Log-Pearson type 3, Gumbel, and Pearson type 3) in order to select the most appropriate distribution fitting the selected observations. These latter were also integrated into the HEC-RAS model process.

Table 1 Annual maximum daily discharge at Koudiat Kourriren station
Full size table

Calibration, verification, validation, and the export of the results obtained depend on the type of flow. The boundary conditions must be applied downstream or upstream of the river reach under study; for the “subcritical” flow, the critical section is located downstream (Javaheri and Babbar-Sebens 2014). The processes calculation outcomes such as water surface elevations, velocity, and flood extents carried out on HEC-RAS were then exported back to HEC-GeoRAS to produce floodplain maps.

Results and discussion

The choice of the most fitting frequency distribution function for the discharge observations has first done graphically (Fig. 11). Then, it was confirmed numerically based on statistical criteria (El Alaoui El Fels and El Mehdi Saidi 2014). In other words, the extreme flow series were adjusted through various probabilistic models. The model that shows maximum fitness between the measured flow rates and estimated flow rates was selected to be used for statistical inference.

Fig. 11
figure 11

Results of the adjustment by the six different frequency distributions for Koudiat Kourriren station

Full size image

The graphical adjustment has shown almost similar results to such an extent that it was difficult to differentiate the most suitable frequency distribution at this level, especially to both distinguish between the Weibull and Gamma graph distribution. To avoid confusion and choose the optimal fitting probability distribution accurately, two criteria of measurement and verification of the statistical models fit quality were summoned. The probabilistic distribution that reveals the lowest values of both criteria is the most suitable for the extreme flow predetermination (El Alaoui El Fels and El Mehdi Saidi 2014). These two criteria are (AIC) the Akaike information criterion (Akaike 1973) and (BIC) Bayesian information criterion (Schwarz 1978). Their formulas are defined as follows:

$$\mathrm{AIC}=-2\mathrm{Log}\ (L)+2\ \mathrm{k}.$$
$$\mathrm{BIC}=-2\mathrm{Log}\ (L)+\mathrm{k}\ \mathrm{Log}\ (n).$$

where k is the number of the statistical model parameters. L is the optimized value of the likelihood function for the estimated model and n is the sample size.

Table 2 represents the results of the AIC and BIC for each selected distribution. As it is shown, the quality assessment of the frequency distribution depends on the values of the previous criteria. The comparison of the values supplied by AIC and BIC allowed at holding among the 6 distributions, the most reliable law for the adjustment. The results reveal that the statistical distribution that is most suitable for the studied variables is Gamma distribution. It is the one that shows the lowest value of both criteria (AIC = 293.576; BIC = 295.758), and consequently the maximum fitting.

Table 2 Comparison of BIC and AIC for annual maximum daily discharge
Full size table

Table 3 provides the frequency analysis of the extreme flows calculated by the gamma statistical model in m3/s with their return periods and their probability of not exceeding.

Table 3 Simulation results of the flow rates adjusted by Pareto distribution
Full size table

The average flow in the main channel of Oued Laou is about 12.15 m3/s. The maximum flow estimated by the 10-year flood is 640 m3/s, and the centennial peak reaches 1320 m3/s. These high volumes of water are considered to be damaging, especially in a vulnerable environment such as the Oued Laou plain.

Mapping and delineation of flood-prone areas

The choice of the river reach is certainly opted for the downstream part of Oued Laou, between the gauging station of Koudiat Kourriren (which is located in the extreme south of the segment) and the outlet in the extreme north. It is the sector that illustrates both the most probable existence of the hazard and assets (Fig. 12).

Fig. 12
figure 12

Map of the location of the studied reach and the Koudiat Kourriren station

Full size image

Among all recurrence intervals of flood scenarios resulting from the frequency analysis of flow observation records, three events related to different periods of return were selected to simulate the spatial distribution of flood flows. The first event scenario is a real flood hydrograph recorded on December 24, 2009. The two other flood scenarios correspond to a 10-year flood interval scenario and a 100-year flood scenario. The simulation was carried out with a one-dimensional model using the HEC-RAS software. It was run under steady flow conditions for each recurrence-interval flood.

The simulation of flood parameters such as water depth (Figs. 13, 14, and 15) and flood velocity (Figs. 16, 17, and 18) highlights significant overflows along the floodplain, especially the 100-year flood event that shows an increasingly worsened spatialization of the flood zone with potentially devastating effects on anthropized environments.

Fig. 13
figure 13

Spatial distribution of water extent and water elevation in the Oued Laou watercourse for a real flood occurred on December 24, 2009

Full size image
Fig. 14
figure 14

Spatial distribution of water extent and water elevation in the Oued Laou watercourse for a T = 100-year flood recurrence interval

Full size image
Fig. 15
figure 15

Spatial distribution of water extent and water elevation in the plain of Oued Laou for a T = 10-year flood recurrence-interval

Full size image
Fig. 16
figure 16

The distribution of the flood flow velocities in the Oued Laou plain for a real flood occurred on December 24, 2009

Full size image
Fig. 17
figure 17

The distribution of the flood flow velocities in the plain of Oued Laou for a T = 100-year recurrence interval

Full size image
Fig. 18
figure 18

The distribution of the flood flow velocities in the Oued Laou plain for a T = 10-year recurrence interval

Full size image

Conclusion

Flood risks can prompt dramatic consequences in terms of property losses. Several agents can aggravate the phenomenon. These agents can be related either to the topography of the landscape or to the anthropogenic activities such as urbanization and agriculture: the case of our steady area. This research aims to develop a hydraulic model based on the Saint-Venant equations in order to define the overflows extent, velocity, and flood flow distribution over the entire study area. The simulation result allows concluding that:

  • The characteristics of the environment, such as morphology and the nature of the land are considered to be a determining factor that inherently controls the process and mechanism of the flow geneses and the propagation of flood flows. In a flat land such as Oued Laou plain, floods cover almost the entire plain area. Upstream, the velocity and depth are high where the channel is deep and slopes are steep, and roughness is low. While going towards the downstream, these variables gradually reverse.

  • The performance and the quality of the model outcomes depend on the quality of the input datasets (the resolution of the digital terrain models (TIN or DEM)). The calibration helps to adjust and improve the simulation, but it remains insufficient. However, the 1D molding still provides accurate flood maps for areas lacking data.

  • Three flood maps have been generated, the first one related to a real flood of 2009, which reaches a flow of 900m3/s), and the two other floods have been obtained from the statistical model. The first is a 10-year flood (640 m3/s), while the second is 100-year flood (1320 m3/s). In terms of damage, the 10-year flood is considered to be a frequent flood that does not show significant damages, except for some crop fields but it remains controllable, whereas, the 100-year flood shows more black spots in terms of buildings, road, or crop fields.