Integrated modelling techniques to implication of demographic change and urban expansion dynamics on water demand management of developing city in Lake Hawassa Watershed, Ethiopia

Hawassa is a rapidly developing city in the Lake Hawassa watershed of Ethiopia that is a continuous change in the face of an urban environment. The urban development has been increasing the challenge to maintain urban services and surrounding environmental quality. These exert a new challenge to the growing gap between urban water demand and supply balance. Correlating urban growth and water demand to a rapidly growing population remains imperative to adaptive urban planning and decision-making. This study quantified urban development with demographic change and urban expansion dynamics. The population statistics and satellite imageries of historical years 1991–2021 and projections to the year 2051 were analysed using the exponential increase model and geospatial techniques. Multiple empirical modelling approaches were employed to link urban water demand with the explanatory variable. The study findings revealed the projected urban population reach more than one million and 79.2% of urbanization by 2051. With the current trend of 8.9% built-up growth rate, urban area will cover 73.6 km2 (45.9%) for the predicted period. The demographic variables and the sprawl of urban expansion jointly influence the water demand with statistically significant (f = 0.000, at α = 0.05) association. An increase in urban water use efficiency also reduces the water demand by increasing the availability of water supplies. Thus, the proposed model can be applied to reset the emerging relationship between the explanatory variables and water demand management. A detailed consideration of the spatially explicit effect on access to and optimization of the urban water supply system is vital to a local-specific solution. Integrating urban land planning with water demand management, therefore, has the potential to minimize the need to construct additional water supply infrastructure and cheer a sustainable urban environment.


Introduction
The urban environment is undergoing unprecedented urban land cover expansion, population growth, and socio-economic development (Xu et al 2019, Shao et al 2021. According to the United Nations UN (2018), twothirds of the world's population live in urban and semi-urban areas. As a result of population agglomeration, the African continent will reach 90 percent urbanization by 2050 (Andreasen et al 2017). Likewise, the 4.2 percent urban population growth rate of Ethiopia exceeds the average rate of 2.2 percent for developing nations (United Nations, Department of Economic and Social Affairs, Population Division UNDESA/PD 2018a). Urbanization and rapid population growth are continuing, and there is a noticeable difference between the resident of urban and rural areas (Terfa et al 2019, Bulti andAbebe 2020).
Hawassa represents a typical rapidly developing city among the cities of Ethiopia. It has been undergone a significant population growth and urban land expansion over the past three decades (Admasu 2015, Kinfu et al 2019. The city's diversified biophysical nature, socio-economic character and tourism attraction have accelerated the rapid urbanization and population growth. In the year 2050, the urban population of Hawassa City will reach over one million, as to the CSA (2007) census report. The city has also experienced uncontrolled urban sprawl that spills out into the extensive periphery areas of the town. Hence, the urban development dynamics have been prevalent and expected to continue in the coming decades, and the environment and several aspects of urban water system will encounter new challenge.
The developing cities are a substantial source of contaminants for freshwater ecosystems and one of the largest consumers of water (UN-Water 2017). According to Jensen and Wu's (2018) observation, as a city grows, the problem of water infrastructures and services that depend on them will rise and become more severe. The urban environment has become an important issue for researchers and planners worldwide in the context of the sustainable development Goals (SDGs), the action focused on cities (African Union Commission AUC 2015, Oldekop et al 2016, United Nations UN 2016. Water is often the basic constraint in human development and life, which has been linked inextricably to the sustainability of the environment. However, secondary cities in developing countries have little research on the most recent challenges to the urban water system. There has been limited reliable knowledge on the long-term spatiotemporal effect of urban expansion and demographic characteristics on water demand management. Mahendra and Seto (2019) argued that developing cities with higher rates of urban land expansion have more disparities in access to the water supply service coverage from their centre. Furthermost, Leeuwen et al (2012) emphasized that the vulnerability of the urban water system depends on exposure and severity of the situation. Lee and Choi (2018) and Dottori et al (2018) also noted that the impacts of the explanatory variables are local scale specific. Hurlimann and Wilson (2018) explained the need for spatially explicit planning approaches to adaptive strategies for sustainable water use. Several studies were focused on a wide range of factors affecting urban water demand related to population growth, socio-economic, and climate change (McDonald et al 2011, Barbier and Chaudhry 2014, Zhao et al 2018, Sanchez et al 2020, Diana et al 2021. The new challenge of urban growth dynamics on water demand remains less understood and discussed (Luo et al 2020, Heidari et al 2021. Most frequently, urban water issues have been considered after negative consequences of water stress and flooding events occurred or developed (Jacobsen et al 2012, Anandhi andKannan 2018).
Urban water-related issues are likely to intensify and become more pressing and characterize future sustainable development. Urban water demand management (UWDM) is a practical strategy that improves the equity, efficient, and sustainable use of water resources and minimizes the need for new water supply construction in urban areas. In this context, an empirical model is applicable for long-term water demand projection and optimization of the water supply system. As Haque et al (2021) mentioned, the accuracy of a mathematical modelling approach depends on the availability of historical data and the identification of significant explanatory variables that affect the system. Li et al (2019) and Dorning et al (2015) mentioned that a quantitative impact of urban development dynamics, as one of the significant explanatory variables, may offer vital information for a future challenge.
The dynamics aspect of urban development is quantified by demographic change and urban land cover dynamics via suitable population projection and geospatial modelling techniques (Mosammam et al 2016, Gonga et al 2018. Therefore, this study intended to develop an empirical model of the UWDM in the context of the sustainable urban environment in Hawassa City in the Hawassa Lake watershed. The study considered the 30-year historical (1991 to 2021) and the future (2022 to 2051) period of socio-demographic factors and urban expansion patterns with the advancement of urban water-saving strategies. The proposed model output can intimately link UWDM with long-term urban land planning and will contribute to the success of future sustainable city development.

