Analysis and Quantification of Water Erosion in Northern Algerian WatershedsS

GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 ANALYSIS AND QUANTIFICATION OF WATER EROSION IN NORTHERN ALGERIAN WATERSHEDS Mahieddine BELLOUT 1 , Djamel BOUTOUTAOU 1 1 University Kasdi-MERBAH Ouargla, Laboratories Exploitation and Valorization of Natural Resources in Arid Zones, Ouargla, Algeria E-mail: [email protected] ABSTRACT Algeria is characterized by a semi-arid climate, particularly vulnerable to erosion of agricultural land, where physical, hydroclimatic, geomorphological and socio-economic conditions are highly favourable to the onset and acceleration of this phenomenon. High concentrations of suspended sediments transported by rivers to dams and reservoirs represent a significant problem for water management and the sustainability of these infrastructures. During this study, solid transport data were collected from the Agence Nationale des Ressources Hydrauliques database. The data set includes a total of 132 hydrometric stations throughout the country. Exploiting pairs of instantaneous liquid flow measurements (m³/s) and instantaneous solid flow measurements (kg/s) has led to the establishment of regressions at different time scales, thus concluding that the power model is the most appropriate based on the coefficient of determination R². The average annual specific erosion varies from one watershed to another, generally between 11,75 and 5978,34 T/km².year. The principal component analysis (PCA) method was used to study the average monthly solid discharges of 132 hydrometric stations, and the results obtained highlight the presence of four hydrologically homogeneous groups. Multiple regression was performed on the four groups to highlight a potential relationship between the dependent variable, specific erosion, and other explanatory variables. The correlations indicate that each group is influenced by parameters distinct from the others, as in the case of group A, where the correlations between specific erosion, on the one hand, and the other hand, the Average slope of a watershed (Im), lithology index (IL), runoff coefficient (RC), and the normalized difference vegetation index (NDVI) are significant. Keywords: Water Erosion; Regionalization; Watersheds; Algeria; Solid Transport. 1 INTRODUCTION Water erosion is a major environmental problem that affects many regions of the world, particularly agricultural areas. Since the 1930s, scientists have begun to study this phenomenon in depth, which has allowed for a better understanding, quantification, and modelling of erosion in different environments. It is a geographical and environmental phenomenon whose impact and severity vary considerably from one site to another. This variability depends on several natural and human factors that influence the intensity of the process and its consequences. This phenomenon is a characteristic of the Maghreb region, whose water and soil potential are seriously threatened [19;17;1;15]. In Morocco, the annual cumulative land losses are estimated at 100 million tons; in Tunisia, water erosion totals 8.5 million hectares, representing 52% of the country's total area [15]; in Algeria, 45% of the Tellian region is affected, which amounts to 12 million hectares; approximately 6 million hectares are currently undergoing active erosion [20]. Algeria has extreme climatic conditions characterized by significant spatial and temporal precipitation variability. In autumn, it is expected to observe intense precipitation, often reaching intensities exceeding 45 mm/h, especially combined with often insufficient vegetation cover and sometimes inadequate management, causing severe floods accompanied by a rapid rise in waters carrying high suspended matter concentrations. This leads to significant consequences both upstream and downstream, affecting ecosystems, infrastructure, and populations, making