Introduction

Background and justification

Water resource, which is the backbone and crucial element of life, is needed in sufficient quantity and quality to meet the increasing demand for domestic, agricultural and industrial processing operations (Fenta et al. 2014; Shanableh and Merabtene 2015; Arefayne et al. 2015). But, its availability is limited because of its natural distribution on the earth surface, in which 97.5% of the global water is saline existing in the oceans and only 2.5% is considered to be available for biological use, i.e., fresh water. About 68.7% of the remaining 2.5% that is fresh water is locked up in glaciers while 30.1% and 0.9% represent groundwater, surface water, and other fresh waters, respectively (Shiklomanov 1998). It is a scarce, crucial and multifunctional natural resource found on the planet and the demand for fresh water is increasing worldwide as a result of urbanization, economic and population growth (Karimi and Bastiaanssen 2015). Because of its scarcity, proper planning and management of such resource in terms of distribution, management, utilization, and environmental functions are crucial for optimizing the resource use sustainably (Karimi and Bastiaanssen 2015).

Access to safe drinking water becomes a problem in many regions of the world particularly in dryland of developing countries because of high growing and competing demand of the resource. Urbanization and population growth in urban areas are not only stressing the capacity and sustainability of the existing water supplies but also it placing the supply at a greater risk of contamination. Moreover, humans are also affecting the global water resources through climate change by increasing atmospheric concentrations of greenhouse gases and aerosols, changes in land use land cover due to deforestation, industrialization, and urbanization will continue to affect the earth’s climate by changing the temporal and spatial distribution of rainfall, temperature, evapotranspiration in particular and the whole hydrologic cycle in general (Sykes 2006).

Mostly, water resource is found as surface water and groundwater forms, surface water is not a reliable source as it is prone to seasonal fluctuations and susceptible to contamination through anthropogenic activities such as point, non-point pollution sources and biological pollutions (Fenta et al. 2014). However, groundwater is more suitable in quantity, readily available and is naturally protected from direct contamination by surface anthropogenic activities (Fenta et al. 2014). It is the largest reservoir of liquid freshwater on the planet and is critical for sustaining life on earth, as it is used to satisfy different human and environmental needs (Zomlot et al. 2015a).

Ethiopia is also endowed with a substantial amount of water resources. The country is divided into 12 river basins; 9 of which are wet river basins; 1 lake basin; and remaining 3 are dry river basins, with no or insignificant flow out of the drainage system. Almost all of the basins radiate from the central plateau of the country that separates into two due to the Rift Valley. Basins drained by rivers originating from the mountains west of the Rift Valley flow toward the west into the Nile River basin system, and those originating from the Eastern Highlands flow toward the east into the Republic of Somalia. Rivers draining in the Rift Valley originate from the adjoining highlands and flow north and south of the uplift in the center of the Ethiopian Rift Valley (Melesse et al. 2013).

There is surface water resource in Birki watershed and the community uses for different domestic and agricultural activities but there is no any information on the amount of groundwater recharge on the specific watershed for wise and sustainable utilization and proper management of the limited water resources. Thus, estimation of groundwater recharge in that specific watershed has its own role in solving the problems related to the management and planning of water resources for sustainable development using hydrological models such as WetSpass model with the help of GIS and remote -sensing techniques.

WetSpass model is a spatially distributed simulation model for water and energy transfer between soil, plants, and atmosphere under quasi-steady state. It predicts spatial patterns of surface runoff, evapotranspiration, and groundwater recharge on a regional scale (Batelaan and De Smedt 2007). The model understands a region as a regular pattern of raster cells. Every raster cell is further sub-divided into a vegetated, bare soil, open water, and impervious materials and seasonal water balance components are calculated for each grid cell.

The WetSpass model was applied in different authors, according to those authors, this model was implemented for groundwater recharge estimation and it simulates with a good and accepted result, which is the main objective of this research work. Therefore, this research study aimed to implement the newly emerged tools of geographic information system/science (GIS)-based WetSpass model for estimation of groundwater recharge for Birki watershed (45 km2), Northern Ethiopia.

Materials and methods

Description of study area

Geographical location of Birki watershed

The study area, which is Birki watershed, is located in the Geba River Basin of Eastern Regional State of Tigray, Northern Ethiopia (Fig. 1). The watershed is found within two districts, i.e., Kilte awelaelo and Atsebi-Wenberta. Geographically, it is located in latitudes of 13.65°–13.75° Northing and longitudes of 39.60°–39.71° Easting with elevation ranging from 1999 to 2514 m above mean sea level with an area of 45 km2 (own processing). The watershed receives mean annual rainfall of 573 mm. The Birki River flows from the eastern escarpment of the Eastern Zone of Tigray to the west contributing the flow to Geba River a tributary of Tekeze River. It is a perennial river but flows are extremely low in the dry season and high floods during the wet season (July to September).