Study area
Hawassa is the largest urban centre in the Lake Hawassa watershed, part of the Great Rift Valley basin of Ethiopia. It is located between 6°55′ to 7°3′ latitude North and 38°25′ to 38°34′ longitude East. Hawassa is a young city that was founded in the 1960s with a modern urban planning concept and named after Lake Hawassa (figure 1). It is the most political, tourist, commercial, and cultural hub among the regional cities of the country. The city has moderate warm temperatures that range from 10 to 30 degrees Celsius and a mean annual rainfall of 976 mm. The average elevation in the city centre and at the lake's surface is 1708 and 1680 meter above sea level, respectively (Hawassa City Administration HCA 2011). The topography of Hawassa is comparatively flat, and the drainage systems slope in the direction of Lake Hawassa.
According to Hawassa city current master plan (Hawassa City Administration HCA 2018), the administrative boundary covers 160.6 km 2 , which is divided into seven urban and one rural sub-cities, including 32 'kebele' or lowest administrative levels. A residential area is the primary land use (>40%), and it is within the central sub-cities and along the lakefront. Other land uses include commercial, industrial, urban agricultural, and green open spaces. Based on CSA (2007) census report, Hawassa's present population is estimated about 454,698 of which 68% are residents in urban sub-cities. With the annual population growth rates for urban and rural areas are 4.08% and 2.08%, respectively. Rapid urbanization and ongoing change in the face of the urban environment is experiencing in the city. An outward rapid urban growth has been primarily driven by natural population growth, rural-to-urban migration, and reclassification where nearby rural areas were incorporated into an urban area.
Hawassa is continuing rapid urban development and challenge essential aspects of the urban water management in the anticipated future. As the subsidiary city now, Hawassa needs to be addressed because it will become a primary city momentarily. Therefore, urban planning and decision-making on environmental issues facing today will have an impact on its capacity to handle emerging challenges to sustainable city development.

