Spatiotemporal variation of water balance components in Mashhad catchment, Iran: Investigating the impact of changes in climatic data and land use

Water stands as one of the pivotal limiting factors of agricultural productivity and livestock; moreover, its availability is a critical factor influencing our welfare. Effective water management in arid and semi-arid regions is essential for meeting the diverse water needs of humans, livestock, and ecosystems. Inadequate water resource practices contribute to land degradation and environmental challenges, underscoring the importance of sound water management practices in these regions (Sharma 1998). Estimating the components of water balance becomes imperative for sustainable land and water management, assessing groundwater depletion, evaluating available water, and preventing land degradation. In semi-arid and arid areas, groundwater serves as a vital water source, playing a crucial role in societal development (UN/WWAP 2006; Holger et al. 2012; Scanlon et al. 2012; Voss et al. 2013; Awan & Ismaeel 2014). However, overexploitation of groundwater negatively impacts both its quantity and quality (Scanlon et al. 2012). Therefore, assessment of the conditions of groundwater recharge in the semi-arid and arid areas is an important challenge and task to determine the sustainability of the aquifers (Yongxin & Beekman 2003; Crosbie et al. 2010).

The recharge of groundwater is the most important factor for determining safe yield (Lerner et al. 1990; Xu 2011; Tesfamichael et al. 2013). The amounts and locations of the groundwater recharge are also crucial to determine the sustainable use of groundwater resources (Sophocleous & Devlin 2004; Devlin & Sophocleous 2005). Groundwater recharge may be estimated using different methods (e.g., Lerner et al. 1990; Simmers 1997; Kinzelbach et al. 2002; Xu 2011). Hence, water-budget methods and models can be implemented with different approaches (Xu 2011). Several models such as SWAT (Arnold et al. 2000), SVAT (Armbruster & Leibundgut 2001), and DREAM (Manfreda et al. 2005), estimate groundwater recharge based on complex hydrological processes. However, in developing countries where hourly and daily climatic data are scarce, their application is limited (Abdollahi et al. 2017). The WetSpass model (a spatially distributed hydrological model) assesses the recharge of groundwater on seasonal and annual time scales (Batelaan & De Smedt 2001, 2007). Numerous studies, including those highlighted in this section, have focused on assessing groundwater recharge and surface runoff using the WetSpass model in various countries (Abu-Saleem et al. 2010; Arefaine et al. 2012; Al-Kuisi & El-Naqa 2013; Tesfamichael et al. 2013; Aish 2014). Recently, Salem et al. (2019) demonstrated that the mean long-term spatiotemporal monthly rainfall in the Drava basin is divided as runoff (29%), evapotranspiration (27%), and recharge (44%). Ashaolu et al. (2020) indicated that 27% of the precipitation in Osun drainage basin, Nigeria, becomes recharge to the aquifer, while the remaining rainfall is lost via evapotranspiration (52%) and surface runoff (21%).

In spite of the water shortage in arid areas of Iran, it is estimated that more than 90% of the freshwater is utilized for agriculture. About 12% of the whole space of Iran (i.e., 19 million hectares) is occupied by agricultural lands. Due to the dependence of most of the country's plains on groundwater resources, especially the Mashhad plain, this area is facing an intense water crisis. The present study concentrates on the Mashhad catchment, Iran, aiming to fill knowledge gaps by meticulously analyzing the spatial and temporal variability of groundwater recharge across different land-use and soil types. The overarching scientific objectives encompass evaluating the WetSpass-M model's capability in estimating water balance component variation, discerning the relative impacts of land-use and climatic data on groundwater level, and assessing mean annual water balance components in diverse land-cover and soil texture combinations. Methodologically, GIS and RS techniques were employed to prepare data, the WetSpass-M hydrological model was applied to assess water balance components (Abdollahi et al. 2012), and non-parametric Mann–Kendall tests and Sen's slope estimator were utilized for trend detection. Finally, GIS tools facilitated the integration of water balance component maps with soil and land-use maps. In the context of Iran's arid areas, where over 90% of freshwater supports agriculture and the Mashhad plain faces a severe water crisis, understanding spatial and temporal recharge patterns becomes crucial for sustainable water management. This study contributes detailed information to inform decision-makers, addressing the pressing need for effective water resource management in data-poor regions.

The WetSpass-M model required diverse input data, including distributed land use, monthly leaf area index (LAI), soil textural type, slope map, monthly groundwater depth, and weather data (rainfall, pan-evaporation values, numeral rainy days per month, temperature, and wind speed). The model's grid maps, configured at 631 rows by 467 columns with a cell size of 250 m × 250 m, were established using a digital elevation model (DEM) derived from topography maps at a 1:50,000 scale provided by the Geological Survey and Mineral Exploration of Iran (Figure 1). A corresponding slope map was then generated based on the DEM.

