Afraz Mehdi, Eftekhari Mobin, Akbari Mohammad, Ali Haji Elyasi, Noghani Zahra

1 Forests, Range and Watershed Management Organization, Tehran, Iran.

2 Young Researchers and Elite Club, Mashhad Branch, Islamic Azad University, Mashhad, Iran.

3 Department of Civil Engineering, University of Birjand, Birjand, Iran.

4 K. N. Toosi University of Technology, Tehran, Iran.

5 Master Research Institute of Forest and Rangeland, Tehran, Iran.

Abstract: In recent years, drought has become a global issue, especially in arid and semi-arid areas. It is without doubt that the identification and monitoring of the drought phenomenon can help to reduce the damages that would occur. In addition, rain is one of the factors which directly affect the water levels of underground water reservoirs. This research applied a linear gradient regression method developed on the basis of GRACE, CHIRPS, and data from monitoring wells to investigate the groundwater storage changes.These data have been analyzed on the Google Earth Engine platform. In order to conduct temporal and spatial analyses, the water levels of the aquifer were generated from the monitoring wells and zoned into five classes. Also, the amount of water storage and rain from the year 2003 to 2017 in the West Azerbaijan Province were investigated using the GRACE satellite and the CHIRPS data, respectively. The results obtained from the GRACE satellite data show that the average water level in the underground reservoirs in Iran had started to decrease since 2008 and reached its peak in 2016 with an average decrease of 16 cm in that year. The average annual decline of groundwater level in the studied time period was 5 cm. A chart developed from the CHIRPS annual rainfall data indicates that the biggest decline in rainfall occurred in 2008, and the declining trend has remained steady. Linear analyses were made on GRACE with CHIRPS results and monitoring wells data separately, from which the correlation coefficients are between 86% and 97%, showing generally high correlations. Furthermore, the results obtained from the zoning of the aquifer showed that in the period of 2004 to 2016, due to the decrease in rainfall and the excessive withdrawal of groundwater, the water levels also decreased.

Keywords: GRACE; Groundwater storage; CHIRPS; Middle East region Received: 16 Sep 2020/ Accepted: 20 Apr 2021

Introduction

The Gravity Recovery and Climate Experiment(GRACE) satellite was the first satellite tasked with the measurement of humidity and water reservoirs in underground layers in different situations (Agboma et al. 2009). This satellite has provided a new and unique way of using remote sensing to measure changes in water storages(Such as ice, snow, surface water supplies, soil humidity, and underground reservoirs). It also provides a new source of information for the science of hydrology in the international community (Strassberg et al. 2007). The most significant benefit of using the GRACE satellite data is that the amount of water storage in different parts of the world can be spatially quantified.Quantification of the changes in total water supplies is essential in hydrological studies, and these investigations can be valuable when conducting studies on drought (Wang et al. 2011).Water in a large watershed is a key component in the hydrological cycle and has a significant effect on the local climate system (Lo et al. 2010).Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) was used to monitor and evaluate the trend of drought in a research conducted by Hosseini-Moghari et al. (2018).

Many researches have been conducted using Satellite Gravimetry and the following are some of them. Voss et al. (2013) studied the depletion of groundwater reservoirs in the Middle East using GRACE data. The results indicated that the depletion rate of water storage in this region was 27.2 mm in water level and 143.6 cubic kilometers in volume per year. Joodaki (2014) investigated on the role of human activities in depleting groundwater in the Middle East. The results indicated a significant decrease in water storage in the west of Iran and the north of Iraq. Forootan et al. (2014)conducted a study on the separation of water storage patterns in the entire Iranian region,showing an average mass decrease of 15 mm per year in most Iranian areas since 2005, especially in the central and northwestern parts. Eftekhari et al.(2017) studied the changes of the Caspian Sea level, found that water levels in the southern region of the Caspian Sea fluctuated within a span of 26 cm to 42.5 cm, and these changes were concluded to be related to climate change and the construction of dams across rivers. Saiedizand (2019) evaluated rainfall predictions provided by the CHIRPS station between 2005 and 2014. The assessment showed that the estimation was acceptable as the correlation coefficients were significant -99%.Eftekhari et al. (2020) analyzed the fluctuations of the Birjand plain aquifer. The GIS results, the GRACE spatial analyses and satellite imagery showed that between the year 2008 and 2018, the water table declined significantly.