this phenomenon particularly concerning. Generally, the 2 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 average annual specific erosion ranges between 2 000 and 4 000 t/km² [6]. The annual loss of storage due to sedimentation in dams is estimated at around 20 million cubic meters [13]. The modelling of the relationship between liquid flow and solid flow is crucial for anticipating the evolution of sediment transport in a watershed, particularly at its outlet or even in the reservoirs of downstream dams [10]. Different models can be used to explain this relationship. Among these models, the predominant one consists of a power function that establishes a link between the concentration of suspended sediments and the water flow rate [14]. Many studies have attempted to describe specific erosion based on various parameters and the hydrological characteristics of the watershed [11;22;21;16;5;4;6]. However, this work needs to be improved due to the need for more quality data. It is based on a regional approach to estimate specific erosion, considering each watershed's specificities and using adapted models and locally collected data. This study aims to establish regressions between liquid flow and solid flow at different time scales for 132 hydrometric stations, quantify solid transport and assess specific degradation in these rivers without measurement data. This requires adopting an innovative methodology that combines numerical models, remote sensing techniques, spatial data and empirical approaches. 2 METHODS 2.1 Presentation of the study area Our study area is located in the northern part of Algeria, between longitudes «2° and 9°» West and East, respectively, and latitudes «33° and 37°» North, covering an area of around 365 000 km2. It extends over a width of around 35 myriameters and 100 myriameters along the coast. Morocco and Tunisia form the western and eastern boundaries, respectively, the Mediterranean Sea, the northern boundary and the southern flanks of the Saharan Atlas, the southern boundary. It is characterized by a Mediterranean climate in its northern part and a sub-desert climate in its southern part (see Figure 1). Figure 1. Location of the study area 2.2 Presentation of data Collecting and formatting data is the initial phase of any statistical study: Duband (1989) states without exaggeration that this represents 30 to 50% of the work. Ambroise (1998) shows that applying any mathematical model presupposes prior knowledge of the data needed to develop it. 3 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 The data used are instantaneous solid flow values, expressed in kg/m3, derived from instantaneous liquid flow values, expressed in m3/s, multiplied by the suspension concentration, expressed in g/l, measured at the catchment outlets. There are two methods for measuring liquid flow: based on the tarage curve from the water heights recorded on a limimetric scale on the one hand, and on the other by depriving the water levels recorded by a float limnigraph. For each liquid flow measurement, a measure is performed to assess the load of suspension material, which is obtained from a sample of water taken on the banks of the water stream and then dried to have a load concentration in g/l; the solid flow is then deducted by the product of the concentration by the corresponding liquid flow Ql. These measurements are carried out at the watershed monitoring station, and the sampling frequency varies depending on the hydrological regime; it is intensified during rainy periods and periods of high loads up to ten minutes apart. The collection of this data is the responsibility of the Algerian National Water Resources Agency (ANRH). Figure 2 illustrates the spatial distribution of the 132 hydrometric stations collected in northern Algeria after these coordinates have been corrected. Figure 2. The spatial distribution of the hydrometric stations used 2.3 Hydromorphometric characteristics of collected watersheds Conventional techniques used to study the physical complex of a watershed are mainly based on manual methods, the results of which are generally unreliable. 