Prioritization of Sub-Watersheds in a Large Semi-Arid Drainage Basin (Southern Jordan) Using Morphometric Analysis, GIS, and Multivariate Statistics

GIS-based morphometric analysis was employed to prioritize the W. Mujib-Wala watershed southern Jordan. Seventy six fourth-order sub-watersheds were prioritized using morphometric analysis of ten linear and shape parameters. Each sub-watershed is prioritized by designated ranks based on the calculated compound parameter (Cp). The total score for each sub-basin is assigned as per erosion threat. The 76 sub-basins were grouped into four categories of priority: very high (12 sub-basins, 15.8% of the total), high (32 sub-watersheds, 42.1% of the total), moderate (25 sub-watersheds, 32.9% of the total), and low (7 sub-watersheds, 9.2% of the total). Sub-watersheds categorized as very high and high are subjected to high erosion risk, thus creating an urgent need for applying soil and water conservation measures. The relative diversity in land use practices and land cover, including variation in slope and soil types, are considered in proposing suitable conservation structures for sub-watersheds connected to each priority class. The adaptation of soil conservation measures priority-wise will reduce the erosivity effect on soil loss; while increasing infiltration rates; and water availability in soil profile. Principal component analysis (PCA) reduces the basic parameters and erosion risk parameters to three components, explaining 88% of the variance. The relationships of these components to the basic and erosion risk parameters were evaluated, and then the degree of inter-correlation among the morphometric parameters was explored. The verification of priority classes obtained through morphometric analysis was tested using Discriminant Analysis (DA). The reHow to cite this paper: Farhan, Y., Anbar, A., Al-Shaikh, N., Almohammad, H., Alshawamreh, S. and Barghouthi, M. (2018) Prioritization of Sub-Watersheds in a Large Semi-Arid Drainage Basin (Southern Jordan) Using Morphometric Analysis, GIS, and Multivariate Statistics. Agricultural Sciences, 9, 437-468. https://doi.org/10.4236/as.2018.94031 Received: March 16, 2018 Accepted: April 27, 2018 Published: April 30, 2018 Copyright © 2018 by authors and Scientific Research Publishing Inc. This work is licensed under the Creative Commons Attribution International License (CC BY 4.0). http://creativecommons.org/licenses/by/4.0/ Open Access