Due to the decrease in rainfall in West Azerbaijan Province and the existence of the Urmia Lake, it is of utmost importance to investigate the reasons behind drought. Therefore, the aim of this study was to investigate the changes in water storage in the West Azerbaijan Province in Iran using GRACE data and the annual rainfall statistics from CHIRPS.

1 Study area

According to the 2016 census, in an area of 42 252 km2, the population of West Azerbaijan Province is 3 265 219 km2, which is 4.08% of the total population of the country and the province ranks the eighth most populous province in the country.This province is located between the latitudes of 35°58'N to 39°46'N and the longitudes of 44°30'E to 47°23'E. West Azerbaijan is surrounded by several countries and provinces, such as the Nakhchivan Autonomous Republic and Armenia in the north, Turkey and Iraq in the west, the Iranian Kurdestan Province in the south, and the Iranian East Azerbaijan and Zanjan provinces in the east.During the wet season, it is affected by the humid air currents of the Atlantic Ocean and the Mediterranean Sea (western currents) with the occurrence of highest amount of rainfall. In winter,cold Siberian air masses infiltrate the region from Russia and reduce the temperature drastically.Sometimes severe frosts prevail for 4 months to 5 months, and precipitation is snowy at high altitudes. As shown in Fig. 1, this region has developed as a plain near the western margin of the Urmia Lake. Due to the fact that this province is surrounded by a mountainous environment, Lake Urmia has been identified as one of the six main catchments of Iran. In addition, the average annual rainfall of the province is 398 mm, which is 220 mm in the north and around Poldasht and increases to more than 450 mm along the western borders. In this province, parts of three watersheds of the country, namely Aras river basin, Urmia lake basin and Zab river basin are rich in water resources.Compared to other provinces, which have roughly 280 mm of rainfall per year, West Azerbaijan is considered to be in a better condition in this regard.

Fig. 1 Geographical coordinates of the West Azerbaijan Province

2 Data and methods

2.1 Data from the GRACE satellite

In 2002, the GRACE satellite was sent into orbit with a five-year mission. This remote sensor satellite was the first satellite tasked with the measurement of humidity and water reservoirs in underground layers in different situations (Agboma et al. 2009). This satellite monitored the changes in the Earth’ s mass in a time series manner by orbiting the planet. By combining several years of time series data we can produce figures of the process of changes in the Earth’ s mass. The monthly changes in the gravitational field of the Earth provide information on both the fixed quantity of mass distribution and the total volume of water storage in any specified watershed (Rodell and Famiglietti, 2002).

In the present study, data regarding water storage (underground reservoirs, soil humidity, and surface water) in the West Azerbaijan Province,between June 2003 and December 2017, were obtained from the GRACE satellite in the unit of centimeters. The data was developed in Google Earth Engine (GEE) using algorithms from three research institutes, i.e. the Center for Space Research at University of Texas, Austin (CSR), GeoForschungs Zentrum Potsdam (GFZ) and Jet Propulsion Laboratory (JPL). These institutes processed the data individually, providing idiosyncratic results,which were used to estimate the maximum and minimum amounts of water storage in the region(Banerjee and Kumar, 2018; Landerer and Swenson, 2012; Swenson and Wahr, 2002). Table 1 shows the results provided by CSR, GFZ, and JPL in Google Earth Engine.

Table 1 Data produced by CSR, GFZ and JPL

In order to calculate the total water storage of the region, specific data was obtained from the GRACE satellite. The CSR, GFZ, and JPL research institutes have created the Science Data System (SDS) for the GRACE project. The main function of the SDS is the processing of raw data from GRACE satellite. The raw data is referred to as level zero (L-0) data. During the next stages, the SDS provides two products which are referred to as level one (L-1) and level two (L-2) data. The L-1 data provides information on the distance between twin satellites, accelerometers, GPS, and other similar types of information. The L-2 data,based on adapted and calibrated L-1 data, provides average and monthly information on the gravitational field. All L-2 data are stored in the GFZ and JPL archives and are accessible 61 days after data collection (Swenson and Wahr, 2002).

