IJME Vol7.1 284
IJME Vol7.1 284
IJME Vol7.1 284
1 January, 2022
1
Laboratory of Analysis and Modeling of Water and Natural Resources (LAMERN), Mohammed V University in Rabat,
Mohammadia School of Engineers, Rabat, Morocco
2
Geographic Information Systems Centre (GISC), Dubai Municipality, Dubai, United Arab Emirates
This work was supported by the Geographic Information Systems Centre (GISC), Dubai Municipality.
Abstract—Dubai's groundwater wells have become at risk of contamination, over-extraction, and over-irrigation for the past
two decades. Thus, the declining water quality and modest irrigation practices in many areas pressured farmers to install
packaged desalination equipment that caused a significant increase in the salinity levels of the groundwater, leading to
contaminants percolation to the aquifers.
This project explores the possibilities of using geospatial analytics such as Nearest Neighbour and Regression Analysis to
model and predict water levels. Such data is essential for the understanding and resolution of many complex water resources
issues commonly faced by hydrologists, water-supply managers, engineers, regulatory agencies, and the public.
It also features ModeflowMap;an innovative Enterprise GIS Solution developed tocollect groundwater quality parameters,
including the water level. This toolwas the primary solution used to collect groundwater data for47,433 wells and boreholes in
Dubai. Itwas designed as a preliminary effort to store the records relevant to the quality, quantity, biological, and chemical
properties, as well as the type and location of groundwater resources. Such data helped track and analyze the status and uses of
groundwater reserves. Furthermore, the findings assisted in improving and promoting the efficient use of water resources in the
Emirate.
Such comprehensive effort is mandatory to fully understand the state of the groundwater resources in the Emirate and support
the development of effective policies tomanage these resources efficiently.
Oneof the key purposes of this study is to highlight the importance of groundwater level measurementsasa fundamental
indicator of the status of thisresource and to foster a more comprehensive approach to the long-term collection of these
essential data.
1. INTRODUCTION
According to the United Nations , a country is defined as 'water scarce' if it has 1000 cubic meters or less of water available per
capita per year. The United Arab Emirates's natural water supply doesn't reach half this level, making it one of the world's most
water-scarce nations [1].One of its most populous cities with more than three million people, Dubai relies primarily on the
desalination plants, which consists 98.8 percent of its water supply [2]. In comparison, most of the groundwater in the UAE is
found in fossil aquifers that receive little to no waterrecharge. For every liter of water flowing into the country's groundwater
reservesfrom infrequent rainfall, twenty-five liters is withdrawn. This extraction rate leads tosevere degradation of the
remaining water supplies. Renewable water resources havedecreased by 42 percent since 2000, and further declines are
expected in the near future.Therefore, acute water shortages are expected in the region by 2025 [3].
Over-consumption is also a problem in the water sector. The UAE is the world's third-largest consumer of water, despite its
arid climate and lack of renewable water sources. Consumers in the UAE use around 400 liters of water each day, compared to
a global average of 250 liters. The UAE relies on its expensive desalination plants to meet this water demand [3].
Moreover, groundwater accounts for up to 94 % of water consumption in the agriculture sector in some emirates like Dubai
and Abu Dhabi. Nonetheless, the current usage of groundwater reservoirs exceeds the natural recharge rates by more than 15
times. Additionally, the Food Sustainability Index (FSI) ranked the UAE at the bottom of 34 countries for sustainable
agriculture in 2017, with the high use of dwindling resources in aquifers as a critical factor [4].
Furthermore, studies of existing productive wells concerning lithology and structures are minimal. Selection of well locations
forgroundwater supply used to heavily rely on traditional field studies using existing waterpoint sites as guidelines. In general,
a systematic approach to groundwaterexploration was also lacking. This research aims to contribute towardssystematic
groundwater analysis utilizing remote sensing, field data collection, DigitalElevation Models (DEM), and Geospatial Analysis
using Geographic Information Systems (GIS),assistingtheassessment of groundwater resources in the UAE.
Therefore, ModeflowMap is introduced in this paper, an innovative enterprise GIS Solution comprising several modules that
cover essential requirements for a better understanding of groundwater conditions. This enterprise solution controls the field
data collection operations, processes high-resolution remote sensing images, extracts vector data layers, ensures data quality
control, enables advanced geospatial analytics, and disseminates the results in the form of thematic maps and dashboards.
Consequently, it supports monitoring groundwater uses in agriculture in real-time. This solution also supports government
entities as a decision support tool, including all actors involved in water management and water policy-makers at the field
level.
In recent years, Dubai'saquifers' conditions have improved thanks tothe measures taken to achieve maximum overall cycle
efficiency for all groundwater uses, including reducing groundwater abstraction to sustainable levels. However, full recovery
will take decades. In addition, a comprehensive set of actions for sustainable groundwater management have been adopted,
notably establishing robust monitoring and regulatory programs to conserve traditional water systems such as aflaj [8].
