GIS-based analytical modeling on evaluating impacts of urbanization in Amman water resources, Jordan

Amman governorate is the largest governorate in terms of population and urbanization in Jordan that is the third most water-scarce country worldwide. It has also limited water resources that were rapidly decreasing as results of groundwater over-pumping and climate changes that generate a serious water crisis. However, the population and urbanization focused on the Northwest of the governorate. The surface water and groundwater resources are available in the Northwest area as well. The overlaying between urbanization and population on one hand and water resources on the other hand resulted in different environmental, hydrological, and hydrogeological problems. Our research investigated these problems using an integrated approach of remote sensing and geographic information systems. Furthermore, our research suggested a spatial plan that would solve the conflict of urbanization's impact on water resources in Amman. Accordingly, the catchment areas that span on the study area and their drainage network were defined.


Introduction
Amman governorate has an area of about 7579 km 2 and about 4,007,000 inhabitants (Perdew 2014;Department of Statistics 2014). The population in the Amman governorate has been increasing dramatically in the last 40 years. The population growth rate is at 5.8%. In addition to this growth, there is a continuous flux of migrations and refugees from neighboring countries like Palestine, Iraq, and Syria which accelerates the urban expansion (Potter et al. 2009;Odeh et al. 2017).
However, in 1921, Amman was established as the capital city of Trans-Jordan and had about 5,000 inhabitants who are mostly from Circassia, Syria and the surrounded cities such as Irbid and Al-Salt. In 1948, the number of inhabitants became about 90,000 as a result first Palestine refugee migration. During 1961During -1979, the population increased from 215,000 inhabitants to 777,800 by the population growth and second Palestinian refugee migration in 1967. The first and the second gulf war in 1991 and 2003 forced more refugee waves to migrate to Jordan in general and to Amman specifically, and the population becomes 851,000, and 1,017,000 inhabitants, respectively. In 2014, the population of Amman becomes 1,698,000 after the migration of Syrians due to the civil war.
Urban area expanding because of inhabitants increasing is a common process in the capital cities such as Amman (Schneider et al 1973, Butler andDavies 2000). However, previous studies show that urban expansion in Amman due to population growth has increased the urbanized area from about 147 km 2 in the year 1987 to 237 km 2 in 2014 that means the urbanized area increased by almost 61% over the last 3 decades. During these decades, urban expansion was at a high-rate stage during 1987-1997 and at a steady-rate stage during 1997-2007and 2007-2014(Al-Bilbisi 2019Al-Bakri et al. 2013).
Geographically, the Amman governorate is in the northwest of Jordan (Fig. 1). It is bounded from the east by Saudi Arabia and from the west by the Palestinian National Authority. Topographically, it has a significant variation in altitude, which ranges between (− 143 m below sea level (absl.) − 1100 m absl.). It has different landforms (Fig. 2) which are influenced by the underlining geological features (Odeh et al. 2015). However, such a change in altitude and large spatial expansion combined led Amman to be influenced by the three common climatic zones in Jordan [The Jordan Valley, The Mountain Heights Plateau, and The Desert or Badai region (Potter et al. 2009;Perdew 2014)]. In general, northwest of Jordan is closer to the Mediterranean Sea where most of the atmospheric troughs come and hence has more rainfall events, while the highlands are commonly the areas that receive the greatest amount of rainfall (Salameh and Bannayan 1993). However, the climate in Amman governorate is subtropical arid, with cold winter (November-April) in the northwest due to the high altitude l that reaches up to 1100 m, and sunny in Fig. 1 The location of the study area (Amman governorate). It is located in the north west of Jordan and has an area of about 7579 km 2 the summer (June-September) that is slightly hot and is controlled by the adjustment desert area (Salameh and Bannayan 1993;Potter et al. 2009;Perdew 2014). Due to the topographic variation, climate is changing spatially. The western highlands of the governorate are close to the Mediterranean Sea and thus affected by that climatic zone, while the eastern part is close to the Arabian Desert and is affected by this climatic zone too (Berndtsson and Larson 1987;Bouraoui et al. 1999).
In our study, we selected rainfall patterns as a primary climatic indicator (Fig. 3) to simplify understanding the climatic patterns in the Amman Governorate. The rainfall patterns are integral to the quantity and quality of water resources (Salameh and Bannayan 1993;Odeh et al. 2017;BGR 2017). The greatest rainfall pattern values (about 500 mm) are concentrated in the northwest, and therefore, most of its water resources are allocated in that area, while the lowest values about 0 mm are in the southeast of the study area (Potter et al. 2009).
