Data screening

The precipitation data has been checked for its homogeneity by using the homogeneity test [9]. The non-dimensional parameter P_i is given by:

P i = P i , a v P a v ∗ 100 [9]

where, P_i, represents precipitation for the month i and it is non-dimensional value, P_i,av is over year’s averaged monthly precipitation of the station i. P_av, is over year’s averaged yearly precipitation of the station.

The consistency of the precipitation data is checked using double mass curve method [10].

P c x = P x ∗ M c M a [10]

where: P_cx is corrected precipitation at any time, P_x is originally recorded precipitation at any time, M_c is corrected slope of the double mass curve and M_a is original slope of the double mass curve.

1) Missing data completion

A number of methods have been proposed to estimate missing rainfall data [9]. Normal ratio method is used for filling missing precipitation data.

P m = 1 n { ∑ k = 0 n ( N m N i ) P i } [9]

where, N_m: Average annual rainfall for the gauge in which data are missing, N_i: Average annual rainfall at gauge i. and arithmetic mean method was used for filling the stream flow data.

2) Areal rainfall estimation

The selected rainfall stations need to be uniformly distributed in order to represent the entire sub-basin, as there might be different geological and hydrological features in different parts of the sub-basin. Thiessen polygon method was used for checking the uniformity of the rainfall station distribution over the entire sub-basin [9]. As the result shows the stations are uniformly distributed over the sub-basin. The location of the rainfall stations in the sub-basin is shown in Figure 2.

The Water Evaluation and Planning software is used in this study. WEAP model basically calculates a mass balance flow sequentially down a river system, making allowance for abstractions and inflows [11]. The elements in the model help to

characterize the water demand-supply system and their spatial relationships [11].

The function of the model is not to describe the hydrological process of the sub-basin accurately, but to be able to simulate the surface water resources of the study area within the available data and using a small number of parameters [11].

Five different sets of “views” are structured onto the working area to form the WEAP model complete: Schematic, Data, Results, Scenario Explorer and Notes [10]. These views are listed on the left side of the screen as graphical icons. Schematic is GIS tool for configuring the system through dragging and dropping to create and position, helps to add other standards in the forms of vector or raster GIS image as background layers. It is an associate access to data and results for any given node. Data view helps to build the model. It produces variables and relationships, enter assumptions and projections using mathematical expressions, and immediately links to excel. Notes help to document data and assumptions. Results view helps to display detailed model outputs in the form of charts, tables and maps. Scenario Explorer is a higher level view of data and results. The slider moves to change the value of connected scenario data variable and WEAP recalculates so that the impact on user-selected key results are displayed [11].

The main menu located at the top provides access to the basic functions of the programs. The current area name, current view, licensing information and other status information are presented in the status bar at the bottom of the screen. The layout of the remaining screen depends on the selected view [11].

From the five methods presented in WEAP model and listed below the rainfall-runoff method is used for this research.

1) Irrigation Demands only method;

2) The Rainfall-Runoff method;

3) The soil moisture method;

4) The MABIA method; and

5) The plant growth method.

Irrigation Demands approach calculates the potential evapotranspiration in the catchment by using crop coefficient, then irrigation demand is identified to cover the parts of evapotranspiration that are not covered by the rainfall [12]. In this method runoff cannot be simulated and infiltration is not calculated. The Rainfall-Runoff method also uses crop coefficient in evapotranspiration calculation for irrigated and “rain-fed” crops [12]. The remaining portion of rainfall which is not consumed by the evapotranspiration is simulated as surface runoff in the river. Among the five methods the soil moisture method is the most complex one. It represents the soil with two layers and also represents the potential for snow accumulation. It simulates the evapotranspiration in the upper layer by considering rainfall and irrigation on agricultural and non-agricultural land, runoff and shallow interflow in the soil moisture [13]. These methods permit the characterization of land cover and/or soil type impacts to the hydrologic process. In the lower soil layer base flow routing and soil moisture changes are simulated. Considerably this method requires comprehensive data to simulate each process in the system, those kinds of data are not available easily [13].

Steps followed by WEAP for surface water potential assessment and demand evaluation [13].

1) Defining the study area, time frame, system components and configuration.

2) Creating the current account, it is water resources potential of the study area.

3) Creating of the scenarios based on future assumptions and expected increases in the various indicators. This forms the core or the heart of the WEAP model since this allows for possible water resources management processes to be adopted from the results generated from running the model.

4) Evaluating the scenarios about the availability of the water resources for the study area.

Results generated from the creation of scenarios can help the water resources planner in decision making, which is the core of this study.

The term calibration refers to the adjustment of a model parameter such as roughness, hydraulic structure coefficients, etc., in order to make the model produces the observed prototype data at an acceptable range of accuracy [11]. The objective function and subjective criteria are the two general approaches to assess the best fitness of data. The objective function is used in this study. The aim of WEAP model calibration is fitting the simulated data to the observed flow data collected from gauging stations, which involves the quantitative evaluation of hydrologic response of the Dabus sub-basin and its tributaries [11].