Data acquisition and processing
This study collected and processed various types of data, including historical demographic data, satellite imagery, Digital Elevation Model (DEM), and ancillary data. The period of 1991 to 2021 demographic data with its tribute was used for the numerical analysis and projection of the urban population and urbanization for the anticipated research period. Purposely considering the historical urban growth pattern over the last 30 years, four cloud-free satellite imagery with three different interval periods were acquired from the US Geological Survey (USGS official website). Worldwide Reference System (WRS) path 168 and 55row cover the whole study area, and the universal Transverse Mercator (UTM) zone of WGS_1984_UTM_Zone_37N.
The ground truth data were collected through GPS and Google Earth images for the accuracy assessment of image classification. The DEM was retrieved from the Shuttle Radar Topography Mission (SRTM) and used to delineate the study area into the sub-watersheds for a spatially explicit analysis of urban land expansion pattern. Ancillary data and information, including the city administrative boundary, the development map, the road map, and the Urban Water Supply System (UWSS) were obtained through field survey, various documents, and literature review. The detail of the attribute data acquisition and their sources are summarized in table 1.

Analysis of demographic change
The criteria used to characterize urban areas exclusive on administrative boundaries to distinguish between urban and rural areas. Hence, the spatial urban extent of Hawassa was considered primarily by defined administrative boundaries, the urban and rural population, satellite imagery, and access to water services and infrastructures. Based on the historical population growth of over 30 years, the population projection for the anticipated future period was estimated. The demographic data for the years 1990 to 2006 assumed the medium variant annual growth rate of the CSA 1984 and 1994. While the constant population growth rate of CSA (2007) was assumed for the period between 2007 and 2051. The demographic attribute data (figure 2) characterized the urban and rural population projection and urbanization for the study period.
The 2021 (study year) population data was taken as a base year for the projection period of 2022 to 2051. Empirical demographic projection models have been used for planning and implementation of population related socio-economic development interventions for anticipated period. The selection of appropriate projection model was based on availability of reliable data, trend of population growth, projection period, and minimizing uncertainties of future population estimate of the study area (Alemayhu and Yihunie 2014; Vanella et al 2020). The demographic projection is signified by exponential increase method (equation (1)) that has been used by CSA for the estimation of the urban and rural populations (CSA 2007). This method is mostly applicable for growing cities having vast scopes of expansion, which assumes the rate of population growth is proportional to present population. The UN Urban-Rural Growth Differential (URGD) Demographic Projection (DemProj) model (equation (2)) was applied to project the percentage of urbanization (United Nations, Department of Economic and Social Affairs, Population Division UNDESA/PD 2018b).
Where: U t and R t are the urban and rural populations respectively after t year; U 0 and R 0 be the same populations at the base period; u and r represent annual population growth rates of the urban and rural area, respectively; Exp is exponential number; U t /T t is urbanization level (%); d is the difference between u and r growth rates (URGD).
The population density is expressed as people per unit area (people/km 2 ) and assumed that can be the driver of a new urban expansion pattern. Using equation (3) and equation (4), the average population density of  CSA 1984CSA , 1994CSA , 2007CSA , 2013. Hawassa city and urban area were estimated, respectively.
Where: D t is total population density (people/area); P t is total population (urban+ rural); A t is total area (urban + non-urban), D u = Urban population density; A u is an urban area.

Analysis of spatially explicit urban expansion dynamics
The satellite imageries data were analysed using Arc GIS 10.7 and ERDAS 2015 image processing software, and thematic urban land cover maps were prepared. Owing to the average quality of the satellite imagery and the smaller size of the study area, the Maximum Likelihood Classification (MLC) algorithm was used as a supervised classification technique for land cover classification. The MLC is the most adopted technique, which has been pertinent for urban land cover change detection, manually corrected and validated (Rana and Sarkar 2021). Even though the image classification process has consumed much time, it has enabled us to refine the classification results to be closer to reality on the ground truth. The spatial resolution of the data used in environmental studies influences the level of detail of land cover classification and the accuracy of resulting maps (Momeni et al 2016). While this study focused on the small urban watershed, higher spatial resolution images enable the detection of urban land cover variation easily and provide high accuracy to inform decision-making. The Sentinel-2 up to 10 m resolution images are freely available and used for urban land cover change detection. The Landsat images represented the past, and the Sentinel-2 images depicted the present land cover of the study area.
Using two satellite images with different spatial resolutions may produce inconsistencies in the output map and influence image classification accuracy. Through appropriate resolution merge techniques, high resolution images can be combined with low resolution to produce a single image of improved quality (Lestiana andSukristiyanti 2018, Kaur et al 2021). However, the study is interested in the comparison of visible objects, such as urban land cover change and expansion, and urban land cover maps are generated using the two satellite datasets for the study period.
Based on the intended objective and familiarity with the study area, four major land cover classes were identified and defined as built-up/urban area, vegetation area, agricultural area, and waterbody. The area of each land cover class was calculated and the land cover change statistics over the study period were derived and recorded. As a result, the study area's spatiotemporal land cover change and urban expansion were determined. Further, the land cover classes were reclassified into built-up areas, and vegetation and agricultural land covers were considered as non-built-up area.
A spatially explicit urban expansion pattern was determined considering the city as a watershed and delineated into 29 sub-watersheds which were defined as Water Management Units (WMUs). During the delineation of WMUs, the DEM image was processed in Arc GIS 10.7 and Arc SWAT modelling techniques, as Ji and Qiuwen (2015) described. The urban land growth rate between each period was determined using equation (5). The future projection was estimated based on the historical built-up growth rate, and so the spatially explicit urban expansion pattern was analysed.
Where: P i is the urban growth rate, LC i and LC f is the land cover of the initial and final years of the period, respectively. Thus, the dynamic aspect of urban development was considered with respect to rapid population growth, percentage of urbanization, population density change, and sprawl urban expansion to the historical and future scenarios.