Fig. 1
figure 1

(Source: Own processing)

Map of the area under investigation. Birki watershed: Northern Ethiopia.

Full size image

Research methodology

The WetSpass model was used as a methodology to simulate temporal average and spatial differences of surface runoff, actual evapotranspiration, and groundwater recharge in seasonal and annual basis for Birki watershed. Generally, to run the WetSpass model, two basic input parameters, which are hydro-meteorological and biophysical data related to the watershed are needed and the necessary data should be prepared in the required grid and database file (DBF) formats.

Basic concepts of WetSpass model

WetSpass is an acronym which stands for water and energy transfer between soil, plants, and atmosphere under quasi-steady state (Batelaan and DeSmedt 2001; Batelaan and De Smedt 2007). It uses both physical and hydro-meteorological input files for simulation of the long-term average spatial patterns of surface runoff, actual evapotranspiration and groundwater recharge which is suitable for studying long-term effects of land-use changes on the water regime in a watershed region (Batelaan and DeSmedt 2001, 2007; Aish et al. 2010). The application of this model is compatible and integrated with GIS ArcView software during the simulation process. It estimates the spatial difference of groundwater recharge at the seasonal and annual basis and it was successfully applied in different countries by different authors, as a result, the findings of those authors showed that groundwater recharge estimation was successfully estimated with a good result (Al Kuisi and El-Naqa 2013).

Model description

The total water balance for a given raster cell (Fig. 2) is split into independent water balance components for the vegetated, bare-soil, open-water and impervious parts of each cell. This allows one to account for the non-uniformity of the land-use per cell, which is dependent on the resolution of the raster cell. The processes in each part of a cell were set in a cascading way. This means that an order of occurrence of the processes, after the precipitation event, is assumed. Defining such an order is a prerequisite for the seasonal timescale with which the processes will be quantified. The quantity determined for each process is consequently limited by a number of physical and hydro-meteorological constraints of the given area under investigation (Batelaan and DeSmedt 2001).

Fig. 2
figure 2

Graphical iterative representation of water balance calculation of a hypothetical raster cell showing surface and sub-surface processes, after Batelaan and De Smedt (2001)

Full size image

Water balance calculation using WetSpass model

Water balance components of vegetated, bare-soil, open-water and impervious surfaces are used to calculate the total water balance of a raster cell using the Eqs. (1)–(3).

$${\text{E}}{{\text{T}}_{{\text{raster}}}}={a_{\text{v}}}{\text{E}}{{\text{T}}_{\text{v}}}+{\text{ }}{a_{\text{s}}}{E_{\text{s}}}{\text{ }}+{\text{ }}{a_{\text{o}}}E{}_{{\text{o}}}{\text{ }}+{\text{ }}{a_{\text{i}}}{E_{\text{i}}}$$
(1)
$${S_{{\text{raster}}}}={a_{\text{v}}}{S_{\text{v}}}+{\text{ }}{a_{\text{s}}}{S_{\text{s}}}{\text{ }}+{a_{\text{o}}}{S_{\text{o}}}+{\text{ }}{a_{\text{i}}}{S_{\text{i}}}$$
(2)
$${R_{{\text{raster}}}}={\text{ }}{a_{\text{v}}}{R_{\text{v}}}{\text{ }}+{\text{ }}{a_{\text{s}}}{R_{\text{s}}}{\text{ }}+{\text{ }}{a_{\text{o}}}{R_{\text{o}}}{\text{ }}+{\text{ }}{a_{\text{i}}}{R_{\text{i}}}$$
(3)

where ETraster, Sraster, Rraster are the total evapotranspiration, surface runoff, and groundwater recharge of a raster cell, respectively, each having a vegetated, bare-soil, open-water and impervious area component denoted by av, as, ao, and ai, respectively. Precipitation is taken as the starting point for the computation of the water balance of each of the above-mentioned components of a raster cell, the rest of the processes, such as interception, surface runoff, evapotranspiration, and recharge follow in an orderly manner.