4 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 Because of the systematic errors in the boundaries of the watersheds collected, and with the advent of new tools such as GIS and remote sensing, it has become easy to determine the shape, relief and typology of a hydrographic network. Therefore, we were obliged to delimit these watersheds to correct the existing errors. 2.4 Analysis method 2.4.1 Establishment of regression between liquid and solid flow The data from the instantaneous values of solid and liquid flows were processed on different time scales: daily, monthly, annual, seasonal (autumn, winter, spring, and summer), wet season, and dry season, for all stations, in order to establish regression models and to have an initial idea of the dynamics of solid transport, specifically the liquid flow-solid flow relationship. The main regressive models are: − The linear model: 𝑌 = 𝑎𝑋 + 𝑏 − The parabolic model: 𝑌 = 𝑎𝑋2+ 𝑏 𝑋 + 𝑐 − The power model: 𝑌 = 𝑎 𝑋𝑏 − The exponential model: 𝑌 = 𝑎 𝑒 𝑏𝑋 − The logarithmic model: 𝑌 = 𝑎 𝑙𝑛𝑋 + 𝑏 2.4.2 Annual solid input The annual flow of suspended solids exported by the various rivers studied is calculated using the formula: 𝐴𝑠=∑(𝑡𝑗+1 −𝑡𝑗)𝑄𝑗 𝑁 𝑗=1 𝐶𝑗 (1) Where: 𝐶𝑗 is the concentration (g/l) measured at time 𝑡𝑗 corresponding to liquid flow 𝑄𝑗 (m3/s), 𝑁 is the number of samplings carried out over the year in question, 𝑡𝑗+1 −𝑡𝑗 is the time step separating two consecutive samplings. 2.4.3 Solid transport modelling The main objective of the modelling is to develop a model of specific erosion as a function of various climatic, hydromorphometric, geological and biophysiographic parameters (topography, vegetation cover, etc.) in t/km2.year, in order to be able to generate a spatial model that will allow us to have the specific erosion at any point of the treated watershed. 3 RESULTS AND DISCUSSION 3.1 Characteristics of the basins at gauging stations The hydromorphometric characteristics of some of the watersheds selected for this study are shown in Table 1. 5 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 Table 1. The hydromorphometric characteristics of some watersheds Features Settings Symbol Unity Station codes 01 19 05 02 03 23 03 16 01 04 04 03 05 08 01 07 05 01 09 05 01 10 01 09 11 01 01 Physical characteristics Surface S km2 428,22 53,84 675,78 107,95 334,39 763,41 3615,56 927,31 961,47 Perimeter P km 122,31 39,42 153,51 72,80 120,78 167,60 390,16 178,41 185,57 Compactness index Kc 1,66 1,50 1,65 1,96 1,85 1,70 1,82 1,64 1,68 Length of equivalent rectangle L km 53,09 16,43 66,61 33,14 54,22 73,40 174,34 77,19 80,90 Equivalent rectangle width L km 8,07 3,28 10,15 3,26 6,17 10,40 20,74 12,01 11,88 Features of the relief Minimum altitude Hmin m 330 12 33 140 557 879 115 396 927 Maximum altitude Hmax m 1 786 665 1 189 1 042 1 862 2 319 1 804 1 695 1 452 Average altitude Hmoy m 781,63 250,34 398,53 467,39 952,73 1 272,81 748,21 907,31 1 165,75 Altitude corresponding to 5% of total surface area H5% m 1 255 476 829 746 1 448 1 762 1 130 1 274 1 276 Altitude corresponding to 95% of total surface area H95% m 426 44 71 204 624 930 332 573 1 032 The slope indices Overall slope index Ig % 1,56 2,62 1,13 1,63 1,52 1,13 0,45 0,9 0,3 Rock slope index Ip % 4,64 5,84 3,82 4,67 4,32 3,87 2,63 3,64 2,15 Average slope of a watershed Im % 23,43 24,15 19,05 16,71 15,51 16,03 20,19 22,85 7,44 Specific gradient Ds m 323,12 192,89 295,84 169,91 277,89 313,18 275,22 276,54 93,52 Network settings hydrographic Length of main watercourse Lcp km 44,85 18,36 49,52 31,64 40,15 69,63 175,35 70,28 66,38 Average slope of main river Ῑ % 2,16 1,99 2,05 1,74 2,25 1,68 1,01 1,28 0,93 Order O 7 5 7 5 6 7 8 7 7 Drainage density Dd