2.2 Data from the CHIRPS station

In the year 1999, the CHIRPS station was jointly founded by the United States Geological Survey(USGS) and the meteorological hazards group of the University of California. This station has provided daily and monthly data with a spatial resolution of 0.05-degree, covering the whole world from 50°N to 50°S since 1981. Furthermore,a second such station was founded on the 12th of February in 2015. This station uses spatial regression models based on Cold Cloud Duration(CCD) and receives the CCD time-series data from the Climate Prediction Center (CPC) and the National Oceanic and Atmospheric Administration(NOAA) stations. The motives behind creating the CHIRPS station were to monitor agricultural drought, to estimate fluctuations in rainfall, to track changes in the environment, and to develop rainfall maps (Funk et al. 2015).

A rustle10 at the front door told me Holly had arrived home from school. Susan invited me to come over to her house after school tomorrow, she announced as she plunked her books down on the kitchen table. Although her voice carried a so-what attitude, I sensed she was pleased by the invitation.

In addition to the data obtained from the GRACE satellite, data provided by the CHIRPS station was utilized in order to calculate the total annual rainfall from 2003 to 2017. In order to analyze the rate of rainfall and to measure seasonal drought, data with resolution of 0.05-degree were obtained from the CHIRPS station by developing networked time series (Funk et al. 2015). Table 2 shows the specifications of data provided by the CHIRPS station.In the current study, in order to validate the GRACE satellite data, 199 monitoring wells from Regional Water Company in the West Azerbaijan Province were selected to conduct water depth zoning between 2003 and 2017 (Fig. 2).

Table 2 Data produced by CHIRPS

Fig. 2 Spatial distribution of the selected monitoring wells in the West Azerbaijan Province

2.3 Methodology

In the current study, GEE was used to process the GRACE data and the CHIPRS database. The advantage of using GEE is that the data are already in the system, and there is no need to upload anything. In addition, there is no need to conduct geometric and radiometric correction on the data.Furthermore, if there is extensive data to be processed, one does not need a powerful processor as the Engine will recall and process the data in a matter of minutes.

In order to analyze the GRACE data and to extract the data for the study region, Equation (1)was used to show changes in mass in the form of water layer thickness. Spherical harmonic coefficients, which are expanded to changes in mass,must first be stated according to the GRACE monthly models. Changes in mass according to the equivalent water layer thickness can be stated as follows (Wahr et al. 1998):

Where: ρave=5517Kg/m3istheaverage surface density ofthe Earth,knrepresentsLove numbers,ΔJnmand ΔKnmarethemonthly variations of sphericalharmoniccoefficients,represents normalizedLegendrefunctions,θis therelative deflection of the vertical axis, Δσ represents the change of surface density,ais the outer radius of the sphere, λ is the average gravity andn,mare degree and order. The level 2 GRACE data were provided monthly in the form of geopotential models for the Earth’s gravitational field. One of the prominent issues in these data is the existence of noise in the harmonic coefficients which increases along with the increase in degree and arrangement. Because of this issue, filters are used to decrease the degree and arrangement coefficients.

3 Results and discussion

3.1 Relationship between variation of groundwater level and precipitation

The GRACE data were used to evaluate the groundwater in the study region. The average groundwater level fluctuations from 2003 to 2017 of the West Azerbaijan Province can be observed in Fig. 3. Although the data were provided by three sources (CSR, GFZ and JPL), the results indicated that the groundwater levels in the region have been declining since the year 2008.

Fig. 3 Data from CSR, GFZ and JPL showing groundwater level changes

Fig. 4 The average water storage (in cm) in the years 2004, 2016 and between 2003-2017

The investigation of spatial variations in the water storage fluctuation can provide us with a better understanding of this regional data. Therefore, three maps of annual water storage fluctuation were developed using data from 2004, 2016, and 2003 to 2017, as shown in Fig. 4. The data were processed in GEE and ARCGIS. Among the 15 years that were investigated in this study, 2016 was found to be the driest year with the most decrease in water storage, while 2004 was a relatively wet year with the most increase in water storage. This decrease is in accordance with the average groundwater level decrease of the West Azerbaijan Province in Fig. 5, acquired from West Azerbaijan Regional Water Organization. In addition, the map of the average water storage between 2003 to 2017 shows that the biggest decrease in water storage occurred in the eastern and south-eastern regions of the province.