Amman gets its water from both surface water and groundwater for agriculture, domestic, and industrial purposes. Both surface water and groundwater in Amman are solely fed by rainfall. The bank full discharge of the river and the recharge of groundwater is accounted for primarily by the receiving rain (Berndtsson and Larson 1987;Bouraoui et al. 1999;Saraf et al. 2004). However, the water resources of Amman are mostly located in the northwest because of rainy climatic zones that prevailed in that area due to the geographic location (nearest area to the Mediterranean Sea) and landforms (mountainous area) where urbanization and population expansion is taking place .
From a geological perspective, Amman is situated in a faulted zone because of the adjustment to the regional Dead Sea transform fault. The faults are mostly strike-slip faults that generate conduits between the aquifers (Odeh et al. 2019). However, the rock units that are cropped out in the study area belong to three geologic periods which are from the oldest to youngest: pre-Cambrian, Cretaceous, and Tertiary (Bender 1974). These rock units could be further classified into five hydrogeological units as follows (BGR 2017) (Fig. 4): 1) Lower aquifer that consists of sandstone. 2) Lower aquiclude that it consists of marl.
3) Upper aquifer that consists of Limestone. 4) Upper aquiclude that it consists of marl. 5) Shallow aquifer that is consisted of chalk and limestone.
However, the groundwater recharge of these aquifers is mostly allocated within the population and urbanizationexpanding zone .
The objective of this research is to investigate the effect of the urbanization expansion on the surface and groundwater resources of Amman. Such a study would enhance sustainable water resource management by generating a spatial plan that recommends where the urban expansion must move to  and where water resources' enhancement projects such as water harvesting needs could be carried out.
Jordan has no study that evaluated the impact of urbanization on the water resources of the Amman governorate which our study is doing (Al Qudah et al. 2015; Al-Bilbisi 2019). However, worldwide, there are tens of water studies assessing the effects of urbanization on water resources, but there is no study evaluating the effects of urbanization on the water cycle variables and hence on the water resources as our study do (Balha et al. 2020;Huang et al. 2020). Furthermore, our research is the first that use this evaluation for sustainable land management by generating a spatial plan for a water harvesting project (Yu et al. 2018; Mohammady 2021).

Methodology
The integrated approach of remote sensing (RS) and Geographic Information Systems (GIS) is now widely used to study environmental hydrology and hydrogeology and water resource problems (Chou 1997;Saraf et al. 2004;Odeh et al. 2015). The integration of RS and GIS allows easy storing, overlaying, and analyzing the geo-data spatially in form of layers (Berndtsson and Larson 1987;Chou 1997;Saraf et al. 2004). The data that are related to water resources and urbanization are mostly in form of geo-data. Therefore, water resources and urbanization could be spatially analyzed and modeled by an integrated approach of RS and GIS (Chou 1997;Saraf et al. 2004;Odeh et al. 2015).
However, the approach that we used to carry out analytical modeling, to achieve our objectives, is an integrated approach of remote sensing and GIS (Fig. 5). The geo-data that were used in the approach are either downloaded from open source on the Internet or were gotten from the authorities and ministries in Jordan or generated by us by analyzing geo-data that we have. These geo-data are summarized in Table 1.
The mechanism of the approach that connected the geodata together could be divided into two parts as follows: (1) Estimation of the water balance equation parameters where the runoff would be representative for the surface water resource and the infiltration be representative for the groundwater recharge. There would be two scenarios where the first is to estimate the runoff and infiltration, assuming that there is no urbanization and the second is to estimate the runoff and infiltration with the layer of urbanization to evaluate the influence of urbanization on surface and groundwater resources. (2) Generate a digitized groundwater level map, drainage network, and catchment area borders that would be used to recommend which directions that urbanization and water harvesting have to toward for.
The precipitation (P) in the study area is mainly rainfall. The rainfall patterns are the driving force for the groundwater recharge in the study area (Berndtsson and Larson 1987;Bouraoui et al. 1999). They were generated according to interpolation by Inverse Distance Weighted (IDW) method as the following equation (Chou 1997): n is the Number of cells that is taken to obtain the unknown value, V is the known rainfall value, d ij is the distance between unknown cell value and known cell value, and P is the power.
The mentioned interpolation was done by ArcGIS 10.3 that is generated by ESRI (Environmental Systems Research Institute) and using the extension spatial analyst. Then, the patterns were converted to vector, so we estimate the evaporation, runoff, and infiltration for each pattern in the attribute of the vector.
The average rainfall values (mm/year) of 30 years of ten climate stations were used. However, the average recorded temperature (T) degrees (°C) for the same period was distributed to the rainfall pattern to estimate the actual evaporation (AE) (mm/year) according to the Blaney-Criddle equation as follows: However, (p) value in the above equation is the actual daily daytime hours to annual mean daily daytime hours expressed as a percent (Schneider et al. 1973).