km/km2 3,81 3,03 3,40 3,05 3,47 3,41 3,13 3,17 3,53 Torrentiality coefficient Ct 32,51 17,26 21,40 17,10 19,12 19,49 17,55 18,30 20,76 Confluence ratio Rc 1,88 1,73 4,11 6,15 1,90 1,89 1,81 1,88 1,94 Length ratio RL 0,97 0,94 0,92 0,92 0,96 0,97 0,96 0,96 0,92 Concentration time Tc H 8,83 4,61 11,66 2,19 8,38 13,54 25,01 12,56 18,09 6 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 The watersheds in our study area present a remarkable morphological diversity, with slope, basin shape and topography directly influencing water management and erosion risks. The most minor surface area watershed is at the SIDI BEKAI gauging station (5,52 km2). The largest surface area corresponds to the watershed at the SIDI BEL ATTAR gauging station (43 952,76 km2). On the other hand, the compactness index is more significant than one, varying between 1,19 and 2,49. Most of these watersheds have a relatively high to high relief, making them particularly sensitive to erosion. The length of the main rivers varies between 4,53 and 743,44 kilometres. We found that concertation times ranged from 2,43 to 89,26 hours. As an illustration, the following figures (Fig 3, 4, 5, 6, 7). show some hydromorphometric characteristics of some of the watersheds studied. Figure 3. Shape map of watershed 160402 Figure 4. Hypsometric map of watershed 140602 Figure 5. Slope map for watershed 150702 Figure 6. Elevation map for watershed 050901 7 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 Figure 7. River order map for watershed 070702 3.2 Liquid and solid flow relationships For all hydrometric stations at different time scales, according to Dagnellie (1992) referring to the coefficient of determination R², the power model is the most representative, as demonstrated by various studies conducted in Algeria [6;20;4;3]. As an illustration, figure 8 and table 2 represents the relationship between liquid and solid flows at different time scales of some watersheds in the study area. Table 2. Adjusted models for the different temporal scales and coefficients of determination Temporal scales Selected model Variation of parameter a Variation of exponent b Example for some watersheds Station code Determination coefficient (R 2) Relationships retained Interannual Power: Qs =a Ql b 0,08 - 85,39 0,79 - 1,91 013301 0,8462 Q(s) = 1,5512Q(l)1,5905 111003 0,8370 Q(s) = 3,8633Q(l)1,226 160202 0,8188 Q(s) = 0,817Q(l)1,6391 Daily 0,02 - 46,46 0,93 - 2,68 040220 0,8608 Q(s) = 3,3858Q(l)2,2042 050801 0,8977 Q(s) = 8,6642Q(l)1,1811 110201 0,8638 Q(s)= 1,0132Q(l)1,8457 Monthly 0,042 - 36,54 0,78 - 2,22 120309 0,8921 Q(s) = 9,0018Q(l)1,2934 160726 0,7932 Q(s) = 0,3257Q(l)1,1612 013001 0,8636 Q(s) = 3,8449Q(l)1,475 Annual 0,03 - 50,69 0,84 - 2,06 012701 0,8437 Q(s)= 49,003Q(l)1,0741 021126 0,8774 Q(s) = 7,3662Q(l)1,0781 051101 0,9051 Q(s) = 12,594Q(l)1,2501 Fall 0,04 - 36,54 0,78 - 2,22 140602 0,8430 Q(s) = 1,503Q(l)1,855 111201 0,9012 Q(s) = 6,9786Q(l)1,3151 120309 0,9560 Q(s) = 8,7911Q(l)1,4071 Winter 0,04 - 36,54 0,78 - 2,22 150703 0,9214 Q(s) = 4,405Q(l)1,1417 030310 0,8309 Q(s) = 2,1859Q(l)1,3114 061403 0,9094 Q(s) = 20,088Q(l)0,7837 Spring 0,03 - 50,69 0,84 - 2,06 070401 0,9004 Q(s) = 4,5536Q(l)1,3204 100701 0,7771 Q(s) = 1,7525Q(l)1,1558 8 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 111501 0,8399 Q(s) = 2,3297Q(l)1,3072 Summer 0,07 - 59,33 0,89 - 2,09 061801 0,8976 Q(s) = 8,9429Q(l)1,3026 070403 0,8445 Q(s) = 21,141Q(l)1,0079 111403 0,9303 Q(s) = 5,9639Q(l)1,3177 Dry season 0,03 - 17,51 0,86 - 1,88 013402 0,8756 Q(s) = 14,802Q(l)1,2108 111425 0,8766 Q(s) = 16,513Q(l)1,4812 160611 0,8381 Q(s) = 11,532Q(l)1,6485 Wet season 0,09 - 48,79 0,94 - 2,12 021201 0,8583 Q(s) = 1,9756Q(l)1,2682 150106 0,8364 Q(s) = 0,4842Q(l)1,4108 050901 0,8820 Q(s) = 3,8024Q(l)1,4696 When expressing solid flows about liquid flows on an interannual scale, there is a very high dispersion of points, with values ranging from 0,08 