Fig. 5 Decrease in average groundwa

Fig. 6 Linear analyses of the CSR, GFZ, and JPL data with the CHIRPS data

To compare data from the three institutions(CSR, GFZ, and JPL), linear regressions were made on the processed GRACE data from each institution with CHIRPS result and data from monitoring wells from 2003 to 2017. The data plots are presented in Fig. 7, which also shows the linear trendlines, linear equations and the R2values.

The correlation coefficients of the linear regression between each of the three institutes(CSR, GFZ, and JPL) and the CHIRPS rainfall data were found to be 0.862 9, 0.861 0, 0.858 2,respectively, indicating that the CSR data was more appropriate for analyzing rainfall. Furthermore, the correlation coefficients between the three institutes and the monitoring well data were found to be 0.961, 0.970, 0.965, respectively, indicating that the GFZ data was more appropriate for monitoring fluctuations in the region’ s water storage. In the present study, data was collected from 199 monitoring wells in the West Azerbaijan Province during the years 2006, 2011, and 2016.The zoning map of water depth was created based on the Simple Kriging method using monitoring well data of the region (Fig. 8).

Fig. 7 Linear analyses of the CSR, GFZ, and JPL data with the data from monitoring wells

Zoning indicates that water levels are lowest in the east of Urmia and the south of Bukan and Qoshachay compared to other areas. Water levels in the north-western region of the West Azerbaijan Province (in Siah Cheshmeh, Maku, and the northwestern and south-western regions surrounding Khoy) are higher compared to other regions. These results indicate that the water levels in this province, during the 10-year period, had decreased significantly. In accordance with the results from the GRACE satellite data and based on the zoning maps, the overall changes of water levels in the region have been found to have a downward trend.Furthermore, the analysis and monitoring of the GRACE satellite sensor models for this region indicate more than 5 cm of decrease per year.Based on the rainfall data graphs (Fig. 9), it can be observed that this downward trend began from 2009. In addition, according to the regional water storage hydrograph in West Azerbaijan, the decrease in water levels of this aquifer that occurred in 2019 can be confirmed.

3.2 CHIRPS annual rainfall data results

In order to assess the rainfall changes as another hydrological factor, in this section the CHIRPS satellite data for the study area and period is analyzed and presented.

In the present study, the amount of monthly and annual rainfall in the West Azerbaijan Province between 2003 and 2017 was calculated using data from the CHIRPS database (Fig. 9 and Fig. 10).Fig. 9 shows that the biggest decrease in rainfall occurred in 2008 and this downward trend has remained steady for the following years.

3.3 Spatial variations in precipitation in the West Azerbaijan Province between 2003 and 2017

In recent years, the issue of climate change and its related problems have caused mass controversy internationally. Out of all factors pertaining to this issue, temperature and water have attracted more attention. Water is one of the most important factors when it comes to hydrological cycle which it directly affects. Spatial variations of rainfall during 2003, 2008 and 2017 in the West Azerbaijan Province can be observed in Fig. 11. The rainfall maps, show that precipitation in the western and southwestern regions of the province is relatively high compared to the eastern region (Rasooli et al.2013). Therefore, in accordance with the GRACE and data from monitoring well, it can be inferred that areas in which precipitation is scarce have less aquifer recharge and the water levels decrease more rapidly.

3.4 Discussion

Fig. 8 Water level zoning in 2006, 2011 and 2016, respectively

Fig. 9 Average monthly rainfall in the West Azerbaijan Province from 2003 to 2017

Fig. 10 Average annual rainfall in the West Azerbaijan Province from 2003 to 2017

Fig. 11 Spatial distribution of annual rainfall during 2006, 2008 and 2003-2017, respectively