Introduction
Rapid population growth(≈ 3% annually), and the persistent need for food production during the 1960s and '70s, caused continuous expansion of rainfed cultivation on marginal areas (average annual rainfall < 250 mm), forest and rangeland, which in turn increased the pressure on soil and water resources. The destruction of vegetation cover historically and land use abuse, coupled with the absence of efficient conservation measures, and compounded by improper farming practices accelerates soil erosion. The impact of land use/cover changes on soil erosion risk in northern Jordan was assessed [1] [2]. Cultivated land with poor conservation measures exhibit a higher rate of soil erosion and decline in soil fertility. The only conservation practice dominant is inefficient old stone terracing where rainfed "mixed farming" is practiced [3]. Further, soil erosion is becoming more serious on moderate and steep slopes which were transformed into cultivated land. Repetitive heavy rainstorms are the major significant triggering factor for extreme soil erosion, landslide activity, and floods. Severe storms with maximum daily intensity in the range of 2.1 -6.66 mm•h −1 are common in the highlands region [4] [5] [6]. Several case studies were carried out on watersheds prioritization in Jordan using morphometric analysis method, multivariate statistics, soil erosion susceptibility, and RUSLE soil loss modeling. For example, fourteen mini-watersheds of W. Shueib (Central Jordan) were prioritized for soil and water conservation [7] based on the method of morphometric analysis [8] [9] [10], and soil erosion susceptibility analysis [11] using GIS. Degradation of vegetation covers including forest, and the existence of W. Shueib structure along the eastern part of the catchment, caused serious soil erosion and landslides activity. Therefore, flooding and sediment discharge into the W. Shueib reservoir have increased during heavy rainstorms. Eleven mini-watersheds (78.6% of the total) are classified in the categories of moderate, high, and very high priority. Thus, they should be considered as potential areas for preferential conservation intervention, and must be prioritized immediately for soil and water conservation practices. Moreover, prioritization was performed for thirty one third-order mini-watersheds connected to W. Kerak (Southern Jordan) using morphometric analysis and RUSLE soil loss modeling W. Wala dam [17] are seriously threatened by the inflow of high sediment loads. Soil erosion loss for the catchments draining to the rift was found to be 1.328 million tons year −1 , which means, 0.14 cm of the top soil is eroded annually [18]. It is obvious that the recorded soil erosion rates in different parts of the highlands in Jordan are greater than the accepted soil loss tolerance limits for the Mediterranean environment which were estimated at from 2 to 12 ton•ha −1 •year −1 [19] [20].
Prioritization of sub-watersheds refers to the "ranking of different sub-watersheds as the order they have to be selected for appropriate soil conservation measures adaptation" [21]. Erosion risk parameters must be calculated to prioritize sub-watersheds for soil conservation. The linear parameters possess a direct relationship with erodibility. Therefore, the highest value of the linear variables was ranked 1, the second highest value ranked 2 and so on, and the least value was rated last in rank. Furthermore, the shape parameters possess an inverse relation with erodibility, thus, the lower their values, the greater the erodibility [22]. Consequently, the lowest value of shape parameter was rated as rank 1 and the second lowest as rank 2 and so on, and the highest value was rated last in rank [23]. The Compound Parameter (C p ) was computed by adding up all the DOI: 10.4236/as.2018.94031 440 Agricultural Sciences ranks of linear parameters, as well as shape parameters, and then, dividing by the number of all parameters. Following the rating of every single morphometric parameter, the ranking values for all linear and shape parameters referring to each sub-watershed are added up for each of the sub-watersheds to achieve a compound parameter (Cp) score based on the average value of these parameters. Further, the sub-watersheds having the lowest compound parameter score was designated the highest priority, the next higher value was assigned as second priority and so on [21]. Highest priority indicates the greater degree of soil erosion in that particular sub-basin, thus, it is considered a potential area for applying soil conservation measures [24]. Several methods have been developed and elaborated for watersheds prioritization. Eleven of the twenty five studies (44% of the total) employed morphometric analysis method in prioritization [8] [10] [13] [21] [25] [26] [27]. Whereas the other 14 studies (56% of the total) adopted the morphometric analysis method combined with one or two of the following methods, such as: sediment yield index method(SYI) [8] [22]; sediment production rate (SPR) method [28]; USLE/RUSLE models for estimating soil loss [7] [29] [30]; soil erosion susceptibility analysis [12]; runoff potential method [31]; land use/cover analysis [32] [33]; sediment yield index(SYI)and land use/cover [23]; weighted sum analysis technique [34]; fuzzy analytical hierarchy process [30] [35]; Snyder's method of unit hydrograph, and land use/cover [36].
In the current study, the authors employed the morphometric analysis method to: 1) Prioritize 76 fourth-order sub-watersheds for soil and water conservation based on the morphometric analysis method using GIS and RS.
2) Generate a spatial map illustrating the distribution of final priority classes representing the 76 sub-watersheds, 3) Explore the relationship of major components determined based on PCA to erosion risk parameters and the basic morphometric parameters. 4) Test the validity of the achieved priority classes by means of Discriminant Analysis, and to determine the nature of Discriminant functions in relation to the character of components, 5) Propose suitable soil and water conservation measures for the W. Mujib-Wala catchment.
Information on soil type, slope categories, and current land use/cover has been provided in order to help in suggesting appropriate soil conservation measures for sub-watersheds in each priority class. Prioritization in the present study demonstrates the role of powerful GIS, RS, and the morphometric analysis method in ranking different sub-watersheds in relation to the order in which they have to be taken for conservation measures [24]. Further, quantitative morphometric analysis of drainage networks is considered the key approach for understanding the hydro-morphological processes acting over drainage basins. Erosion risk parameters can be measured and calculated using DEM's and Arc GIS software. Morphometric analysis of linear and shape parameters are the

