Applicability of Urban Water Simulation Models for Estimating Urban Water Balance of Kabul City: A Review

Abstract

Computational models have gained recognition as effective tools for estimation of urban water balance. Beyond personal skills, the selection of an appropriate model requires an understanding of the city’s water system, the capabilities of the model and data requirements. Kabul represents a rapidly urbanizing city with limited water and sanitation infrastructure. To analyze the urban water balance of Kabul, ten prominent open-source and commercial hydrological models were evaluated. The characteristics of the models along with their data requirements, calibration parameters, and applications are assessed through a review of previous studies and user manuals. The study demonstrates that assessing Kabul’s urban water balance requires explicit consideration of processes such as snowmelt, groundwater abstraction, surface water–groundwater interaction and irrigation. The urban water balance of Kabul and cities with similar conditions can be effectively modeled using tools such as MIKE SHE, SWAT, and WEAP. The flexibility of the MIKE SHE model and its ability to use time-varying raster data make it a viable option for analyzing water balance under changing land cover and climatic conditions. Lumped models account for limited spatial variability and rely on empirical fitting. In contrast, physically based models reduce reliance on empirical calibration. However, they are more data-intensive and complex than simpler conceptual models.

1. Introduction

Many cities around the world are facing water scarcity due to rising demand and dwindling water resources. Urbanization has reduced infiltration, leading to increased surface runoff and reduced groundwater recharge [1,2,3,4,5,6]. It is believed that climate change may further add to the severe conditions by reduction in snowfall, intensification of rains, and prolonged droughts [7,8,9,10].

Kabul, the capital of Afghanistan, has experienced rapid population growth since 2002 [11]. Urbanization has not only reduced groundwater recharge but has also increased groundwater extraction, as no external water source has been developed, and the demand for all the domestic and industrial water has been fulfilled by local groundwater [12]. Studies show that groundwater quality has deteriorated badly and is overexploited in recent years [13,14]. Due to the declining water table, public buildings and households are compelled to dig new wells. In some areas, shallow wells have gone dry, and residents are seen waiting in long lines to collect water for their homes. Many households, particularly in the informal hillside settlement areas, rely on expensive tanker-delivered water [15]. In October 2024, the United Nations Children’s Fund (UNICEF) warned that if no action is taken, Kabul’s groundwater might dry up by 2030 [16]. Figure 1 shows location of Kabul and the region using Streetmap in GIS.

To identify sustainable solutions to water scarcity, it is essential to analyze Kabul city’s urban water balance. The concept of water balance has been widely applied in urban water management studies. As described by C. S. B. Grimmond in the journal Water Resources Research in 1986, the urban water balance is based on the principle of mass conservation and accounts for all inflows and outflows within an urban area. The urban area is considered as a single unit extending from roof level to the subsurface boundary below, where no net water exchange occurs [17]. Modeling urban water balance helps in understanding the water system, analyzing problems, and evaluating alternative solutions [7,18]. Over the last two decades, computational models have gained recognition as effective tools for assessing the impact of climate change and anthropogenic activities on urban water balance. Due to the increasing emphasis on urban water management in Australia and Europe, most hydrological models have been developed in these regions, and their applications have largely concentrated there [19].

Models are typically designed to address specific research questions, and many may not be directly applicable to cities in developing countries, which face unique challenges in terms of infrastructure development and data availability. A significant portion of the population in these cities lacks access to centralized water supply infrastructure. Domestic, commercial, and industrial users pump water from wells without any monitoring or records of water use [11]. Even though many such cities are facing widespread droughts and shall face an unprecedented water crisis in the future, limited research has been conducted on modeling the urban water balance [20]. Unplanned settlements, undocumented water use, and the lack of monitored discharge data make estimating the water balance challenging and uncertain [21,22].

The selection of an appropriate hydrological model for the analysis of the urban water balance necessitates a review of peer-reviewed studies employing hydrological models. The function of review articles is to aid readers in the compilation and synthesis of information [23]. However, many available review papers on hydrological models focus on model types, structural characteristics, and the degree of integration. In their review article, Mitchell et al. [24] screened 65 models based on five-point criteria and found only seven models (UVQ, Hydro Planner, Krakatoa, Mike Urban, Urban Cycle, and WaterCress) to be integrated urban water models. Bach et al. [25] proposed a typology for the classification of models according to the degree of integration, and the urban water simulation models were classified into four levels of integration. Kant et al. [26] conducted a review of various hydrological models, addressing their typology, methodological framework, and the outcomes of these models in facilitating comprehension of the multifaceted interactions within the hydrological cycle. In order to assist researchers and decision-makers in selecting a model, Pena-Guzman et al. [19] reviewed key components and applications of models developed between 1990 and 2015. Mosleh et al. [27] have developed a methodology for selecting a hydrological model based on its advantages and applications; however, it does not identify the specific data requirements of each model. In this review study, ten hydrological models that were most frequently used in urban water balance studies were evaluated to determine their suitability for estimating the water balance components of Kabul city. The findings were compiled and presented in this review paper to support researchers in urban water management.

This study aims to assist researchers in selecting an appropriate model to simulate the urban water balance in their study area. Kabul represents a rapidly urbanizing city with limited water supply and sanitation infrastructure under semi-arid climatic conditions, and the findings of this study may assist researchers working in analysis of urban water balance of cities with similar conditions. The study found MIKE SHE to be capable of simulating the urban water balance of Kabul city due to its ability to represent multiple interconnected hydrological processes, such as groundwater–surface water interactions, irrigation, and groundwater abstraction, while incorporating time-varying distributed spatial data. However, the model requires considerable computational resources and extensive spatial datasets for effective application. The findings also highlight the need for improved model adaptability across diverse urban contexts and enhanced data collection and management practices, thereby providing insights relevant to both model developers and urban water managers.

2. Materials and Methods

This study evaluates the applicability of computational models for urban water balance assessment in Kabul. In the first step, a comprehensive review of the published literature and institutional reports was conducted to examine the natural and anthropogenic factors affecting the urban water balance of Kabul. For each component of the urban water balance, the availability of relevant data is evaluated, indicating whether it can be obtained from online sources, acquired from organizations working in the water sector, or is currently unavailable.