AVHRR and MODIS products were used to obtain the long-term monthly LAI from 1985 to 2013 (available at: ftp://ftp.glcf.umd.edu/glcf/GLASS/LAI/AVHRR), these maps were resampled based on the required pixel size for the study area. Monthly snow cover data from 2000 to 2013 were acquired from MODIS products, while data from 1985 to 2000 were derived through regression analysis of snow, temperature, and rain using Integrated Land and Water Information (ILWIS) software. Groundwater level data (for 50 observation piezometric wells) spanning from October 1985 to September 2013 were sourced from the Regional Water Authority of the Khorasan-Razavi Province. Distributed monthly groundwater depth was determined by subtracting groundwater levels from topographical elevation data. Monthly hydro-climatological variables, such as groundwater depth, precipitation, temperature, pan evaporation, and wind speed, were converted into grid maps using spatial analysis tools. These hydro-climatological data were collected from synoptic and climatology stations (35 weather stations and one hydrometric station) operated by the Iranian Ministry of Water Resources and Meteorological Organization from October 1985 to September 2013.

The WetSpass-M model, the latest version of the WetSpass model, served as the basis for estimating water balance. This raster-based, quasi-physically distributed monthly hydrological model facilitated the estimation of interception, runoff, evapotranspiration, and recharge for each pixel. The simulation process, initiated with data reading, progressed through monthly water balance components per pixel, encompassing interception, runoff, evapotranspiration, and recharge. The model's flexibility was enhanced through its spatial computational engine, developed by Abdollahi et al. (2012, 2017) in IronPython.

indicates surface runoff (mm/month), C_sr shows the coefficient of surface runoff (implication of the portion of monthly rainfall which participates in surface runoff), C_h and I_m are a soil moisture-dependent coefficient and the monthly interception, respectively.

Researchers such as Eckhardt and Gonzales et al. have corroborated the effective performance of the WHAT system in assessing water balance evaluation and calibrating hydrological models (Eckhardt 2005; Gonzales et al. 2009). In this investigation, the WHAT system was used to partition base flow from river flow (Lim & Engel 2004), employing three separation modules: the one-parameter digital filter technique (Lyne & Hollick 1979; Nathan & McMahon 1990; Arnold & Allen 1999; Arnold et al. 2000), the local-minimum method (Lim et al. 2005), and the Eckhardt recursive digital filter (Eckhardt 2005). River flow data from the Moshang station, situated on the primary perennial stream, was used for filtration. Iranian Ministry of Water Resources collected and provided these data at daily and monthly intervals covering a period from October 1985 to September 2013. The hydrograph separation tool was applied to daily base data, and monthly base-flow and direct-runoff were computed based on the daily data average over the study period (336-time steps).

Model calibration involved iterative adjustments to input calibration parameters (a, LP, x, α, β, ∅, and MF, corresponding to interception, surface runoff, runoff delay, evapotranspiration, storage parameter, recharge contribution, and snowmelt, respectively) through trial-and-error process. Sensitivity analysis was conducted on three key parameters (LP, a, and α) as a technique to simplify the model calibration (Abdollahi et al. 2017). Calibration and validation of the WetSpass-M model were performed via the comparison of model-simulated data and filtered river flow data using the WHAT method (direct-runoff and base flow).

Analyzing monthly runoff dynamics in the Mashhad catchment (Table 3) unveils variations from 0 to 41.9 mm/month, with average and standard deviation values of 3.2 and 5 mm/month, respectively. It discovered that about 14% of the total average monthly precipitation (22 mm) in the Mashhad basin became surface runoff (3.2 mm) during the study period. This result is higher than compared to the findings of Teklebirhan et al. (2012), Al-Kuisi & El-Naqa (2013), and Mathenge et al. (2020). For instance, Teklebirhan et al. (2012) indicated that as low as 7% of precipitation became surface runoff on the basin located in Northern Ethiopia. On the other hand, this is lower than what was found by Tesfamichael et al. (2013), Salem et al. (2019), Ashaolu et al. (2020) and Zeabraha et al. (2020). For instance, in Adigrat area, Northern Ethiopia, Zeabraha et al. (2020) observed that about 16% of the annual precipitation became surface runoff. Generally, the comparison with the previous studies highlights both higher and lower percentages, emphasizing the unique hydrological characteristics of our study area.

The estimation of actual evapotranspiration in the Mashhad basin (Table 3), encompassing various land-use classes and soil types, discloses monthly variability (0–63.9 mm) and a substantial contribution (57%) to the water balance. This aligns with earlier research emphasizing the pivotal role of evapotranspiration in watershed water loss (e.g., Tesfamichael et al. 2013,Ashaolu et al. 2020; Gebru & Tesfahunegn 2020; Mathenge et al. 2020; Zeabraha et al. 2020). Monthly groundwater recharge, spanning October 1985–September 2013, exhibits variability (0–38 mm), with an average of 6.5 mm (Table 3). Significantly, 29% of the average monthly rainfall contributes to recharge, a proportion exceeding comparable dry regions (e.g., Adelana et al. 2006; Teklebirhan et al. 2012; Al-Kuisi & El-Naqa 2013; Babama'aji 2013; Tesfamichael et al. 2013; Zeabraha et al. 2020) but falling below the values reported by Salem et al. (2019).