Derivation of Erosion Risk Parameters
The morphometric analysis method was employed to conduct prioritization for the 76 sub-watersheds. Five linear morphometric parameters (R b , D d , F s , T r , and L o ) and five shape morphometric parameters (R f , B s , R e , C c , and R e ) (see Table 1) were calculated as a first step to compute the compound parameter (Cp) in order to categorize the sub-watersheds into priority classes. The linear parameters comprise the bifurcation ratio, drainage density, stream frequency, texture ratio, and length of overland flow.
The bifurcation ratio (R b ) refers to the ratio of the number of streams in lower order (N u ) to the number of the streams of the next higher order (N u + 1), and it computed as: A is the basin area u L is the total stream length Stream frequency (F s ) is computed as the ratio of the total number of streams (N u ) of all orders in a watershed to the catchment area (A) [41]. It denotes the texture of the drainage network, and is expressed by the following equation: u N is the total number of streams A is the area of a drainage basin Texture ratio (T r ) is defined as the ratio of the total number of streams of the first order (N 1 ) to the perimeter of the drainage basin. Texture ratio is deter- u N = the total number of streams of all orders P = perimeter (km) Length of overland flow (L o ) is the length of water over the ground before it gets concentrated into definite stream channels [41], and is determined by the equation: L parameter is related inversely to the average slope of the channel and is equivalent to the length of sheet flow to a large extent [41].
The shape parameters comprise form factor, shape factor, elongation ratio, compactness coefficient, and circularity ratio. lowing: higher values of R f indicate a more circular shape of a catchment, whereas small R f values (<0.45) imply that the basin is elongated [41]. A catchment characterized by high R f has high peak flows of shorter duration.
Shape factor (B s ) refers to the ratio of the square of the basin length to the area of the basin, or Shape factor provides a notion towards the circular nature of the watershed.
The greater the circular character, the greater is the fast response of the catchment following an intense rainstorm [44].
Elongation ratio (R e ) is denoted by Variations in geology and climate cause prominent variation in R e values. Low values of R e indicate that the watershed is more elongated. When R e values approach 1.0, the shape of the watershed becomes circular [45].
Compactness coefficient (C c ) parameter is developed by Gravelius [46] as a shape measure of a watershed. C c is defined as the ratio of perimeter of a watershed to the circumference of the circle area, which is equal to the area of the watershed, or When the C c value is 1.0, it denotes that the catchment is a perfect circle. If the value is 1.28, the basin is more square-shaped, whereas the basin is considered a very elongated one, when the C c value > 3.0 [47].
Circularity ratio (R c ) parameter has been developed by Miller [48], and calculated according to: (A) is the basin area, and (P) is the perimeter. If R c is close 1.0, the shape of catchment is circular. Low, medium, and high values of R c indicate young, mature, and old stages of geomorphic development of the watershed.

Digital Elevation Model and Morphometric Parameters
Morphometric analysis for prioritization was conducted using the ASTER Glob-    [49]. Moreover, five basic parameters, five linear parameters, and five shape parameters were considered in prioritization of the 76 sub-watersheds based on morphometric analysis as illustrated earlier (Table 1)

Morphometric Analysis
Basic parameters comprise: A, P, N u , L b , and L u (

Prioritization of Sub-Watersheds Based on Morphometric Analysis
Morphometric The spatial distribution of priority classes is illustrated in Figure 7 and Table   3. Out of the 76 sub-watersheds, 12 sub-basins ( The third class of sub-basins is designated as moderate priority class ( Figure 7 and About half of these sub-basins are located in the rainfed farming area in the western part of the catchment (Figure 8(a)), while the rest of the sub-basins are located within the degraded rangeland and bare land. Scattered irrigated agriculture is predominant based on pumping wells (Figure 8(b)). It is also practiced downstream of the W. Wala floodplain and fluvial terraces using surface water from Wala reservoir (Figure 8(c)), the extracted water from springs, and groundwater wells. Similarly, irrigated agriculture is growing rapidly down the  dam of the W. Mujib on the floodplain, and fluvial terraces using surface water from the reservoir and pumping wells as well. The fourth priority class of sub-basins is assigned a low priority (Figure 7 and Table 3). It consists of 7