A web-based literature search was conducted using Google Scholar to identify both open-source and commercial hydrological models capable of simulating components of the urban water balance. Peer-reviewed studies employing hydrological models for urban water balance analysis were collected and organized in Citavi 6.20. Ten models including ABIMO, MIKE SHE, Aquacycle, UVQ, WEAP, WABILA, SWAT, SWMM, HEC-HMS, and WaterCress were found to be more frequently used in these studies and were therefore selected for detailed evaluation. Models such as SuWaMBA [28], CWB [28,29], SUEWS [30], Hydro Planner [31], DUWSiM [32], and WaterMet2 [33] were not included in the detailed evaluation due to their limited applications in urban water balance studies. Along with peer-reviewed articles, user manuals, technical documentation and academic theses employing the selected models were reviewed to determine the applicability and limitations of each model for urban water balance assessment in Kabul and the findings are synthesized in the Results and Discussion section. The schematic in Figure 2 describes the method of review and the model selection process.

Ten hydrological models were selected for detailed evaluation, and a sequence of screening criteria was applied to identify the most suitable model for simulating the urban water balance of Kabul and similar cities. The first screening criterion evaluated whether the model could conceptualize key hydrological processes, including snowmelt, irrigation, surface water–groundwater interaction, and groundwater abstraction. The second criterion assessed whether the required input data were available or whether key parameters could reasonably be estimated using satellite products or values reported in the literature. Based on these two criteria, several potentially suitable models were identified. However, models primarily developed for long-term simulation were excluded because they are less suitable for detailed distributed hydrological process simulation.

Models capable of simulating daily hydrological processes were further evaluated based on their spatial representation. Lumped models were considered Unsatisfactory due to the significant heterogeneity in the topography, geology, land use, and urban development characteristics of Kabul city. The remaining semi-distributed and fully distributed models were then assessed according to their process representation approaches. Some models were mainly conceptual in nature, whereas others combined empirical methods for runoff estimation with physically based approaches for selected hydrological processes such as groundwater flow, evapotranspiration, and lateral flow. Preference was given to the model capable of representing all major hydrological processes using physically based methods, including overland flow, unsaturated flow, saturated flow, and channel flow. In addition, the capability to incorporate time-varying distributed raster datasets was considered an important advantage for analyzing urban water balance under changing land cover and climatic conditions.

3. Results and Discussion

3.1. Water Cycle of Kabul City and the Availability of Data

Kabul is the capital of Afghanistan and is located between latitude 34°31′ north and longitude 69°12′ east at an altitude of 1800 m (6000 feet) above sea level. Increased temperature due to climate change has resulted in a decrease in snowfall and an increase in heavy rains [34,35]. Increased temperature is melting the snow rapidly, resulting in less recharge of the aquifers [9]. This phenomenon is affecting the hydrological cycle, and cities that depend on local groundwater face challenges in meeting their water demand [20]. Climatic data, including precipitation, temperature, relative humidity, and wind speed, can be obtained from Ministry of Energy and Water (located in Kabul). Data on evaporation is not available and can be estimated based on minimum and maximum temperatures using Penman–Monteith or Hargreaves Equations.

Topography plays a crucial role in hydrological models and hydrologically corrected digital elevation models (DEMs) can be used for this purpose. DEM with a spatial resolution of 30 m can be downloaded from United States Geological Survey Shuttle Radar Topography Mission (SRTM) datasets [35], and a higher resolution DEM (5 m) can be obtained from the Ministry of Urban Development and Housing. The DEM, river network, and plan of the drainage canal were used for the identification and limitation of the study area. Figure 3 shows the study area, rivers, canals, landcover and location of climate stations managed by MEW.

Inflowing rivers are an important component of the Kabul water cycle and need consideration as they provide water for irrigation, recharge groundwater, and contribute to the outflow recorded at the end of the study area. Discharge in rivers passing through Kabul (Maidan, Paghman, and Logar) is highly dependent on snowfall. River leakage from these rivers is the main source of recharge for aquifers of Kabul [36]. Discharge recorded at gauges shows that in recent years, flow in these rivers increases in the month of February instead of March and is higher in peak months as compared to historical records. Wazirabad canal and Kabul River are the main two outlets from Kabul city. Average daily discharge data is recorded at the Tangi Gharo outlets but discharge data for the Wazirabad drainage canal is not available. Recorded discharge data at upstream and downstream of the city can be obtained from the MEW and can be used for the setup, calibration and validation of a hydrological model. Data about x-sections of the rivers, which provides information about the hydraulic head, width and depth of the rivers, is not available; however, DEM data can help in its estimation. Qargha Dam, Ghazi Dam, and a natural water lake called Kol-e-Hashmat Khan are surface water storage bodies that reduce runoff and recharge groundwater through infiltration. Location of weirs can be marked on USGS areal imagery but data about other technical details such as depth, flow and height are not available.

Despite certain central urban areas, centralized sanitation systems are absent. Consequently, individual houses and apartment buildings collect blackwater, containing human waste, in sanitation tanks that are later emptied by tankers and transported to wastewater treatment plants [37]. Greywater, primarily kitchen and sink water, is directly discharged into nearby roadside drains, ultimately flowing into the Kabul River or Wazirabad canal without any treatment and its seepage into aquifers has deteriorated the quality of groundwater in Kabul [37,38,39,40,41]. Except for some large canals, data of the drainage network is not available.

The agricultural areas shown in Figure 3 use water from rivers. In the dry period of the year, when rivers are dry, farmers use groundwater for irrigation. In recent years, farmers have been using solar-powered pumps to abstract groundwater, resulting in overexploitation [42,43]. The Ministry of Agriculture, Irrigation and Livestock located in Kabul, has records of the annual area under cultivation, and its yield in each province; however, the distributed data of crop types is not available. Land cover of the study area shown in Figure 3 can be downloaded for Afghanistan from the data repository of the International Center for Integrated Mountain Development (ICIMOD) [44]. The data has classified most of the agricultural areas in the western Paghman districts as fruit trees and grapes as a major crop in the northern districts 15 and 18 are classified as vineyards (grapes). Surface roughness, detention storage, and leakage coefficient are highly sensitive parameters that affect the overland flow and can be estimated based on land cover [45]. Vegetation parameters such as leaf area index (LAI), root depth, and crop coefficient (Kc) can also be estimated based on land cover data [46].