WetSpass model data inputs preparation and their sources

GIS-based hydrological models such as WetSpass model was used for analyzing groundwater systems in steady-state condition and needs long-term average hydro-meteorological data and spatial patterns of watershed-based biophysical layers as the main inputs. WetSpass needs the parameters in seasonal basis, as a result, four months of June, July, August, and September are considered as summer (main rainy season) and the remaining 8 months are considered as winter (dry season) in the case of Ethiopian condition particularly at the study area. Grid maps and parameter tables are required as inputs for the model and were prepared with the help of ArcGIS tools and Erdas Imagine software. These grid maps were a land-use land cover, soil texture, slope, topography and groundwater levels, precipitation, potential evapotranspiration and wind speed. The input files prepared as parameter tables were also prepared in a database file format (dbf); these are summer and winter land use land cover, soil texture and runoff coefficient. All the inputs and their sources were mentioned as follows in Table 1.

Table 1 Input parameters and sources for WetSpass model
Full size table
Table 2 Long-term average seasonal hydro-meteorological data of nearby principal stations [precipitation (pre), temperature (temp), wind speed]
Full size table
Table 3 Long-term average seasonal hydro-meteorological data of nearby principal stations (potential evapotranspiration winter & summer, PET_Win, Sum), mean precipitation (mean_ppt), average temperature and annual average PET
Full size table

Biophysical input data and hydro-meteorological grid maps preparation

Land use land cover map of the watershed was prepared using a supervised image classification method with an overall accuracy of 82% and Kappa statistics of 0.78 using Erdas Imagine. The interpolation technique used here is an inverse distance weighted (IDW) method and it is used for estimating missing values between known measurements. Hydro-meteorological and soil textural map of the watershed was also prepared using inverse distance-weighted (IDW) interpolation technique. Around 65 soil samples were taken from the watershed with soil depth of 0–15 cm and with a distance of 100 m among the different land use types in the dry season and taken in to the laboratory to identify their textural classes, meteorological data were collected from the National Meteorological Agency(NMA), PET was calculated using Cropwat-8 software and groundwater depth were collected from the regional water resources office and it was prepared by subtracting static water level from the surface elevation of the boreholes and springs found within the watershed. All these parameters were prepared for the year 2015 as spatial data. The reason for using this interpolation method is that it is fast and easy to implement for a specific purpose with few and scattered measured data, each value estimated in an IDW interpolation is a weighted average of the surrounding sample points. Weights are computed by taking the inverse of the distance from an observer’s location to the location of the point being estimated. Long-term hydro-meteorological data of the study area was prepared as follows:

Using the above long-term annual average hydro-meteorological data of the nearby principal stations, grid maps of Birki watershed was prepared using the IDW interpolation method in ArcGIS environment. And also the required physical data should have grid file format to read by the model for effective simulation, so, the inputs were prepared using Erdas imagine and ArcGIS environment. The parameters were mentioned in Tables 2, 3.

Parameter tables preparation/lookup tables

Look up tables are also important for running the WetSpass model, so that, four parameter tables were prepared, they are summer and winter land use land cover, soil texture and runoff coefficient parameters in DBF (database file) format. Basically, the model user guide and some other literature reviews were used to adjust and develop the parameter values to the watershed characteristics. In this section, excel (xls) file to dbf file format converter software was used to prepare the lookup tables and these parameter tables are mentioned in (Tables 4, 5, 6).

Table 4 Look up parameters for winter land use land cover
Full size table
Table 5 Look up parameters for summer land use land cover
Full size table
Table 6 Look up parameters for soil texture
Full size table

Materials and softwares used

The materials and software’s used in the entire research work were discussed as follows (Table 7).

Table 7 Materials and software are used in this research
Full size table

The overall research methodology framework adopted for estimation of groundwater recharge using GIS-based WetSpass model for Birki watershed was mentioned as illustrated in (Fig. 3).

Fig. 3
figure 3

Research methodology frameworks and different activities and tools used to achieve research objectives

Full size image

Research results

Groundwater recharge estimation using WetSpass model

The WetSpass model was implemented for Birki watershed to estimate long-term seasonal and annual groundwater recharge using 10 years (2006–2015) hydro-meteorological data of surrounding principal stations of Atsebi, Senkata, Wukro and Mekelle stations with the main objective of estimating groundwater recharge for Birki watershed. The results of our simulation using such a long-term biophysical and hydro-meteorological input indicated that groundwater recharge amount varies among seasons. The recharge value during the main rainy season in June to September ranges from 0 to 41.09 mm/year with a mean value of 24.1 mm/year (Fig. 4a), while the recharge during long dry season ranges from 0 to 1.53 mm/year with mean value of 0.82 mm/year (Fig. 4b), and mean annual groundwater recharge ranges from 0 to 42.6 mm/year with mean value of 24.9 mm/year, which accounts 7.4% of the total long-term mean annual precipitation 573 mm on the entire watershed as shown in (Fig. 4c).