Proposed Soil and Water Conservation Measures
Morphometric analysis using erosion risk parameters was employed to prioritize 76 fourth-order sub-watersheds for soil and water conservation. The compound parameter values were calculated and the prioritization rating is given in Table  (Figure 7 and Table 3). Exploitation of land resources over the last 3000 years [56] has contributed to several serious environmental problems, i.e., [58]. Stone terraces should be placed in long rows along the contour at various intervals depending on the length and steepness of slope [58]. Additionally, bench terraces can be constructed on slopes even steeper than 25˚, and when stones are not sufficiently available to build contours stone terraces. Such conservation techniques have been adopted by the rainfed farmers since the Nabatean period, some 3000 years ago [56]. Structural conservation measures should be integrated with technology enhancing farming practices, i.e., rotation, strip cropping, contour strip intercropping cultivation [59], and contour plowing so contributed to the variance in the original data. It is also obvious that each component is more strongly correlated to some parameters that are considered more effective compared to others. Table 4 illustrates the Eigen-values, variance proportion, and cumulative proportion variance. Further, the results showed 88% of variances represented by 12 parameters which include five basic parameters, and one erosion risk parameter are highly loaded on PC 1 and PC 3 . Most of the basic and erosion risk parameters exhibit loading values > 0.9, whereas R c , F s and T r parameters display loading values < 0.9 (0.798, 0.898, and 0.703 respectively). The higher the loadings, the stronger the correlation. Three major components were resulted based on PCA analysis and accounts for 88% of the total variance explained by the basic, and erosion risk parameters and the C p parameter. The most effective variable in PC 1 to PC 3 are shown by bold font in Table 5.   third component corresponds to the form factor (R f ), and elongation ratio (R e ); therefore this component refers to the sub-watershed "Shape component."

Discriminant Analysis (DA)
It has been argued that prioritization based on the morphometric analysis method by utilizing the erosion risk morphometric parameters is time consuming in comparison with other approaches including the Principal Component Analysis approach, which allows for more effective parameters for prioritizing watersheds [24]. However, a prioritization for the Zarqa River has been carried out recently based on the morphometric analysis method and the PCA approach [13]. The output revealed that both methods did not produce similar results as stated earlier by Gajbhiye and Sharma [24] in a study on Shakkar River Catchment, Madhya Pradesh, India. Such disagreement in the results related to the two case studies is probably attributed to differences in physical conditions between Central India and Northern Jordan. The present study introduces a prioritization scheme for 76 sub-watersheds on the basis of the morphometric analysis method (the linear and shape parameters), and to test the achieved priority classes using Discriminant Analysis (DA) technique. The resultant priority classes displayed in Figure 7 and Table 3, are proved to be statistically valid. Consequently, the utilization of the morphometric analysis method is jus-tified as a successful method of prioritization. Statistical validation also implies that erosion risk morphometric parameters are efficient parameters in prioritization of watersheds for soil and water conservation measures. The intention of statistical testing of sub-watersheds pertaining to the four priority classes is to test the hypothesis that there are significant differences between the four priority classes achieved earlier, and if the hypothesis can be accepted to establish a system of coordinate axis which discriminates between the four priority classes identified (1: low priority to 4: very high priority). Statistical analysis was conducted on four data matrices representing the four priority groups (i.e., 7 × 11; 25 × 11:32 × 11; and 12 × 11) with the associated ranking values (connected to linear and shape parameters), including the Cp scores. The F test of Wilks Lambda obtained is F ratio 174.9, with the degree of freedom V1 = 3 and V2 = 72. Referring to the table of percentage points of the F-distribution, with V1 = 3 and V2 = 72, it is found that at 99.9 percent of confidence, the tabulated value is 5.78, which is significantly exceeded by the computed F ratio (174.9). Subsequently, their is a great significant difference between each of the priority groups (very high, high, moderate, and low), and the four priority classes are completely separate and distinct. Moreover, 98.0 percent of the difference between the four the four priority classes is attributed to discriminant function 1 (93.7 percent) and discriminant function 2 (4.3 percent). Further, it was observed that the discriminant function 1 is positively correlated with six erosion risk morphometric parameters (the linear and shape parameters). Correlation values range from 0.413 to 0.999, and the C p values are very strongly correlated with discriminant function 1 (0.970). By contrast, the correlation of discriminant function 2 with erosion risk parameters is relatively weak (0.17 -0.237). The scores of each sub-basin of the priority groups (shown in Table 3, and illustrated in Figure 7) on the discriminant function 1 and 2 were plotted in Figure 9. The plot displays completely disconnected priority clusters. The present results show that prioritization based on morphometric analysis is substantiated to be statistically valid, consistent and reliable, and of high capacity using GIS tools. The potential of the morphometric analysis approach as developed an elaborated earlier [8] [9] [22] is highly appreciated, and thus is, recommended for prioritization research.