to 85,39 for component a and 0,79 to 1,91 for exponent b. This dispersion can be explained by the fact that several factors control concentrations. On a daily scale, the parameter a fluctuated from 0,02 to 46,46, while the exponent b also varied between 0,93 and 2,68. Based on a monthly scale, it is possible to observe a decrease in the dispersion of points, characterized by a variation in the parameter a from 0,042 to 36,54 and in the exponent b from 0,78 to 2,22. The relationships on this scale are exciting and can be used in various studies. On the annual scale, the dispersion of the points is not significant, with the parameter a and the exponent b varying respectively from 0,03 to 50,69 and from 0,84 to 2,06. Furthermore, just like monthly relationships, these relationships are exciting and can be used in various studies. On a seasonal scale, the parameter a and the exponent b have different ranges of variation from one season to another, as each season has a rather distinct hydrological behavior. Autumn is characterized by precipitation on relatively dry soil, which promotes erosion and leads to significant solid concentrations. During the winter, the soil is relatively moist, making it more resistant to erosion, except when significant liquid inputs lead to considerable solid contributions. During the spring season, vegetation enhances the soil's resistance to erosion and often leads to a decrease in solid particle concentrations. During the summer, most waterways are dry; however, solid concentrations can be remarkable during floods. y = 1,5512x1,5905 R² = 0,8462 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 050 100 150 200 250 300 350 solid flow measurement (Kg/s) liquid flow measurement (m3/s) hydrometric station 01 33 01 (Interannual scale) y = 3,39x1,3206 R² = 0,901 0 500 1 000 1 500 2 000 2 500 0 5 10 15 20 25 30 solid flow measurement (Kg/s) liquid flow measurement (mᵌ/s) hydrometric station 07 04 01 (daily scale) 9 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 Figure 8. Relationship between liquid flows and solid flows at different time scales 3.3 Monthly distribution of solids input The histograms illustrating the monthly distribution of solid inputs (Figure 9) show significant temporal and spatial variation in solid transport. The analysis of these monthly values shows that the amount of sediment transported throughout the year varies monthly and from one watershed to another. The solid transport in autumn remains the highest for most watersheds, far surpassing other seasons, due to several combined factors that promote erosion and intensify sediment transport, such as intense rainfall generating rapid runoff that carries away large quantities of fine sediments. y = 16,513x1,4812 R² = 0,8766 0 1000 2000 3000 4000 010 20 30 40 solid flow measurement (Kg/s) liquid flow measurement (mᵌ/s) hydrometric station 11 14 25 (seasonal scale -Dry season-) y = 4,262x1,2196 R² = 0,858 0 1000 2000 3000 0 20 40 60 80 100 Débit solide mesuré (Kg/s) liquid flow measurement (mᵌ/s) hydrometric station 07 07 02 (seasonal scale -Wet season-) y = 2,5196x1,1781 R² = 0,8192 0 5 000 10 000 15 000 20 000 25 000 0 200 400 600 800 1 000 solid flow measurement (Kg/s) liquid flow measurement (mᵌ/s) hydrometric station 02 12 01 (monthly scale) y = 0,0793x1,6109 R² = 0,9681 0 5 000 10 000 15 000 20 000 25 000 30 000 35 000 01 000 2 000 3 000 4 000 solid flow measurement (Kg/s) liquid flow measurement (mᵌ/s) hydrometric station 15 01 06 (Annual scale) 0 0,2 0,4 0,6 Solid transport in millions Tons Months hydrometric station 01 22 01 0 0,04 0,08 0,12 0,16 Solid transport in millions Tons Months hydrometric station 02 13 01 16 GeoScience Engineering Vol. 70 (2024), No. 2 geoscience.cz pp. 01–16, ISSN 1802-5420 DOI 10.35180/gse-2024-0108 [4] BOUANANI, A. Hydrology, solid transport and modelling study of some sub-basins of the Tafna (NW - Algeria). State doctoral thesis. Tlemcen, Algeria: University of ABOU BEKR BELKAID, 2004. [5] BOUCHELKIA, H., BELARBI, F. and REMINI, B. Quantification of suspended sediment yield by statistical analysis: the case of Mouillah watershed. Water and Environment Journal. 