Rodell et al. (2009) used terrestrial water storage change observations from the NASA GRACE satellites and simulated soil-water variations from a data-integrating hydrological modeling system over the Indian states of Rajasthan, Punjab and Haryana. Their findings showed applicability of GRACE data for groundwater depletion detection as we used in this research. Roudgarmi and Farahani (2017), assessed groundwater quantitative change in Tehran Province and their results showed there was an annual decline of 1.43 m in the groundwater level, which has been confirmed by comparing the similar studies in other parts of Iran. Rasooli et al. (2018) used Tropical Rainfall Measuring Mission (TRMM) satellite to estimate precipitation in the study area from 2006 to 2010.The results showed that the estimated values of TRMM and CHIRPS satellites are in good consistency, which is in conformance of this research. Shami et al. (2019) conducted a study in the Iranian region between 2003 and 2017, and Joodaki (2014) conducted another study in the south of Iran between 2002 and 2015, both of which were done using GRACE satellite data.Their results showed a more than 10 cm decrease in water levels between 2008 and 2017, and a negative trend in underground water storage in the southern and western regions of Iran. During the 8-year period of investigation in these studies, the total decrease of around 367 cubic kilometers of groundwater was reported. Both of these research results confirm the present research findings. Abou et al. (2019) monitored groundwater storage depletion using GRACE data in Bakhtegan catchment, Iran. They compared GRACE-derived water mass data against groundwater volume variations measured in situ and confirmed the efficiency of GRACE-derived data in catchments smaller than 200 000 km2as well as what this research revealed. Thomas and Famiglietti (2019)in their research identified the climate-induced groundwater depletion in GRACE observations.Similar to this research, they revealed that climate variability, especially precipitation has a great influence on groundwater depletion. Also, they confirmed that groundwater pumping is influencing too. These results are in accordance with the current research finding where the rainfall decrease in study period was determined by CHIRPS data and the main water source in the study region was from groundwater extraction,which caused the depletion of groundwater level as the GRACE results shows. Kazempour-Choursi et al. (2019) monitored the drought in the study region using TRMM and Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data. In this study, drought monitoring during 2000 to 2014 was performed using MODIS sensor and the estimated rainfall values were from TRMM satellite. The results of this study also indicate that with the decrease in rainfall, we have faced a decrease in groundwater level and in other words,there is a direct relationship between the decrease in groundwater level and an increase in drought indices of MODIS satellite.

The results indicate that GRACE satellite used in this study have a high ability to calculate the groundwater changes and precipitation values with considerable reliability. We can conclude that the trend of groundwater decline in the region is quite clear through satellite data and field data (Fig. 5).Also, the decrease in rainfall and climate change are the main reasons for the decrease in groundwater levels. Of course, man-made factors such as the construction of dams and the development of agriculture and urbanization in the region can also accelerate the decline of groundwater. Frequent construction of dams on the rivers has caused the lack of water supply to the farmers and the farmers have to drill illegal wells and abstract groundwater uncontrollably.

4 Conclusions

The depletion of underground water storage is one of the greatest crises of the century and leaves no nation untouched, forcing all countries to find ways to manage and preserve this precious resource.Due to limitations of using observation wells, data provided by the GRACE satellite was utilized in order to assess the fluctuations and variations in the water storage of the West Azerbaijan province.It was found that the GRACE data processed by the three algorithms (CSR, GFZ, and JPL) and the CHIRPS rainfall data have a relatively high correlation coefficient of around 0.86, indicating the decrease of water storage followed that of the precipitation in this region. Based on the GRACE satellite data on the fluctuations of annual water storage between 2003 and 2017, it was found that the biggest annual increase in water levels(5 cm) occurred in 2004, and the biggest annual decrease (16 cm) occurred in 2016. The average decrease in water storage during the period under investigation was 5 cm per year. In addition, the average decrease in water storage in the eastern,southeastern and the northern regions of the province were compared. Gravimetric satellite algorithms show a general decrease in groundwater level in the region. Also, due to climate change, the amount of precipitation has decreased in the study period, which indicates a direct relationship between the decrease of groundwater level and that of precipitation. The results indicate that the remote sensing satellites used in this study have a high ability to estimate groundwater changes and precipitation values with good reliability. With all the analysis, we can conclude that the decrease in rainfall, excessive withdrawal and the dams constructed by the farmers can have a significant impact on water storage in this region. Also,results showed that GRACE and CHIRPS data are valuable source for groundwater change detection.