Distributed hydrological models need soil grid data for the estimation of irrigation demand and the simulation of flow through the unsaturated zone. A raster dataset known as the Harmonized World Soil Database developed by the Food and Agriculture Organization (FAO) of the United Nations can be downloaded from the FAO website. HWSD v2.0 is a comprehensive global soil inventory that provides detailed information on soil morphology, chemistry, and physical properties, with a spatial resolution of 1 km [47]. The soil map of the Logar and Kabul Valley can also be retrieved from the archives of the European Soil Data Centre (ESDAC). The map, originally produced in 1977 by the French Scientific Mission in Afghanistan in collaboration with the National Centre for Scientific Research of the French Republic, is available in PDF format. It identifies 24 distinct soil texture classes. However, soil profile data describing soil types at different depths, which are required to simulate water movement in the unsaturated zone using the Richards equation and the gravity-flow method, are not available. Data of well lithology is available from MEW and can also be found in the sector report (ground water) of the study for the Study for the Development of the Master Plan for the Kabul Metropolitan Area, however these well are located in urban areas and the data can only be used for preparation of soil profile in urban areas.

The geological map of the study area prepared by the United States Geological Survey (USGS) can be obtained from Ministry of Energy and Water. Distributed data showing the depth of each geological layer is not available; previous studies have conceptualized the x-section of geological layers in Kabul and Logar sub-basin. The study titled “Inventory of Groundwater Resources in Kabul, Afghanistan” [48], conducted by the USGS, includes an unscaled geological cross-section of the study area. The online available report of the study conducted by the German Federal Institute for Geosciences and Natural Resources (BGR) developed geological sections for the Kabul River valley and Logar River valley. This study conceptualized three geological layers: loam, sand and gravel, and conglomerates/sandstone [49]. It also estimated the depth of each layer based on borehole data. The findings of this study were further refined in “The study for the development of Master Plan for Kabul Metropolitan Area” [50].

Since November 2006 MEW has recorded monthly ground water table elevation in 127 wells. However, there is a data gap between May 2009 and November 2013. The Danish Committee for Aid to Afghan Refugees (DACAAR) has also Monitored groundwater table elevations and electrical conductivity in four wells within the study area, with records available since 2000. Point data of water table elevation can be interpolated to get raster representing initial potential head and can also be used for calibration of spatially distributed models with a groundwater simulation. Simulation of groundwater flow requires data on water extracted from groundwater resources. Due to less coverage of the centralized water supply system, such data is not available. Domestic, industrial, and agricultural users abstract water from wells without monitoring. Moreover, due to the presence of informed settlements and incomplete data, precise population estimates for neighborhoods remain elusive. Land cover data and open access distributed population data such as Gridded Population of the World (GPW) [51], Land Scan [52], WorldPop [53], and Global Population Data [54] can be used for the estimation of population.

Modeling urban water balance requires consideration of all inflows and outflows within the study area. Kabul demonstrates notable heterogeneity in its topography, geology, and urban characteristics. To model the urban water balance of Kabul city, it is essential to select a model capable of representing this heterogeneity and the key components of the city’s water cycle, including rainfall–runoff, snowmelt, river–groundwater interaction, irrigation, and groundwater abstraction.

Table 1. Hydrological processes and available/unavailable data for Kabul city.

Component of Water Cycle	Data Available	Source	Data Not Available
Rainfall–Runoff/Snowmelt	Precipitation, temperature	MEW, AMD, MAIL	Evaporation
Overland Flow	DEM Land cover	USGS, ICMOD	Detention storage, surface roughness, surface subsurface leakage and paved area fraction
Urban Drainage	Map and x-section of few canals	Kabul municipality, MOUD	Drainage network, discharge data at canals
Rivers and Lakes	River path, discharge data Location and area of dams and lakes	MEW	River cross-sections, geometry of weirs and other structures, depth and storage capacity of lakes
Irrigation	Land cover, shapefiles of irrigation canals	MAIL, GIS online	Crop types, irrigation demand
Unsaturated Zone	Soil texture data	UN-FAO, MAIL	Soil profile data
Saturated Zone	Water table elevation in wells Well lithology Cross section of geological layers	MEW, DACCAR Literature	Raster data for initial potential head Raster data for depth geological layers, groundwater abstraction

presents the available and unavailable data for each hydrological process, along with their respective sources.

3.2. Modeling Urban Water Balance

Water balance models consider both natural and anthropogenic conditions to determine the quality and quantity of water at desired spatial and temporal scales, depending on the input data’s temporal and spatial resolution. Although all urban water balance models follow the principle of mass conservation of water, they may vary in input data requirements. Grimmond and Oke (1986) [17] defined urban water balance as a box with a unit surface horizontal area that extends from roof level to a depth in the ground below where no net exchange of water occurs throughout the interest, and expressed it as below:

P + I = r + E + ∆S

(1)

where P is precipitation, I is the piped-in water supply, r is net runoff, E is evapotranspiration, and ∆S is the net change in water storage. The water balance equation forms the core of nearly all hydrological models and accounts for changes in water storage within a system. One portion of the received precipitation infiltrates into the soil and combines with groundwater while the other portion of the rainfall evaporates from the roof and ground surfaces. The remaining rainfall becomes runoff and joins the drainage system. Mitchel et.al [55] divided the runoff (r) in Equation 1 into stormwater runoff (Rs) and wastewater runoff (Rw) and expressed the equation as below.

∆S = (P + I) − (Ea + Rs + Rw)

(2)

Change in storage in Equation (2) represents the sum of the changes in groundwater and surface water storage. Separating these two components gives Equation (3), where the water infiltrated into the groundwater aquifers represents the following:

dS/dt = (P + I) (Ea − f − Rs − Rw)

(3)

Each component of Equation (3) represents a critical process within the urban water cycle. Precipitation serves as the primary input and can be represented in models as either a uniform or spatially distributed variable, depending on the spatial resolution and complexity of the modeling framework. Actual evapotranspiration (Ea) is typically estimated from reference evapotranspiration values or climatic parameters, with adjustments based on soil moisture availability. Wastewater discharge (Rw), another key component, is directly related to urban residents’ water consumption. In areas with centralized water supply and sanitation systems, it can often be quantified using data from water meters or wastewater treatment facilities. Stormwater runoff (Rs) and infiltrated water are largely governed by rainfall intensity, land surface characteristics, and the hydrogeological properties of the catchment and subsurface materials.