For the runoff, we used a coefficient that changes by the land cover and landforms the main two factors that control overland flow in our case study. Accordingly, we classified the studied area with different runoff coefficients as follows (Schneider et al. 1973;Butler and Davies 2000): (1) Multi-units, attached urban with a runoff coefficient of 0.65 in the NNW and NNE of the study area.
(2) Multi-units detached suburban with a runoff coefficient of 0.50 in the west and SW of the study area.
(3) Steep non-clastic hard rock (mostly limestone) with a runoff coefficient of 0.35 in the east and NE of the study area.
We obtained a groundwater level map for the studied area for the Federal Institute for Geosciences and Natural Resources so-called BGR. By ArcGIS 10.3 software, we georeferenced the map and then digitized the groundwater level of the study area in form of shapefile (vector). The National Aeronautics and Space Administration (NASA) generated through Shuttle Radar Topography Mission (SRTM) 2000 a 30 m resolution Digital Elevation Model (DEM) for the study area. We extracted the drainage network water divide and drainage network density by the spatial analyst extension according to the approach that is described by Odeh et al 2015. ( The Landsat image is one of the most common satellite images that are used to detect the land cover over on the land surface (Chou 1997;McMaster 2002;Szypuła 2016). We used a Landsat image 2014 with 8 bands and 30-m resolutions. However, only the visible bands were used to carry out a supervised classification by minimum distance method to classify the study area into three classes: urbanization, rock, and soil. In the minimum distance method, we depend on vectors of each class and calculate the Euclidean distance from each unknown cell on the satellite image to the vector for each class. The cells are classified to the nearest class (Odeh et al. 2015McMaster 2002;Szypuła 2016). The accuracy of the classification reaches up to 0.79 that is acceptable for our purposes (Table 2).

Results and discussion
The urban area expands for accommodating the growing local and migrated inhabitants is a common process in the capital cities and governorates (Potter et al. 2009). Amman is a capital city and governorate that its urban area was expanded rapidly during the last decades as a result of the migrations fluxes from the surrounding countries and the high population growth rate of the native people (Potter et al. 2009; Department of Statistics 2014).
Water resources and moderate climatic conditions are attractive factors for urban expansion (Schneider et al. 1973;Stephenson 2003). In general, human beings in Jordan do prefer neither hot weather nor cold one and need water resources for drinking, construction, agricultural activities and domestic usages ). However, the three mentioned factors are allocated in the northwest of the study area (Fig. 6). The high elevations and near the Mediterranean Sea generate moderate conditions in the northwest while the highest rainfall quantity as a surface water resource and nearest depth to the groundwater level (groundwater resource) (Fig. 7) is in the northwest too. The influence of urbanization on water resources has not been studied adequately in yet Amman. The urban landscape has common negative effects on the hydrological cycle and hence on the water resources (Schneider et al. 1973;Saraf et al. 2004). This includes increasing runoff, evaporation, and decreasing the infiltration and hence decreasing the groundwater recharge too (Jyrkama et al. 2005). The influence of urbanization does not stop on the quantity, but extends to affect the quality too. This is because urbanization has point sources for pollution such as wastewater treatment plants; wastewater pipes defects, fuel stations, and factories (Butler and Davies 2000;Stephenson 2003).
Land policy use regulations to restrict how the land can be used. It has great importance to specify and orient in where the urban expansion should be carried out (Schneider et al. 1973;Butler and Davies 2000;Potter et al. 2009,). Amman governorate was established by unprofessional land policy according to weak urban planning that led there are capital mistakes in generating the spatial distribution of the land use. Accordingly, the urbanization was expanded in our case study over the area that has the highest amount of rainfall, because the citizen and the refugees were searching for moderately climate conditions (Potter et al. 2009;Perdew 2014;Odeh et al. 2017,).