Fig. 4
figure 4

Shows WetSpass model-based groundwater recharge simulation maps of the Birki watershed during summer season (a), winter season (b) and annual (c) using biophysical and hydro-meteorological characteristics of the watershed

Full size image

According to the mean annual groundwater recharge values simulated by the model and as illustrated in (Fig. 4c) there are high groundwater recharge values in northwestern parts of the Birki watershed but the lowest groundwater recharge was observed in the southeastern escarpments of the watershed. This variation is due to the biophysical and climatic variations of the watershed and to say some relationships, sandy soils have high groundwater rechargeability than silt and clay soil textures in combination with shrubland use land cover classes but water bodies have the lowest groundwater recharge in the watershed. Combination between the biophysical components and the groundwater recharge values simulated shows that, areas which have gentle slope type, lowest elevation value, with shrubland class and sandy textural classes have high groundwater recharge values whereas hilly slope areas, mountainous areas with bare land class and clay soil textural classes have lowest groundwater recharge in the watershed. This difference depends on the climatic, geological, biophysical and topographic characteristics of the watershed.

Groundwater recharge potential of Birki watershed

Understanding groundwater potential of a given watershed was important for wise utilization and proper management of the water resource and for developmental water resources programs such as for designing hand dung wells and springs. Based on the yearly groundwater recharge simulated by WetSpass model, Birki watershed has three (high, medium and low) levels of groundwater recharge potential. This groundwater recharge potential was developed based on natural break classification method in ArcGIS environment.

There is high groundwater recharge potential in the northwestern of the watershed with 33% of the total area coverage of the watershed and medium recharge potential was found around the central and northeastern parts (Fig. 5). In other words, the southeastern, as well as some parts of central, have low groundwater potential. The detail areal coverage of the recharge potential was depicted in (Fig. 6). Based on the groundwater recharge potentiality classification, as a result, the watershed has three classes, these are high (33%), medium (38.1%), and low (29.2%) groundwater potentiality classes in the entire Birki watershed.

Fig. 5
figure 5

Groundwater recharge potentiality map for Birki watershed

Full size image
Fig. 6
figure 6

Areal coverage distribution of groundwater recharge potential in percent

Full size image

Groundwater safe yield for Birki watershed

Groundwater used to support crop growth by irrigation can be assessed by means of a safe yield map, which indicates how much groundwater can be abstracted in a sustainable way without depleting the groundwater resources. Safe yield is usually expressed as a percentage of the groundwater recharge. Several authors suggested different percentages, from the least conservative 100% to a reasonably conservative 10% (Gebreyohannes et al. 2013). Sustainable yield must be considerably less than the groundwater recharge to sustain both the quantity and quality of streams, springs, wetlands and groundwater-dependent ecosystems (Sophocleous 2000). Hence, in this study, a reasonably conservative estimate of the sustainable yield of 25% of the groundwater recharge value was adopted from the following formula\(, \text{i}.\text{e}., \left(\right)\)(Gebreyohannes et al. 2013). Where SY is the safe yield groundwater abstraction rate expressed in m3/d/ha and R is the total annual groundwater recharge expressed in (mm). The resulting safe yield map of the Birki watershed is presented in (Fig. 7). The value ranges from 0 to 0.29 m3/d/ha, with mean value of 0.17 m3/d/ha. Annually, 0.17 m3/d/ha groundwater can be extracted safely from the watershed. The highest groundwater safe yield values were observed in the northwestern parts of the watershed but the lowest values were also found in the southeastern parts of the watershed (own processing).

Fig. 7
figure 7

Groundwater safe yield map of Birki watershed

Full size image

Discussions

This study aimed at estimating long-term seasonal and annual average groundwater recharge value, to map out groundwater recharge potential and groundwater safe yield map estimated by WetSpass model for Birki watershed, located in northern Ethiopia. The result showed that mean annual average groundwater recharge of the Birki watershed was 24.9 mm/year. The total mean annual rainfall of the watershed (573 mm) contributes to 7.4% as recharge to groundwater, 7.1% of surface runoff and 85.5% of evapotranspiration. About 96.5% of the groundwater recharge in the watershed was happened in summer season, i.e., main rainy season and 3.5% in the winter season (dry season) because of precipitation, temperature, potential evapotranspiration, and soil moisture variations. Moreover, there is high duration, intensity and amount of precipitation distribution, high soil moisture and evapotranspiration in the summer season which accelerated groundwater recharge.