To simulate these components, hydrological models employ various methodological approaches that differ in how they treat physical processes and in their data requirements. Physically based models use physical parameters such as soil hydraulic conductivity, porosity, slope, and land use characteristics, and typically solve partial differential equations to capture spatial and temporal variations in hydrological processes. These models are data-intensive but offer high accuracy and process representation [56,57]. In contrast, conceptual models simplify the representation of the hydrological system by employing empirical relationships and aggregated indicators in place of physically measurable parameters [56]. The catchment is typically modeled as a series of interconnected storage sections, such as surface water, unsaturated soil, and groundwater, and fluxes are estimated using simplified or calibrated equations. When applied to an entire catchment with average input parameters, these are considered lumped models [58]. If spatial heterogeneity is accounted for by subdividing the catchment into smaller units (e.g., grid cells or hydrologic response units), the model becomes semi-distributed or fully distributed, depending on the level of spatial detail [56]. Regardless of the model type, calibration and validation are essential steps to ensure reliable predictions. Models are commonly calibrated using observed data such as surface runoff, groundwater levels, or lysimeter measurements. This helps in the adjustment of certain parameters to achieve alignment with real-world behavior.

3.2.1. Aquacycle

Aquacycle was developed by Grace Mitchell [55] and can be freely obtained from the Cooperative Research Centre for Catchment Hydrology, Department of Civil Engineering of Monash University [59]. It simulates components of water balance at various spatial scales by using conceptual, lumped modeling approaches at a daily time step [60].

Aquacycle requires dividing the study area into clusters and unit blocks. A unit block represents an industrial, commercial, or institutional land block. Input data for each unit block helps the model estimate total previous and impervious land as well as the occupancy. Using this information and climatic data, the model estimates evapotranspiration, stormwater runoff, and potential rainwater storage. Input data of indoor water usage enables the model to estimate wastewater generated and whether it can be allocated for reuse [59]. These capabilities enable the model to analyze the impact of wastewater reuse and rainwater harvesting on urban water balance [61]. The model requires data on 15 measured parameters, and the values of 16 calibration parameters need to be calibrated. Eleven of these parameters relate to stormwater flows, three to wastewater flows, and two to water use. To test the performance of the Aquacycle model in an urban catchment, stormwater runoff and wastewater discharge data are required [55].

3.2.2. Urban Volume and Quality (UVQ)

The Urban Volume and Quality (UVQ) model is an extension of the Aquacycle model with the incorporation of containment balance [62]. It is a daily time step, conceptual model developed by Grace Mitchell and C. Diaper for simulation of water and containment flows within an urban water system [40]. UVQ operates on the same principles as Aquacycle; therefore, the study area should be divided into neighborhoods with similar water demand patterns, land use types, and soil infiltration characteristics [63].

UVQ contains approximately 300 input fields to accommodate a wide range of urban water scenarios and contaminant indicators; only 18 of these are calibration parameters [63]. In addition to climatic data, water usage data, land use data, demographic data, contaminant load for potable water, rainwater, bore water, evaporation, runoff from roof, road, and paved area, kitchen, bathroom, laundry, and toilet, and fertilizer application are also required. Calibration can be carried out with wastewater and stormwater flow and contaminant data [62].

3.2.3. ABIMO

ABIMO was initially designed by the German Federal Institute of Hydrology for hydrologic analysis in the quaternary area of central Germany and was later modified for urban areas. This lumped model uses conceptual representations to estimate evapotranspiration, surface runoff, and seepage water from total precipitation [64]. This is done using look-up tables and empirical formulas that are based on land use, soil type, and climatic data [65]. In the first step, ABIMO uses the Bagrov equation to estimate actual evapotranspiration and subtracts it from total precipitation. In the second step, the model divides the remaining water between infiltration and surface runoff as per the soil type, depth to groundwater, surface imperviousness, and sewer connectivity of the study area [64]. Look-up tables used by ABIMO are prepared for Berlin city and need to be updated for use in cities other than Berlin. The model can be used for analyzing the impact of land use change on evapotranspiration, runoff, and infiltration, and therefore can estimate the potential increase in surface runoff due to urbanization [66].

3.2.4. WABILA

WABILA is the short form of Wasserbilanz, which is the German word for “water balance”. It is a lumped, conceptual water balance model developed by Prof. Dr.-Ing. Mathias Uhl, Malte Henrichs, and Julian Langner at the Institute for Water Resources Environment (IWARU), Münster University of Applied Sciences, Germany [67]. The model is designed to assist urban planners and water managers in evaluating the impacts of land use changes and Water Sensitive Urban Design (WSUD) measures on the local hydrological cycle [68].

Using the provided land use data, WABILA aggregates the study area into land use classes with similar hydrological properties. It uses climatic data to estimate annual evapotranspiration, infiltration, and runoff for each land use class using a simplified water balance equation [69].

The main purpose of WABILA is to serve as a decision-support tool and to analyze the effects of different stormwater management strategies. Different urban planning scenarios and WSUD options can be evaluated for their impact on runoff generation, infiltration, evapotranspiration, and urban water retention. Being an event-based lump model, it is difficult to use WABILA for continuous and detailed simulations.

3.2.5. WaterCress

The WaterCress model was developed by the Australian company Water Select. It is an open-source, freely available model and can be downloaded from the company’s website https://www.waterselect.com.au/index.html (accessed on 20 May 2024). It is a conceptual model that uses simplified, lumped representations of hydrological processes (e.g., runoff, infiltration, evaporation) [70].

The user can select one from the eight available rainfall methods to estimate runoff from a rural catchment, but for an urban catchment, only the Initial Loss–Continuing Loss (ILCL) model can be used. In addition to estimating runoff from precipitation, the model can include direct inflows. Based on the input data, generated runoff and added direct inflow can be routed to storage nodes and then to demand nodes, depending on the required demand quality and the quality of the available water. A successful run of the model provides statistics on flows and storage within the water system over the simulation period and can help analyze the performance of the current water system against desired outcomes. It can be used to explore alternative system layouts at the feasibility stage [71]. Rainwater and treated wastewater can be directed to domestic, industrial, and agricultural water demand, and their impact can be analyzed on urban water balance [71]. The model has been used for water resources studies in Australia and has helped in analyzing the performance and implications of alternative system designs [72].