Analysis of urban water demand and supply system
The urban water demand and supply system was reviewed from relevant literature and official documents of HWSSSE. The analysis considered the historical water demand and supply, the urban expansion pattern, and average per capita water consumption. The per capita demand for the historical period up to 2015 was assumed the UN recommended an average 50 litre per person per day (l/p.d) which is comparable with the urban water supply design criteria (Ministry of Water Resources MoWR 2006). The value is given for the year 2006 and the annual growth rate factor of 1.5% is considered for the living standard and socio-economic activities.
The projection year (2016-2051) urban per capita demand was assumed Ministry of Water, Irrigation, and Electricity MoWIE (2015) standard for urban water supply service level. The population number and access to the water supply at the premises were considered in the National One WaSH program and SDGs plan. Thus, 100 l/c/day for a primary city (more than 1 million urban population), and 80 l/c/day for an intermediate city (100,000 to 1 million urban population) were assumed.
Urban water demand (UWD) may be categorized as domestic and non-domestic water users. The domestic water demand has been the major user of the water supply for various residence use. The domestic water demand of Hawassa accounts for 65%-70% of the total urban water demand. The domestic water demand, hereafter UWD is estimated by multiplying the urban population and the per capita water consumption.

Multiple linear regression model
Multiple linear regression develops a model by establishing a linear relationship between two or more independent variables with the dependent variable . The independent (explanatory) variables were urban population, urbanization, urban land expansion and population density change. A stepwise empirical analysis was performed in Excel to explore the correlation between each explanatory variable and their potential impact on UWDM. Using equation (6), the empirical model was developed.
Where: Y t is representing Urban water demand (m 3 /year); X 1 , X 2 , X 3, K X n are independent variables; β 1 , β 2 , β 3 , β 4 K β n are coefficient of each independent variable; ε t is the error due to variability in the observed responses.

Demographic change analysis
The growing urban population is the function of natural population increase, rural-to-urban migration, and industrial development that contributes to the pursuit of socio-economic and job opportunities. As summarized in figure 3, the base year (2021) urban population of Hawassa is 308,436, so considered as intermediate city (100,000 to one million inhabitants). Over the next 30 years, the projected urbanization of Hawassa will reach 79.2%. The urban population predicted to be 632,497 and 1,300,013 in the years 2036 and 2051, respectively. Hawassa will become a primary city (more than one million inhabitants) in the middle of the century. The future urban population will be higher than the projected counts, as an increase in urbanization and internal migration are expected. Hence, error bars with 5% values across urban population considered to minimize uncertainty.
The rapid urban population growth is the key driver of the increase in a built-up area, that challenges the growing need for new urban infrastructures. Urbanization is, therefore, population-driven land use change and expressed as percentage of urban population. It is accompanied by the need for basic urban infrastructural development, which resulting in sprawl urban expansion. Furthermore, the population density as a function of population per unit area affects urban expansion patterns, the design and access to the water supply system. Figure 4 summarizes the population density of the city and urban area for the anticipated study period. The historical to the present scenario has relatively low urban population density with dispersed settlement patterns, and built-up areas are expanding in all directions. A new urban development primarily occurs on agricultural land and some expansion into the vegetation areas on the periphery of the town. The future scenario shows higher urban population density than the historical trend. This implies there is rapid urban population growth than urban expansion and relatively dense settlement through clustering around the urban center.