However, the urbanization in that area increases the overland flow and generates in some regions in the city of Amman a flash flooding which caused tragic economical losses reach up to tens of million American dollars, since part of the flash flooding was outlet in shopping area spots (Potter et al. 2009;Perdew 2014;Odeh et al. 2017). Figure 8 shows how urbanization covered aggressively the water resources in the study area where it condensed mostly in the North West. Before urbanization, the runoff of water was generating streams in the governorate during the winter seasons. A famous river was in downtown that is so-called Sayl Amman (The Amman Creek) which was killed by the urbanization (Potter et al. 2009;Perdew 2014). Tens of springs their discharges were stopped as a result of decreasing the groundwater recharge where the elevations of groundwater level become below the elevation level of the springs (Salameh and Bannayan 1993;Potter et al. 2009;Perdew 2014). However, we evaluated the influence of urbanization on the surface rainfall and then on the groundwater recharge according to the techniques of overlaying layers and then erased in GIS (Fig. 9). We found that this approach is very useful, although it is simple. Figure 10 shows the estimated groundwater recharge quantity before and after the attack of urbanization over the study area. The groundwater recharge rates (mm/year), that were estimated by the water balance equation, were multiplied by the area (km 2 ) of each pattern to have groundwater recharges quantities. However, to analyze the values spatially, we carried out thematic maps that show how the recharges quantities change within the location in the study area and correlate that with the location of the urbanization areas. We found that the highest amount of groundwater recharges are on the area of the highest amount of rainfall in the North West are the urbanization is condensed too. We found that there is a tremendous loss of groundwater recharge because of urbanization reaching up to 55.00 MCM/year overall in the study area. These losses are directly proportional to the urbanization area. In the groundwater recharge patterns of 4 and 5, the urbanization areas are maximum and, therefore, the groundwater recharge losses are maximum that reaching up to 12 Million Cubic meters per year. In groundwater recharge pattern 1, there are no groundwater recharges losses, because there is almost no urbanization and, therefore, there are no losses. In groundwater recharge pattern 9, there is no recharge as the amount of rainfall over that pattern is zero. However, our model is the first for the study area and there is no other model that estimated the groundwater recharge to compare with Odeh et al. (2019), Al Qudah et al. (2015, Al-Bilbisi (2019).
The groundwater recharge losses of the urbanization have several negative impacts on the groundwater resources such as decreasing the groundwater level and increasing depth to water rapidly, increasing the groundwater salinity, and decreasing the discharge of the remained springs (Odeh et al. 2019). A further transient numerical groundwater model for the study area is a must to achieve sustainable groundwater management for the study area (Jyrkama et al. 2005;Odeh et al. 2015Odeh et al. , 2019.
There are several decisions that have to be taken from the decision and policymaker to stop the degradation in water resources of Amman in addition to the sustainable water resource management according to transient groundwater modeling as follows: (A) There must be a land policy to protect the groundwater recharge area from the attack of urbanization. The zone of groundwater recharge is the same zone of soil, forest, and agricultural activity, so the recommended policy would enhance the protection of soil and the ecosystems too.
(B) A river restoration project could be carried out in the downtown of Amman city to reduce the influence of flooding and improve the water resources in Amman.
(C) A water harvesting project for the flash floods that run during the winter must be carried out in the area that has a limited water resource to generate sustainable water resources development. Figure 11 shows the catchment basins in the study area. There are two factors that increase the probability of flash flooding: the high rainfall intensity and the small catchment area size (Bue and Conrad 1967;Butler and Davies 2000). The two factors are available in the North West catchment areas. Furthermore, most of these catchments are covered by urbanization that triggers the runoff and hence the flash flooding.

Fig. 10
The groundwater loss over the groundwater recharge patterns because of the urbanization Drainage network density is the measurement of the length of the river channel per unit area of the catchment. It has the unit of km-1. The hydrologist and geomorphologist use it to give vision about the landscape shape and runoff potential, respectively (Bue and Conrad 1967;Butler and Davies 2000). In our research, recognizing the potential of the runoff has great importance to specify the location of water harvesting sites. However, runoff magnitude and hydrograph peak increase with drainage density (Butler and Davies 2000;Saraf et al. 2004).
In the southeast of the study area, there are larger catchment areas that have strong drainage network density (Fig. 12) that encourage us to recommend a water harvesting project there, because the intensity of the drainage network intensity increases the efficiency of water harvesting. Such a water project would improve the urban development where the groundwater recharge is minimum, and furthermore, the soil units and ecosystems are limited there; therefore, the optimized area for urbanization would be there too in terms of soil and ecosystem management too.

Conclusions
High population growth rate and refugees are common reasons for urbanization. Water resources are an attractive feature for urbanization in the arid region because of water scarcity and stress. Urbanization expansion has negative impacts on water resources in terms of quality and quantity. However, this negative is extended to affect the soil availability, ecosystems, and landscape. Urbanization plays a major role in controlling the hydrological cycle in the cities through increasing the runoff and evaporation. As a result, the infiltration that represents the groundwater recharge would decrease too. The integrated approach of GIS and remote sensing generate several geo-data values that estimate the variables of the water balance equation. GIS could analyze water balance equation spatially and correlate it with the land-cover data. Tremendous of groundwater recharge losses could be generated by the urban expansion through increasing the runoff. Sustainable water resource management needs a professional spatial plan that controls the attacking urban expansion on water resources and moving toward the area that has less rainfall quantity, far groundwater level, away from any water creek or spring discharge. We found that river restoration and overland flow harvesting projects are requested in the