3.2.6. WEAP

WEAP is a conceptual, semi-distributed, and scenario-based water management model that is developed by the Stockholm Environment Institute (SEI) [73]. Like WaterCress, it uses a node–link structure for the representation of the water system and calculates a water and pollution mass balance for every node and link in the system. Water flow is simulated to meet instream and consumptive requirements, based on demand priorities, supply availability, infrastructure constraints (e.g., capacity, flow limits), and environmental flow requirements.

It provides users with a choice to use one of the five available methods for the simulation of catchment processes such as evapotranspiration, runoff, infiltration, and irrigation demands. These methods include (1) the rainfall–runoff and (2) irrigation demands only versions of the simplified coefficient approach, (3) the soil moisture method, (4) the MABIA method, and (5) the plant growth model or PGM [73]. The model is widely used for water resource allocation in large catchments and needs data on water resources and water demand [74,75]. It can analyze future scenarios based on alternative assumptions about climate change, water demand, and supply policies [76,77,78].

WEAP and WaterCress are more often used as water resources allocation models; both use node–link structures for the representation of the water system and need demand quality and quantity data from domestic, agricultural, and industrial users [79]. The model can be integrated with GIS for data input and visualization of results [80]. The model can be dynamically linked with the MOFLOW model for simulation of groundwater flow. In the reviewed studies WEAP coupled with MODFLOW has been used for the simulation of monthly water balances for the aquifer, root zone, rivers, canals, and reservoirs [75,81,82].

3.2.7. SWMM

The Storm Water Management Model (SWMM) is an open-source model that is developed, continuously updated, and distributed by the United States Environmental Protection Agency. It is a dynamic, semi-distributed hydrological and hydraulic model used for the simulation of runoff quantity and quality from urban areas. Since 1971, SWMM has been applied globally for planning, analysis, and design of combined sewers, sanitary sewers, and other drainage systems, and estimation of stormwater runoff and analysis of flooding in urban areas [83].

To conceptualize a catchment in SWMM, the area under investigation is divided into sub-catchments, making it a semi-distributed model. Rainfall on each sub-catchment is handled using simplified processes, such as the Horton, Green-Ampt, or Curve Number methods [84]. However, the hydraulic component of SWMM is physically based, as it uses the Saint-Venant equations to simulate unsteady, dynamic open-channel flow [85]. The SWMM is widely accepted by the design community for urban stormwater applications and is used for planning, designing, analysis, and research purposes. Applications of the SWMM include designing drainage systems, analysis of urban flooding, and analyzing the impact of green infrastructure, e.g., rain gardens, permeable pavements, or bioswales, on the quality and flow of stormwater runoff [86,87,88,89]. The recently updated version, EPA SWMM 5.2, has the capability to incorporate seepage in links, thus making it possible for the users to measure infiltration of water in canals, rivers, and ditches [83]. However, the model cannot be dynamically linked to aquifer storage due to the reason that groundwater flow is not directly simulated within the model.

3.2.8. HEC-HMS

The Hydrologic Engineering Center’s Hydrologic Modeling System (HEC-HMS) is a widely used hydrologic simulation software developed by the Hydrologic Engineering Center (HEC) of the U.S. Army Corps of Engineers. It is designed to simulate rainfall–runoff processes in natural and urban watersheds [90].

HEC-HMS is a conceptual, semi-distributed hydrologic model that uses a modular and component-based approach [91]. The basin model represents the overall hydrologic system and is used to calculate losses due to infiltration and evapotranspiration. The transform method converts this excess precipitation into direct runoff by generating a hydrograph at the watershed outlet, accounting for watershed surface characteristics and geometry. The baseflow method simulates groundwater contributions to total runoff. Finally, the routing method models the movement of water through channels and rivers across sub-basins. Together, these components allow HEC-HMS to simulate the rainfall–runoff processes of complex watershed systems with flexibility and adaptability [91,92,93].

HEC-HMS is widely applied in flood forecasting, watershed hydrology, dam safety studies, urban drainage planning, and climate change impact assessment [94,95,96,97,98]. Its integration with HEC-GeoHMS (a GIS extension) facilitates spatial preprocessing and visualization. Climatic data, topographic data, land cover, soil texture and stream network are the mandatory items required for successful runoff modeling while other parameters can be calibrated with records of stormwater gauges/wastewater treatment plan, lysimeter data and snow depth.

3.2.9. SWAT

The Soil and Water Assessment Tool (SWAT) is an open-source, river basin-scale, hydrological model developed by the United States Department of Agriculture (USDA) Agricultural Research Service (ARS). It is an integrated, continuous-time, semi-distributed, process-based model that operates at a daily time step [99,100].

The hydrologic cycle simulated by SWAT is based on the water balance equation. The subdivision of the study area into Hydrological Response Units (HRUs) makes the model semi-distributed and enables it to consider differences in evapotranspiration, infiltration, and runoff for various crops and soils. Runoff is predicted separately for each HRU and routed to obtain the watershed’s total runoff [100]. Although SWAT uses the empirical Curve Number method to estimate surface runoff, it also applies physically based methods like Darcy’s Law for groundwater flow, the kinematic wave for lateral flow, Penman–Monteith or Priestley–Taylor for evapotranspiration, and the variable storage method for channel flow [100,101].

SWAT is a widely used tool for simulating the quantity and quality of surface and groundwater resources, as well as evaluating the environmental impacts of land use change, land management practices, and climate change on the quality and quantity of water resources in a watershed [99,102,103]. SWAT simulates canopy interception, infiltration, evapotranspiration, soil water movement, surface runoff, groundwater recharge, and streamflow [104]. The web database https://www.card.iastate.edu/swat_articles/ (accessed on 12 June 2024) shows that the model is used in more than 6204 journal articles for various applications. After filtering the database for the keyword “Land use change,” it was noted that in nearly 150 journal articles, the model is used for analysis of the impact of land use change on hydrological response.