Conclusions
Soil erosion by water has seriously threatened rainfed farming and rangeland over most of the terrain units of the W. Mujib-Wala watershed. High soil erosion rates have also increased sediment supply to the Mujib and Wala reservoirs during exceptionally heavy rainstorms, which are common in southern Jordan. The GIS-based morphometric analysis method and RS, were employed to prioritize 76 sub-watersheds, and the relationship of major components determined based on PCA to erosion risk morphometric parameters was explored. Then efficient conservation measures were suggested, especially for sub-watersheds where rainfed farming and grazing are practiced. Four priority classes were recognized: very high (12 sub-basins, 15.8% of the total), high (32 sub-basins, 42.1% of the total), moderate (25 sub-basins, 32.9% of the total, and low (7 sub-watersheds, 9.2% of the total). Moreover, 44 sub-watersheds (57.9% of the total) are ranked under very high and high priority, and are subject to high erosion risk, thus creating an urgent need for applying soil and water conservation measures so as to maintain rainfed farming and grazing sustainability. Based on C p values and ranking priority, sub-watershed 54 with a C p score of 22.8 receives the highest priority. The next in the priority list is sub-basin 47, having a value of 23.3 (Table 3). Likewise, sub-basins nos. 6 (Table 3), and with moderate priority as well. The W. Mujib-Wala watershed has been subjected to severe soil erosion over the last 3000 years, resulting in immense destruction of its vegetation cover. Supplementary information regarding current land use/cover, soil, slopes and topography were used as a guide in suggesting suitable soil and water conservation measures. The recommended measures were in accordance with priority ascribed in order to minimize negative impact of soil and land resources, rainfed farming, rangeland, and sedimentation in the Mujib and Wala dams. The expected advantages of expanding soil conservation measures, modernization of old soil conservation structures, enhancing farming practice, and rangeland management over sub-basins ranked as very high and high priority are manifested in the following: soil erosion loss control so as to protect soil from future erosion; minimizing sediment yield production to control sedimentation in the Mujib and Wala reservoirs; and reduced peak flow across the sub-watersheds and the entire W. Mujib-Wala catchment. Principal Component Analysis was used to reduce the original 16 basic and erosion risk morphometric parameters to three significant components which account 88% of the variance explained by the basic and erosion risk parameters. Out of ten parameters, six erosion risk variables were strongly correlated with PC 1 , and PC 2 where both explained 46.2% of the total variance. One erosion risk parameter (circularity ratio) was highly correlated with PC 1 , and most of the erosion risk parameters exhibit loading values >0.9. The limitation of the morphometric analysis method of prioritization stated earlier was tested statistically using Discriminant Analysis. The results showed that the four priority classes significantly differ from each other; thus, prioritization based on the morphometric analysis approach is consistent, reliable, accepted, and of high capacity using RS and GIS technology. The present results are aimed so as to assist decision-makers in identifying priority sub-basins which need immediate adoption of appropriate conservation measures, and land management practices.