Spatially explicit urban expansion dynamics analysis
Succeeding land cover classification, the accuracy of classification was assessed with 50 random ground truth sample points for each land cover class and Google Earth imagery for each year. The accuracy assessment techniques were used alike to early studies (Rwanga andNdambuki 2017, Dessu et al 2020). The result in table 2 shows, the overall accuracy with kappa coefficient for the year 1991, 2006, 2016 and 2021. The results of overall accuracy values are greater than 85% and kappa coefficients are between 0.80 to 0.85, that imply an agreement and accuracy requirement for the land cover change analysis, as Terfa et al (2019) discussed.
The urban land cover of Hawassa changed from 11.6 km 2 (7.2%) in 1991 to 42.5 km 2 (26.5%) in 2021 with 8.9% average built-up growth rate. As table 3 indicates, the urban land growth rate varies with the urban development process of the city over three periods with different intervals. Considering the current trend builtup growth rate, future urban land cover of the city for the years 2036 and 2051 predictable to be 58.1 km 2 (36.2%) and 73.6 km 2 (45.9%), respectively.
The result is also consistent with similar studies of Wondrade et al (2014) and Degefu et al (2021). The outward urban expansion mainly in the North (N), Northeast (NE), South (S), and Southeast (SE) directions. Figure 5 illustrates, the spatially explicit urban expansion extent over the last 30 years concerning urban water demand, access to water supply service and coverage in the urban setting. Specifically, spatially explicit urban  , and the shoreline of Lake Hawassa, which are the urban centre and populated area. Further, the rapid urban expansion in the 2006, 2016 and 2021 continued to the direction of SW and widespread to the periphery of the urban area. The sprawl urban expansion is extended into agricultural, vegetation, and environmentally sensitive areas with inadequate water supply service coverage. The urban land cover change analysis result also revealed that between 1991 to 2021 built-up area increased by 19.2% (30.9 km 2 ), agricultural area reduced from 79.7% (128 km 2 ) to 49.1% (78.9 km 2 ), and urban expansion pattern also extended to Hawassa Lake front.
The extent and direction of urban expansion of Hawassa has been mainly associated with its physical setting, political, administrative boundary condition, demographic change, and development plan of the city. Excluding a hill in the south direction of the city, the topography of Hawassa is relatively flat (figure 1), and not a constraint for the expansion. The city is suitable for future outward urban expansion in all directions, apart from Lake Hawassa. The study of Degife et al (2019) also revealed comparable results on the expansion pattern of Hawassa.
The urban land cover dynamics has a direct impact on hydrological characteristics of watershed. The urban water supply distribution network of the city mainly covers the urban centre and the year 2006 land cover area. The access to and a new design requirement of the water supply distribution system is mainly dependent on the spatial urban expansion pattern. Hence, the spatially explicit urban land cover extent map highlights the potential of spatial urban planning strategies to local-specific UWDM and maintain sustainable urban environment.
3.3. Urban water demand and supply system analysis Adequate and safe water supply is the basic urban service that extensively influences the socio-economic development and well-being of the community. The water supply sources of the town include 79% groundwater (deep well) and 21% surface water (river treatment and springs) which are connected to the urban water supply system. The coverage of urban water supply service in the base year has reached 68%. The existing quantity of water supplied to the town is 14,210.12 million m 3 , which is a tenfold time increase as compared with 1,223.6 million m 3 for the year 1991.That comprises without considering water losses due to the operation and efficiency of the distribution system. The water supply system was designed in 1985 and has progressively improved in size, length, and density of the distribution network. In terms of access to the urban water supply system, the spatial extent of the water supply pipelines covered the central area, but the periphery areas of the town could not adequately reach. The spatial extension of the pipeline over the settlement area is a pre-condition to supply water with adequate quantity, quality, and pressure to water users. The efficiency of the water supply system can be determined primarily by the density of pipelines within the distribution area. Figure 6 illustrates, the domestic water demand from 1991 to 2021 gradually increased from 0.845 to 9.01 million m 3 . The projected domestic water demand is estimated 18.47 and 47.45 million m 3 for the future scenario of years 2036 and 2051, respectively. The growing population and socio-economic change are directly related to an increase in domestic water demand and per capita water consumption of the city. The trend analysis may only give a qualitative description of the impact of demographic factors on urban water demand and supply challenges. Understanding emerging influencing factors on UWDM becomes a basis for developing a reliable water demand projection to balance the water demand and supply in the long run.
In this regard, analytical modelling approaches have become necessary for understanding the long-term sustainability of the urban environment. A multiple empirical model is used to renovate the UWDM to the dynamic circumstance of socio-demographic factors and urban expansion. Subsequently, two plausible scenarios of the historical and anticipated future period are determined to integrate UWDM and urban planning to adaptive decision making for the changing environment.