3.2.10. MIKE SHE

MIKE SHE is a physically based, distributed hydrological model originating from the concept proposed by Freeze et al. (1977) [105]. It was first adopted by a consortium of three European organizations (System Hydrologique Européen) in 1977. Since the mid-1980s, DHI Water & Environment [106] has further enhanced and expanded it.

MIKE SHE integrates the entire land phase of the hydrological cycle and covers the major processes of the hydrologic cycle, including evapotranspiration, overland flow, unsaturated flow, saturated flow, and channel flow [106]. Key natural and anthropogenic processes affecting the water balance, such as snowmelt, groundwater abstraction, surface water–groundwater interaction, and irrigation, can be conceptualized in a flexible manner. The flexibility of MIKE SHE comes from its ability to allow users to choose between physical and conceptual methods for simulating water movement in the overland flow, unsaturated zone, and saturated zone. The selection of the appropriate method is generally based on data availability, modeling objectives, and computational capacity. Either the finite difference method or sub catchment-based methods can be adopted for the simulation of overland flow. Richards’ equation, the two-layer unsaturated zone method, or gravity flow can be used to simulate water movement in the unsaturated zone. Similarly, the model provides the option to select either the finite difference method or the linear reservoir method for simulating water movement in the saturated zone [107].

MIKE SHE is a catchment-scale model that has been widely applied at the watershed scale to simulate hydrological components of the water balance [108,109,110,111]. Surface water bodies such as rivers, lakes, and dams can first be modeled in MIKE+ and then linked with MIKE SHE. River paths and cross-sections can be imported from GIS shapefiles or derived from a digital elevation model (DEM). It uses raster data in DFS2 (Data File System 2-dimensional) format. The MIKE Zero Toolbox can be used for conversion of raster data from ASCII format to DFS2 format. Time-varying distributed data for Manning’s coefficient and surface–subsurface leakage can also be incorporated, making the model a suitable tool for estimating the impacts of climate change and anthropogenic activities on surface water and groundwater resources in urban environments [46,112,113]. According to Qiang et al. [114], the model can overcome missing-data problems and effectively represent complex hydrological systems. However, the most frequently cited challenges are the long computational times required for complex simulations and the high demand for spatially distributed data. These constraints have historically limited its application, particularly in regions with scarce data availability. To facilitate a direct comparison,

Table 2. Applications, input data requirements, calibration parameters, and calibration data of the reviewed models.