The average annual evapotranspiration for 28 years was calculated by adding monthly evapotranspiration maps (Figure 8(b)) and is higher in regions with higher vegetation cover and agricultural land. Table 6 reveals the variability of evapotranspiration among the various combinations of soil and land use. Silty loam and silty clay loam soils exhibit the highest evapotranspiration, while the lowest value was obtained for clay soil. The irrigation farming, rain-fed farming, and orchard land uses display the greatest amounts of evapotranspiration, meanwhile, the lowest values are revealed in rangeland and residential land-use classes. Tesfamichael et al. (2013) reported that evapotranspiration in their case study was more affected by land-use class than soil texture and the same result was found in our study area which shows that evapotranspiration is more changeable within land cover than soil texture (Table 6).

The annual recharge maps were created based on monthly simulated recharge maps (Figure 8(c)). The lowest values of groundwater recharge were observed in the flat parts of the catchment, probably due to a mixture of desirable circumstances such as more surface runoff and more evapotranspiration and more vegetation cover. Soil textural types and land-cover classes had a significant impact on groundwater recharge. The average annual groundwater recharge amounts for various mixtures of soil and land-use categories are indicated in Table 7. It seems that loamy and sandy clay soils with rangeland have the highest amounts of groundwater recharge, basically due to the high permeability of these soils. Agricultural lands with the soil classes of silty loam, silty clay, and sandy clay loam soils have the lowest values, which is clearly due to the high transpiration, evaporation losses, high temperature, and low precipitation in these flat regions. The higher standard deviation amounts of the recharge for various soil textures (Table 7) show that soil texture has more impact on recharge than land-use classes. This underscores the predominant impact of soil characteristics on recharge, consistent with what was reported by Babama'aji (2013), while is inconsistent with the findings of Ashaolu et al. (2020) who indicated that recharge in Nigeria was more affected by land-use type.

In this study, we employed the WetSpass-M model, a raster-based quasi-physically distributed monthly hydrological model, to comprehensively assess the monthly water balance components in the Mashhad catchment. Utilizing a combination of the WetSpass-M model, WHAT (WHAT), and non-parametric trend analysis tests, we evaluated various hydrological aspects and identified trends in the data. Spatial analysis techniques were integral in preparing the extensive input data required for the model. Our results highlight that the average monthly evapotranspiration, constituting 57% (12.7 mm) of the monthly precipitation (22 mm), is primarily influenced by precipitation and vegetation cover. Notably, long-term average monthly evapotranspiration increased with rising monthly precipitation, with rain-fed and irrigation farming lands, along with silty loam soils, exhibiting the highest evapotranspiration. Examining the monthly runoff, we observed variations ranging from 0 to 41.9 mm, with an average of 3.2 mm, representing 14% of the monthly precipitation. The highest surface runoff occurred in irrigation farming areas on silty loam and clay soils. Monthly recharge of the Mashhad catchment ranged from 0 to 38 mm, averaging 6.5 mm per month, constituting 29% of the monthly precipitation, with the highest amount occurring in sandy clay soils with rangeland.

Spatially, the largest surface runoff amounts were observed in the southwest and central regions of the catchment, correlating with higher rainfall. Additionally, areas with low elevation exhibited higher evapotranspiration and lower groundwater recharge compared to the high-altitude regions. Combining simulated maps using the WetSpass-M model with soil and land-use maps revealed that evapotranspiration varies more within land-use classes compared to soil texture types, while soil texture had a greater impact on recharge and surface runoff compared to land use in the Mashhad Basin. Only groundwater depth showed a statistically significant declining trend post the 1990s, while precipitation and simulated water components trends were not statistically significant. The increase in irrigation farming area, residential area, and orchard land cover from 1986 to 2013 suggests a potential influence on the declining groundwater levels in the Mashhad mega city. The significant decline in groundwater level at a rate of 0.66 mm per year, coupled with changes in land cover, suggests that human activities are leading to over extraction.

In summary, the integrated approach of the WetSpass-M model, coupled with comprehensive spatial analysis and trend assessments, provides valuable insights into the intricate hydrological processes of the Mashhad catchment. Our study contributes to a deeper understanding of water balance dynamics, aiding in the formulation of informed water resource management strategies amidst evolving land-use patterns and climate variations. Our study has provided valuable insights into the recharge dynamics of the study area. The observed trend changes in recharge are subject to certain limiting conditions, including the spatiotemporal variability of precipitation, the hydrogeological characteristics of the study area, and the impact of human activities on groundwater resources. Therefore, the results may not be applicable to all regions or under all circumstances. Nonetheless, our findings highlight the critical importance of sustainable water management practices to mitigate the impact of human activities on groundwater resources. Further research is needed to better understand the complex interactions between climate, land use, and groundwater resources and to develop effective strategies for sustainable water management in arid and semi-arid regions.

The authors appreciate the Iranian Meteorological Organization (IRIMO) and Iran Water Resources Company for providing the observational daily hydro-meteorological and hydrogeological data of the study area. We also are grateful to the Department of Civil, Environmental, Architectural Engineering and Mathematics, University of Brescia, Italy for providing scientific support for this research.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.