The model calibration is attained by using the EF (Nash-Sutcliffe Efficiency) and R² (coefficient of determination) since the Software (WEAP), has no automatic calibration routine; due to this reason the changes implemented were tested manually by comparing the simulated and observed time series flow [11].

Nash and Sutcliffe efficiency criteria

The objective assessment relies on creating measures of errors using statistical parameters for evaluations of the model. These error methods include mean error (ME), mean square error (MSE) and the model coefficient of efficiency (EF) [14]. The MSE measures the mean square of the errors, which is the difference between the simulated and observed flow values. The EF is used in assessing the predictive power of hydrological models [15]. It is the non-dimensional parameter and scaled version of MSE ranging between 0 and 1, most of the time it gives clearer evaluation of the model results and performances. The efficiency value 1 indicates a perfect match between the observed and simulated discharge whereas, the efficiency value 0 indicates that the model prediction is accurate as the mean of the observed data [16].

E Q = Q m − Q o [10]

ME = ∑ i = 1 n ( Q m ( i ) − Q o ( i ) n ) = ∑ i = 1 n ( E Q ( i ) n ) [10]

MSE = ∑ i = 1 n ( Q m ( i ) − Q o ( i ) n 2 ) = ∑ i = 1 n ( E Q ( i ) n 2 ) [10]

EF = [ 1 − MSE S 2 Q o ] [10]

where: E_Q = Difference between simulated and observed flow, Q_o = Observed flow, Q_m = Simulated flow, ME = Mean Error, MSE = Mean Squared Error, and EF = Model Efficiency Coefficient, and S = Variance.

Coefficient of determination, R²

The coefficient of determination is R² is defined as the squared value of coefficient of correlation of the simulated and observed streamflow. Coefficient of determination is also used for this research for calibration. It is estimated using the following formulae [11].

R 2 = [ ∑ i = 1 n ( Q o − Q ¯ s ) ( Q o − Q ¯ o ) ] 2 [ ∑ i = 1 n ( Q s − Q ¯ s ) ] 2 [ ∑ i = 1 n ( Q o − Q ¯ o ) ] 2 [11]

where Q_o = observed flow,

Q_s = Simulated flow,

Q ¯ o = Mean observed flow,

Q ¯ s = Mean simulated flow.

WEAP examines water demand in an area based on disintegrated end-use approach [17]. By applying WEAP21 one can use economic, demographic and water-use information to develop scenarios that analyze how total and disaggregated consumption of water grow over time in all sectors of the economy [17]. Calculation of demand for domestic, livestock and agriculture entities are based on a disintegrated accounting for several measures of social and economic activity such as population served, livestock population size and area of land irrigated [17]. In this study 10 percent of the domestic water demand is assigned for industrial demand, by taking into consideration the future growth of industry in the area. Regardless at present time there is small industrial activity (only one site of gold mining) in the sub-basin. Environmental flow requirements (EFR) are also considered in this study to sustain the ecosystem of the area by taking 10 percent of the total annual flow. Standard Water Use Method was selected, in which the modeler determines each water user type (activity) for each end-user and inputs the correct annual water use rate for each activity [17].

Current account year (Base year: 2020)

The current account year is the base from which all the other scenarios developed from.

Types of demand considered

1) Domestic demand

2) Industrial demand

3) Irrigation demand

4) Livestock demand

5) Environmental flow requirement (EFR) demand

The total water demand for the year 2020 is 867 MCM and it is percentage to each demand sector is shown in Figure 4.

A scenario is a probable description of how the future may emerge based on articulate and internally consistent set of assumptions about key interactions and driving forces in the hydrologic processes [18].

In order to contain future water demand sophistication four different scenarios have been developed under three different time horizons which are listed as follows:

1) Scenario 1: Reference scenario (2021-2050)

Under this scenario the demands are met with 100 percent coverage and the largest percentage of demand comes from the environmental flow requirement and lowest being the domestic water demand (urban and rural) as shown in Figure 5. This implies for this scenario if situation indicated prevailed the surface water from the sub-basin is sufficient, so no additional source of water is required.

2) Scenario 2: Domestic water demand change due to population growth rate

Under this scenario as it is indicated in Figure 6 the domestic water demand is analyzed for the three different growth rate probability: high, low, and medium and all the demands met with 100 percent coverage. This shows if one of the three conditions prevails in the future, the surface water from the sub-basin is sufficient and no need for additional source of water for supply.

3) Scenario 3: Irrigation demand projection

Under this scenario the irrigation area is increased and as the result indicates there is unmet demand under irrigation projection scenario for the months of Apr and May. The unmet demand is presented in Figure 7.

4) Scenario 4: Increased water demand scenario

Under this forth scenario the domestic demand i.e. both for rural and urban population, the demand has been increased regardless of the surface water potential. However, the water resources potential in the sub-basin met the demand with 100 percent coverage, so if the mentioned scenario prevailed there is no need for considering any additional sources of water Figure 8 shows the coverage of the demand under this scenario.