Model Name	Applications	Input Data	Calibration Parameters	Data for Calibration
Aquacycle	Simulation of water supply, stormwater, and wastewater flows [55,60,115,116] Evaluation of the impact of alternative scenarios on UWB [61,117,118,119,120] Analyze the impact of land use change on UWB [119,121]	Precipitation and potential evaporation Water used for kitchen, bathroom, laundry, and toilet applications Number of unit blocks, average block size, and cluster area Area of roof, garden, road, paved, open space, irrigated agriculture, and rainfed agriculture Average household occupancy imported water (if any)	Percentage areas and capacities for Store 1 and Store 2 Initial loss and effective areas for roof, paved, and road surfaces Baseflow index and recession constant Surface runoff inflow percentage Infiltration index and recession constant Irrigation triggers for gardens and public open spaces	Stormwater outflows Wastewater outflow
UVQ	Analysis of the impact of potential scenarios of water management on urban water and containment balance [122,123,124] Quantification of flows and contaminant loads to urban aquifers by using UVQ in linkage with the pipe leakage model NEIMO [63,125,126,127]	Daily precipitation, potential evaporation, and temperature Water used for kitchen, bathroom, laundry, and toilet applications Area of roof, garden, road, paved, open space, irrigated agriculture, and rainfed agriculture Average household occupancy Imported water (if any) Contaminant load for potable water, rainwater, bore water, evaporation, runoff (from roof, road, and paved area), kitchen, bathroom, laundry and toilet, and fertilizer application	Percentage of previous and impervious areas Initial loss and affective areas of roof, paved, and road areas Initial losses for roof, paved, and road surfaces Percentage of surface runoff contributing as inflow Baseflow index and recession constant Infiltration index and recession constant Trigger-to-irrigate values for gardens and public open spaces	Stormwater outflows and concentrations Wastewater outflow and concentrations
ABIMO Model	Estimation of the components’ water balance [2] Analysis of the impact of land use change on UWB [66,128]	Precipitation and potential evaporation Land use and imperviousness Sewer connectivity Soil types Groundwater table	Runoff coefficients Infiltration rates Evapotranspiration factors Soil storage capacities Percolation thresholds	Discharge at outlets/WWTP Lysimeter or pan evaporation data
WABILA	Evaluation of the hydrological impacts of urban development and WSUD strategies [67,68]	Annual precipitation, potential evapotranspiration Digital elevation model Proportion of different surface types/land cover such as roofs, paved areas, unpaved areas WUSD measures Performance coefficients such as retention and infiltration rate of WUSD measures Hydraulic conductivity and storage capacity	Rainfall–runoff coefficient (a) Groundwater recharge (g) Evapotranspiration coefficient (V) Baseflow index Soil water holding capacity	Discharge at outlets Discharge at wastewater treatment plants Lysimeter data
WaterCress	Assessment of the security of urban water supply with non-traditional sources under climate change [129,130,131] Simulation of runoff, recharge, and recovery for different rainfall, catchment, and aquifer conditions [71,132,133] Regional optimization of water supply options using alternative sources and multiple objectives [72,134,135]	Precipitation, potential evaporation Topographic maps (catchment areas, boundaries, etc.) Identification of roof, paved, open areas Water quantity and quality demand by each water user group Water supply system and water storage facilities Capacity of storage Transfer rates into and out of facilities	Runoff coefficients for different surfaces Infiltration rate Water demand and usage Rainwater tank size First flush diversion volume Reuse fraction or demand priority Wastewater generation fraction Treatment efficiency or losses	Records of stormwater gauges Flow data of wastewater treatment plants
WEAP Model	Urban water use and demand trend analysis [76,79,136]. Assessment of water resource allocation impacts on different demands [74,82,137] Integrated water resource management through coupling with MODFLOW [81,138,139]	Precipitation, temperature, and potential evapotranspiration Population, per capita water use, and industrial water demand Reservoirs, treatment plants, and pumping stations. Transmission and return flow losses Catchment boundaries and area, land use maps, and soil types Safe yield and recharge rate of aquifers Agricultural area, crop types, and irrigation methods River and stream flow, reservoirs	Runoff resistance factor, root zone conductivity, hydraulic conductivity, maximum root zone storage, deep water storage capacity, initial storage fraction, crop coefficients), leakage co., baseflow recession constants, aquifer conductivity, abstraction limits or efficiencies and evaporation rates	Flow data at catchment outlets. Lysimeter data Snow depth
SWMM	Assessment of hydrological impacts of urbanization and climate change in urban areas [140,141,142,143,144,145] Evaluation of green infrastructure and low-impact development effects on urban water balance components [82,85,146,147,148,149]	Precipitation, temperature, potential evapotranspiration Sub-catchment area, width, %slope Soil properties Conduct shape, depth, roughness, initial flow, maximum flow, entry/exit loss coefficient, seepage loss rate Water table elevation, Aquifer bottom elevation X, Y coordinate, inflows, invert elevation, max depth, initial depth, ponded area, surcharge depth for each junction	%Imperviousness, N-perv, N-Imper, Depression storage for previous and impervious areas Surface roughness, initial retention, maximum flow, entry/exit loss coefficient, seepage loss, invert elevation of junctions, storage curve parameters Porosity, wilting point, field capacity, conductivity, unsaturated zone moisture Water quality calibration parameters such as built-up rate, wash off coefficients, and decay rates	Records of stormwater gauges Flow data of wastewater treatment plants Lysimeter data
HEC-HMS	Rainfall–Runoff modeling in ungauged watersheds [150,151,152] Urban runoff prediction and simulation [153,154] Flood forecasting and early warning modeling [155,156,157] Assessment of land use and land cover change impacts on runoff dynamics [158,159,160,161,162,163,164,165]	Precipitation, temperature data, and potential evapotranspiration Land cover Soil texture Digital elevation model Reach length Reach slope Stream network	Curve number, initial abstraction, initial loss, constant rate, suction head, hydraulic conductivity, initial moisture, maximum deficit Lag time, concentration time, storage coefficient Manning’s n, flow width, slope. % Impervious area	Records of stormwater gauges Flow data of wastewater treatment plants Lysimeter data Snow depth
SWAT	Analysis of the impact of urbanization and climate change on catchment water balance [104,166,167,168,169,170,171,172,173,174] Integrated assessment of urbanization impacts on surface and groundwater using coupled MODFLOW [175,176,177,178]	Precipitation, temperature, solar radiation, humidity, and wind speed (optional). Land use/land cover Soil texture and properties tables Digital elevation model River path, river sections and river inflow Type of vegetation	Curve number for runoff Baseflow alpha factor Groundwater delay time Threshold for return flow Available water capacity in soil Soil evaporation compensation factor Plant uptake compensation factor Effective hydraulic conductivity in the main channel Manning’s n for the main channel Surface runoff lag time Parameters related to nutrients and sediments	Stream flow records Sediment load Nutrient concentrations Lysimeter data Snow depth
MIKE SHE	Simulation of hydrological components of the total water balance in large watersheds [107,108,109,110,111,179] Simulation of surface runoff, surface storage change, and groundwater storage in urban areas [45,112] Flood risk modeling and mapping [180] Assessment of the influence of urban geology and spatial resolution on simulated shallow groundwater levels and flow at city scale [181,182]	Rainfall, temperature, and potential evapotranspiration River path, cross section, and inflow data Leaf area index, root depth, Kc irrigation command areas, crop types, irrigation demand and irrigation method Topography, Mannings coefficient, Detention storage, initial water depth, Surface–subsurface leakage (optional), paved area fraction (optional) Soil texture data Lower level of geological layers Vertical hydraulic conductivity Horizontal hydraulic conductivity Specific yield Specific storage Initial potential head of groundwater Groundwater abstraction	Root depth, leaf area index, and crop factor (Kc) Leakage coefficient Channel roughness (n) Manning’s (M) Detention storage Unsaturated zone Water content at saturation, Field capacity and wilting point of soil types ET surface depth Vertical hydraulic conductivity Horizontal hydraulic conductivity Specific yield Specific storage	Surface water level/discharge Groundwater level Lysimeter. Snow storage depth

summarizes the applications, input data requirements, and calibration data of all ten reviewed models.

4. Conclusions

Kabul demonstrates notable heterogeneity in its topography, geology, and urban development. Rivers provide water for irrigation and constitute an important source of groundwater recharge. Agricultural areas within and surrounding the city are dependent on both river water and groundwater for irrigation purposes. Consequently, irrigation and interaction between river and groundwater systems are critical components of the urban water balance. In the city of Kabul, groundwater supplies nearly 100% of domestic, commercial, and industrial water demand, meanwhile approximately 78% of the population lacks access to centralized water supply system and relies on private wells for their water needs. Consequently, groundwater abstraction represents a significant component of the city’s water cycle and must be incorporated into urban water balance modeling. These characteristics suggest that urban water balance modeling in Kabul, and in similar semi-arid and data-scarce cities, requires models capable of representing key hydrological processes such as snowmelt, river–groundwater interactions, irrigation water use, and groundwater abstraction, while also remaining applicable under limited data availability.

Lumped models conceptualize the catchment as a single unit or as a few homogeneous subunits. This methodological shortcoming results in an inadequate accounting for spatial variability. Aquacycle, UVQ, and WaterCress mandate the division of the study area into hydrologically homogeneous units, designated as unit blocks. In a city such as Kabul, where urban development is largely unplanned and the topography is uneven, the number of required unit blocks would be substantial. This phenomenon is attributable to the considerable variation in features such as roof area, garden area, and road area across different parts of the city. These models require data on indoor water use and household occupancy rates, which is not available for Kabul. In the absence of reliable data, estimation of these parameters would lead to a significant degree of uncertainty.