Correlation analysis of urban land expansion and demographic change
The growing urban population and urbanization are significant driving variables, that resulted in a sprawl spatial urban expansion pattern. Thus, it is determined through an increase in a built-up area. The correlation analysis result reveal that the urban expansion and urban population growth are positively correlated with a coefficient (β) value of 0.052, explains 89.02% of the variation of response variable and significant at 95% conference level (α < 0.05).
Urbanization level is expressed as the percentage, reflecting the portion of the urban population of the city. The urban land expansion is positively correlated to urbanization (β = 1.83, R 2 = 0.9187) and statistically significant (α < 0.05). A sprawl of urban expansion expands at the extensive margin. Bigelow et al (2017) also explained the environmental consequences of urbanization. The urban population density change also explains 81.7% of urban expansion variation. The correlation between urban population density and urban land expansion is positive with β = 4.9 and statistically significant.
After checking the assumption satisfaction of explanatory variables, the analysis was on the combined impact of the three independent variables on urban land expansion. The Analysis of variance (ANOVA) result indicates the overall F statistic value of 398.035 and the significant F value of 0.000. The coefficient of determination value (R 2 ) implies the combined influence of the independent variables can explain 99.23% of the variation in sprawl urban expansion extent. Hence, the equation: Urban expansion = 0.08 (urban population) + 1.18 (urbanization) − 5.63 (population density) −19.72 can be applied to estimate the expected urban expansion based on the combined influence of the explanatory variables. The coefficient values represent the mean change in the response variable to change in a unit of the independent variable.
The model output shows that urbanization and urban population growth are the most impactful explanatory variables to sprawl urban expansion. The urban population density change has the opposite influence. The effect of urban population density depends on whether a change is occurring under the businessas-usual scenario on behalf of the extensive periphery areas converted to built-up areas or an increase in the development tightly clustered around the town.

Empirical model of urban water demand management
The effect of demographic variables and urban expansion along with the progress of water use efficiency were considered to model UWDM. The corelation analysis result in figure 7(a) reveals the influence of urban population growth on UWD has a positive value of β. The growing urban population explains 99.2% variation of the response variable with good degree fitting and statistically significant (p < 0.05). Figure 7(b) shows the regression relationship of urbanization and UWD with positive value β and R 2 = 0. 653. Similarly, the correlation result in figure 7(c) shows a positive value of β and the urban expansion explains 81.54% of the variation, but no obvious statistically significant (p > 0.05) correlation. Meanwhile, the influence of sprawl urban expansion is considerable on the access to and the requirement of additional water supply distribution networks design. As shown in figure 7(d), urban population density and UWD are positively related, but their correlation is not statistically significant.
Furthermore, the significant relationship between the response variable and the combined influence of the explanatory variables was analysed. The null (Ho) and alternative (H1) hypotheses for testing are defined. Ho: the UWD has no statistically significant relationship with the combined influence of the explanatory variables. H1: the UWD has a significant correlation with the explanatory variables combined in the study area.
The regression model test output shows a correlation between UWD and the independent variables, and high predictability with a predicted R 2 value of 99.43%. The result also reveals the overall F statistic value of 349.077 and the significant F value of 0.000. There is a statistically significant relation and rejecting the null hypothesis. Hence, this implies that the four explanatory variables combined have a statistical association with the response variable and jointly impact the UWD of the city.
The empirical model proposed (equation (7)) can be used to estimate UWD. With coefficients (β) value, the high impact of urbanization followed by urban population growth as compared to urban land expansion and urban population density change.