2011, vol. 19, pp. 29–41. [6] DEMMAK, A. Contribution to the study of erosion and solid transport in Northern Algeria. Doctor of Engineering thesis. Paris: University of Pierre et Marie Currie, 1982. [7] DUBAND, D. Applied statistics in hydrology. National Higher School of Hydraulics, France, 1989. [8] LABORDE, J.P. Data analysis and automatic mapping in hydrology. Elements of Lorraine hydrology. Doctoral thesis. National Polytechnic Institute of Lorraine, National School of Applied Geology and Mining Exploration of Nancy, 1984. [9] LABORDE, J.P Automatic classification of variables using Dynamic Principal Components. Quantitative and applied spatial analysis. 1991, vol. 30, pp. 53-66. ISSN 0751-7297. [10] LEKFIR, A.; BENKACI, T. and DECHEMI, N. Suspended sediment assessment using fuzzy logic approach, application to Beni Amrane reservoir (Algeria). Journal of Water Science. 2005, vol. 19, no. 3, pp. 247-257. [11] MEDFOUNI, M.N.; KORICHI, K. and MAROUF, N. Artificial Neural Networks vs Long Short-Term Memory Prediction of Solid Flow in Tafna Basin (North-West Algeria). Ecological Engineering & Environmental Technology. 2024, vol. 25, no. 3, pp. 275–286. Available from: https://doi.org/10.12912/27197050/181208. [12] PEARSON, K. On lines and places of closest fit to a system of point in space. Philosophical magazine. 1901, vol. 2, no. 11, pp. 557-572. [13] REMINI, B. L’envasement des barrages. Bull. Réseau Erosion. 2000, vol. 20, pp. 165 -171. [14] REMINI, W. and REMINI, B. Sedimentation in dam in North Africa. Larhyss Journal. 2003, vol. 2, pp. 45– 54. ISSN 1112-3680. [15] SEMARI, K. and KORICHI, K. Water erosion mapping using several erosivity factors in the Macta Basin (North-West of Algeria). Acta Hydrologica Slovaca, 2023, vol. 24, no. 1, pp. 101-112. Available from: https://doi. org/10.31577/ahs-2023-0024.01.0012. [16] SENOUCI, F. Specific degradation and solid transport in semi-arid zones – Case of the Oued Ouahrane watershed. Master's thesis. Chlef, Algeria: University of Hassiba BEN BOUALI, 2013. [17] TERFOUS, A., MEGNOUNIF, A. and BOUANANI, A. Study of suspended solid transport in Oued Mouilah (North West Algeria). Journal of Water Sciences. 2001, vol. 14, no. 2, pp. 173–185. Available from: https://doi.org/10.7202/705416ar. [18] TOUAÏBIA, B. and ACHITE, M. Contribution to the mapping of specific erosion in the Wadi Mina basin in the semiarid zone of northern Algeria. Hydrological Sciences Journal. 2003, vol. 48, no. 2, pp. 235֪–242. Available from: https://doi:10.1623/hysj.48.2.235.44691. [19] TOUAIBIA, B.; AIDAOUI, A.; GOMER, D. and ACHITE, M. Temporal quantification and variability of sediment discharge in a semiarid area in northern Algeria. Hydrological Sciences Journal. 2001, vol. 46, no. 1, pp. 41–53. Available from: https://doi.org/10.1080/02626660109492799. [20] ZEGGANE, H. Study of the hydrological behavior of watercourses in Algeria, case of the Isser watershed, Algeria. Doctoral thesis in science. University of Kasdi-Merbah, Ourgla, Algeria, 2017. [21] ZEGGANE, H. and BOUTOUTAOU, D. Quantification and multivariate analysis of water erosion in the Mediterranean region. A case study of the Isser basin. northern Algeria. In: TECHNOLOGIES AND MATERIALS FOR RENEWABLE ENERGY, ENVIRONMENT AND SUSTAINABILITY: TMREES. 2016. Available from: https://doi.org/10.1063/1.4959411. [22] ZEYNEB, T.; NADIR, M. and BOUALEM, R. Modeling of suspended sediment concentrations by artificial neural network and adaptive neuro fuzzy interference system method – study of five largest basins in Eastern Algeria. Water Practice and Technology. 2022, vol. 17, no. 5, pp. 1058–1081. Available from: https://doi.org/10.2166/ wpt.2022.050.