Conceptual models do not require detailed physical input data such as soil properties, aquifer characteristics, or land use distribution. Instead, they simplify the hydrological cycle into a set of empirical relationships. The calibration of these models is achieved through the adjustment of calibration parameters, ensuring that the model outputs align with observed data, such as river discharge or water table levels. The effectiveness of this approach is contingent upon the availability of comprehensive time series data for all major rivers, canals, and groundwater tables. Nevertheless, for calibration to be both meaningful and accurate, it is imperative that the various components of the water balance be thoroughly understood and quantified. In regions with limited data, such as Kabul, where documentation of certain components of the water balance is lacking, the calibration of conceptual models may yield misleading results. The utilization of ABIMO and WABILA models is incompatible with the implementation of continuous simulation. In addition, the look-up tables utilized by ABIMO have been prepared for Berlin city and require updating for application in cities other than Berlin. This process requires the establishment of a network of stormwater measurement gauges and the accumulation of several years of data.

The SWMM has the capability to simulate river flow by representing rivers as open channels and defining upstream inflow as direct inflow to upstream nodes. Recent versions of SWMM also permit the inclusion of river leakage; however, this leakage cannot be dynamically linked to aquifer storage because groundwater flow is not directly simulated within the model. As demonstrated in

Table 3. Model characteristics and capabilities.

Model	Spatial Representation	Dimensionality	Hydrological Approach	Typical Temporal Scale	Software Type	Hydrological Capabilities
Aquacycle	Lumped	1D	Conceptual	Daily	Open-source	Simulate urban water balance (UWB)
UVQ	Lumped	1D	Conceptual	Daily	Open-source	Simulate UWB and water quality
ABIMO	Lumped	1D	Conceptual	Annual/long-term	Proprietary	Simulate UWB
WABILA	Lumped	1D	Conceptual	Monthly or annual	Proprietary	Simulate UWB
WaterCress	Lumped	1D	Conceptual	Daily	Open-source	Simulate UWB and allocation of water resources
WEAP	Semi-distributed	1D	Conceptual	Daily or monthly	Proprietary	Simulate UWB, surface routing, resource allocation, irrigation, snowmelt, and groundwater–surface water (GW–SW) interaction; groundwater simulation possible when linked with MODFLOW
SWMM	Semi-distributed	1D/quasi-2D	Conceptual	Minute to hourly	Open-source	Simulate UWB, surface routing, resource allocation, irrigation, snowmelt and GW-SW interaction,
HEC-HMS	Semi-distributed	1D	Conceptual	Hourly to daily	Open-source	Simulate UWB, surface routing, snowmelt and GW-SW interaction.
SWAT	Semi-distributed	1D	Physical/conceptual	Daily	Open-source	Simulate UWB, surface routing, resource allocation, irrigation, snowmelt, and GW–SW interaction; groundwater flow simulation when coupled with MODFLOW
MIKE SHE	Fully distributed	3D	Physical-based	Hourly to daily	Proprietary	Simulate UWB, surface routing, resource allocation, irrigation, snowmelt, GW–SW interaction, and groundwater flow simulation when coupled with MIKE+

, of the ten evaluated hydrological models, WEAP, SWAT, MIKE SHE, and HEC-HMS are the only ones capable of representing groundwater flow processes. HEC-HMS and WEAP employ conceptual methods, including linear reservoirs and recession curves, to simulate groundwater movement and baseflow contributions. In MIKE SHE, groundwater flow can be simulated using either conceptual approaches, such as the linear reservoir method, or physically based approaches, such as the 2D finite difference method, depending on the study objective and data availability.

To simulate river–groundwater interaction, it is necessary to couple the SWAT and WEAP models with the MODFLOW model. Alternatively, the MIKE SHE model can simulate these processes through coupling with the MIKE+ River model. A review of the relevant studies indicates that SWAT-MODFLOW is generally more hydrology-oriented and suitable for analyzing watershed processes, recharge, and land use impacts. Conversely, WEAP-MODFLOW is more management-oriented and better suited for water allocation and urban water management scenarios.

Physically based models are founded on fundamental hydrological processes based on measurable physical characteristics such as topography, soil type, vegetation, and aquifer geometry. While these models also benefit from calibration, their reliance on physical parameters allows them to estimate unmeasured components more reliably and reduce dependence on empirical fitting. Therefore, in catchments with incomplete flow data, physically based models provide a more robust and valid approach for simulation of the urban water balance. A notable disadvantage of physically based models is their requirement for detailed input data and acquisition of such data can be challenging, especially in data-scarce regions. Additionally, these models tend to be more complex and computationally demanding, making their setup, calibration, and use more difficult compared to simpler conceptual models.

A review of case studies indicates that the MIKE SHE model provides a comprehensive and fully integrated framework for simulating both surface water and groundwater processes. The model’s capability to couple various critical components of the urban water balance enables it to incorporate multiple interrelated processes, such as groundwater–surface water interactions, irrigation, and anthropogenic influences like urbanization, sewer leakage, and groundwater abstraction. The model can be utilized to simulate the urban water balance of Kabul, and the impact of urbanization can be analyzed using time-varying distributed data for surface roughness, surface–subsurface leakage, and paved area fraction.

Available data for Kabul indicate that the information required for simulating overland flow using the finite difference method can be obtained. Distributed parameters such as surface roughness, detention storage, surface–subsurface leakage, and paved area fraction can be estimated from land cover data and values reported in the literature. In addition, the data required for conceptualizing the river network in MIKE+ River are available, while river paths and cross-sections can also be derived from the high-resolution digital elevation model (DEM) available for the study area. Unsaturated flow can be simulated using the two-layer method because detailed soil profile data are not available. Although the data required for simulation of the saturated zone using the 3D finite difference method are limited, reasonable parameters estimation can be achieved using the results of previous studies, available data, and the geological characteristics of the study area. Groundwater abstraction can also be estimated by assuming a per capita water demand based on the population distribution derived from land cover data.

As with any model, there are limitations to consider. The most frequently cited challenges are the long computational times required for running complex simulations and the high demand for spatially distributed data. These constraints have historically restricted its application, particularly in regions with limited data availability. However, recent advancements in information technology have led to substantial improvements in computational speed and enhanced access to extensive, high-quality datasets. Given the ongoing advancements in remote sensing, big data analytics, and geoinformatics, it is reasonable to expect a gradual reduction in these limitations.