( )
Where: Y t is Urban Water Demand (m 3 /year); X 1 is urban population growth (number); X 2 is urbanization (%); X 3 is urban expansion (km 2 ); X 4 is urban population density (People /km 2 ). The model result suggests the significant influence of urbanization and growing population on UWD. The increase in population density has shown the opposite effect. The effect of the population density on UWD depends on whether the density of urban growth is occurring in the intensive area or the periphery area of the town. The sprawl urban expansion has converted agricultural and vegetation areas to built-up. The net impact of urban expansion pattern on UWDM will depend on the quantities of water required by urban water users. The urban land planning approach also influences the level and density of future development. Thereby influencing the water demand and supply, the spatial urban expansion pattern and quality of surrounding environment.
The challenge of spatially explicit sprawl expansion and relatively high-density development on the water supply system may differ. The UWDM on the sprawl urban expansion pattern may imply low access to water supply services and increase in the design cost of new water supply distribution networks. The high-density development mean optimization of the water supply system by cost minimization with reduced pipeline length is required. It may not imply relatively increased access to water supply services. The intervention measures and strategic decision-making will improve the access to and coverage of the distribution networks to the standard. It is also important to note that an increase in water use efficiency likely improves future domestic water use by decreasing water loss due to efficiency gain for the plan period. The water supply system efficiency assumed was 65% and could be substantially improved. Based on the established reduction of water loss plan of Ministry of Water, Irrigation, and Electricity MoWIE (2015), a plausible range of 15%-25% water loss factor is acceptable for the future anticipated period. The urban water demand reduction through the progress of water saving technologies as planned by the city will be incorporated to the net changes in water demand. Considering urban water forms (i.e., domestic wastewater reuse and water harvesting) as potential water sources, which fit to the purpose and conservation strategies will improve water availability and quality of urban environment.
The significant influence of climate change is not in the scope of this study. The proposed empirical model should be modified to incorporate the climatic elasticity of domestic water demand due to change in climate variables. The study mainly focuses on the demand side of the urban water management equation. A complementary study on the design of urban water distribution networks will also provide comprehensive information for adaptive urban planning and decision-making to sustainable urban development.

Conclusions
This study analysed impact of the socio-demographic change and urban expansion dynamics along with the advancement of urban water use efficiency on UWDM. The empirical modelling approach is enabled to correlate the complicate relationship between UWDM and the attribute of urban development dynamics of Hawassa as developing city. The developed empirical model also helps to understand and better account for UWD projection adjustment with the emerging explanatory variables.
However, the proposed model result may be subject to some uncertainty and consideration is required for future improvement of the model. The significant sources of uncertainties, including the availability of reliable population data and assumptions. The study should consider the impact of climate change for the improvement of model accuracy. The domestic water demand for in-door and out-door uses are different, therefore, effect of the explanatory variables on each uses should be evaluated for better UWDM strategies. Even with these uncertainties, the quantitively analysis of UWDM and new impact of urban development dynamics are closely linked and integrate UWDM with urban planning for adaptive decision-making.
The spatial dimensions of urban water demand related to the current and future water supply distribution networks coverage should be considered to develop area specific strategies for the sustainability of the urban environment. The urban land expansion dynamics is also a significant impact driver for other aspects of the urban water cycle, such as an increase in impervious surface areas, and challenge urban stormwater management. Furthermore, future studies are required to address this issue in an integrated urban water resources management approach. The urban density-specific water requirements should also be included for consideration in future studies. The study will recommend that recognizing the possible impact of future urban development dynamics on water demand management will reduce the vulnerability of urban water system and helps to plan innovative strategies to adapt or mitigate those impacts.