There is high groundwater recharge and evapotranspiration than surface runoff in the watershed because the watershed is a conserved type of watershed and there is high shrub land coverage, in which this facilitates soil infiltration rate and evapotranspiration by reducing surface runoff in the watershed. Recent studies showed that the WetSpass model was implemented in different regions by different authors and their findings showed that their simulation results were acceptable, these simulation results were stated as follows as cited by Zarei et al. (2016).

Abu-Saleem et al. (2010) evaluate the water balance components using WetSpass model for the Hasa basin in Jordan. According to their results, mean annual groundwater recharge and surface runoffs were 0.98 and 23.64 mm/year, respectively. In other words, about 0.64% and 15.4% of the annual precipitation convert to groundwater recharge and surface runoff, respectively, and the major part of the precipitation (83.96%) is lost as evapotranspiration.

In Northern Ethiopia, Arefaine et al. (2012) simulate the water balance components including groundwater recharge, evapotranspiration and surface runoff using WetSpass model. Results show that the mean annual groundwater recharge, evapotranspiration, and surface runoff were found to be 66, 440 and 40 mm, respectively. Therefore, 12% of the precipitation becomes recharge while evapotranspiration and surface runoff are 81% and 7% of the precipitation, respectively, compared to our simulation results surface runoff and evapotranspiration values have some similarity than the recharge value, this may be due to different biophysical factors within the area.

According to the above results, WetSpass model simulates the water balance components in Birki watershed which is 7.4% as recharge to groundwater, 7.1% surface runoff, and 85.5% evapotranspiration have some similarity and difference with the results shown above, this difference can be observed because of variation in hydro-meteorological components, geological formations, climatic variability, physical properties of soils, watershed area, input data and their spatial resolution and conservation practices with a given watershed, such as Birki watershed is a highly conserved watershed that is way the runoff is so small. The groundwater recharge potential was highest in the northwestern parts of the watershed with characteristics of the flat slope, sandy soil, and dominated by shrubland classes and there is low groundwater recharge potential in the southeastern part which has clay soil type, hilly slope and covered by bare land. Understanding the groundwater recharge potential areas of a given watershed is important for designing and developments of hand dung wells as sources of potable water used for different domestic and agricultural activities, based on the groundwater recharge map of the watershed, the potential map was created using natural break classification method into three levels, high, medium and low recharge values. As a result, there is high groundwater potential in the northwestern part but low recharge potentials in the southeastern escarpment of the watershed.

Conclusions

Water resource is a crucial element of life, so as to maximize its use proper planning, management, and wise utilization is crucial in the twenty-first century. To do so, the development and application of GIS and remote-sensing techniques make the assessment and modeling of water resources easy and effective for such a purpose. The WetSpass model which is a simulation model for groundwater recharge is based on biophysical and hydro-meteorological properties and it is important to estimate long-term annual average groundwater recharge in seasonal and annual basis for wise utilization, proper management and future planning of water resources.

The long-term annual and seasonal groundwater recharge for Birki watershed is 42.6 mm/year, summer is (41.09 mm), winter is (0.82 mm) and about 96.5% of the recharge happened during the summer season and remaining 3.5% in the winter season. This differences in recharge value are attributed to differences in rainfall amount that leads to the differences in the biophysical properties of the catchment and mean annual precipitation (573 mm) contribute to 7.4% as groundwater recharge, 7.1% of surface runoff and the remaining 85.5% is lost as evapotranspiration in the Birki watershed.

There are high groundwater recharge value and evapotranspiration with low runoff in the watershed, because the watershed is the conserved type with high coverage of shrublands, in which this facilitates infiltration rate and evapotranspiration, but decreases runoff production. There is high groundwater recharge potential in the northwestern parts of the watershed and low groundwater recharge potential in the southeastern escarpments of the watershed.

Groundwater safe yield ranges from 0 to 0.29 m3/d/ha, northwestern parts have high values of groundwater safe yield than the other parts of the watershed and the lowest values are in the southeastern part and annually, in average 0.17 m3/d/ha groundwater can be extracted safely without diminishing the groundwater resources and annually, 1.1205 million m3 water recharging to the groundwater from the total annual precipitation in Birki watershed.