Fig.2 represents a spatial distribution of wells density in the Dubai Emirate. The total records used for this study are 47,433.
During this study, ModeflowMap Mobile was used to collect groundwater well information in order to validate the results
obtained by desktop application; the results of this Survey also supported calibrating the Hatta groundwater models to ensure
accurate results. Furthermore, these inspections allow the collection of irrigation activities information in the crops and either
change or add any relevant agronomic data, including the following:
Table 1.Boreholes Sample Data featuring boreholes ID, boring date, location, ground level, water level, depth and diameter
BORINGD DMDGD_LE GD
BO No EASTING NORTHING DEPTH DIAMETER
ATE VEL WATER_LEVEL
12827 2008-08-14 490490 2768071 0 28.02 15 0.15
11651 2007-05-30 488967 2770785 30.44 22.6 30 0.15
11642 2007-05-30 488955 2770813 30.56 22.6 30 0.15
13328 2004-09-10 489250 2769778 31.07 22.55 30 0.13
11652 2007-06-07 489068 2770884 30.5 22.5 30 0.15
11617 2004-09-02 487267 2770775 29.72 22.5 29.9 0.13
11650 2007-06-06 489092 2770842 30.47 22.4 30 0.15
11653 2007-05-25 488987 2770870 30.46 22.4 30 0.15
11641 2007-05-27 488986 2770832 30.57 22.1 30 0.15
13336 2004-08-28 488317 2769271 28.22 21.5 29.9 0.13
43542 2008-04-17 489326 2771018 31.75 20 35 0.15
13322 2004-09-06 488906 2769514 28.13 20 30 0.13
43543 2008-04-17 489305 2771030 31.82 19.8 35 0.15
43544 2008-04-19 489305 2771045 31.12 19.6 35 0.15
43545 2008-04-20 489331 2771032 30.99 19.5 35 0.15
Throughout this project, 47,433 wells were visited in Dubai. Most wells are located on farms, along Dubai's tunnel, and in
forests except some remote desert locations. Trained teams with experienced staff and dedicated advanced equipment were
assigned to these remote locations, where off-road driving capabilities and extensive safety measures were mandatoryfor the
Survey's success. For each surveyed well, more than 100 parameters werecollected.
Additionally, the operational status and the use of the wells were assessed. Groundwater level, salinity, temperature, and pH
were also measured. The teams recorded pumping operation time, well discharge, and selected farm data.
The soil data collected for every depth was validated in the laboratory and documented at first in a metadata sheet as a PDF
document. Then, a database model was designed to store this information and link it to the main wells and boreholes records
through a unique ID for every soil type.Automating the electronic transformation was performed using FME Workbench [9].
Fig. 5 demonstrates the soil types recorded and validated by a local laboratory for BH-BDT-44. It confirms the depth of the
sample collected, core recovery elements, thickness and description of the structure found.
For best groundwater data analysis, different formats ofinputs are required, including raster, vector, and alphanumeric, to be
controlled by shared desktop-GIS, web-GIS, and Mobile-GIS environments. The solution was developed to combine data from
drones, field agronomic inspections, cadastre, and information about irrigation rights, among others.
This project relied on themultispectral drone imagery captured by the field team,which was used for the generation ofthe
Normalized Difference Vegetation Index (NDVI) maps [10]. Such practice is essential for soil segregation from vegetation
cover, differentiation between crops at different crop stages, and detection of plants under stress. Hence, it has been proven that
there are strong correlations between NDVI data measured at certain crop stages and crop yield. Hence, tracking the crop
growth at crucial steps helps estimate the crop yield and addresses issues early [11-12]. Table 3 lists the different levels of
acquisition performance of the Trimble UX5 Drone.
The Trimble UX5 HP drone used during the field survey is equipped with a modified color-infrared (CIR) Sony NEX5R fitted
with a 16 mm lens. On each flight day, roughly 100 ground-based normalized difference vegetation index (NDVI)
measurements were collected with the Trimble Green Seeker Handheld at a constant height of 80 cm above the target, of which
the center point was georeferenced to 2 cm accuracy using a Trimble R8RTK GNSS system [13]. Fig.6 shows an example of
the drone outputs such as an ortho-rectified image with elevation contours.
One of the most efficient and oldest distance statistics is the nearest neighbor index. It is advantageous because it is a simple
tool to understand and calculate. Since two botanists developed it in the 1950s [15], it was primarily for fieldwork, but it has
also been used in many fields for a wide variety of problems[16]. It was also used inseveral other distance statistics
implemented in groundwater analysis and geostatistics.
Where Do is the observed mean distance between each feature and its nearest neighbor:
n
Do = (∑k=1 dk) /n (3)
And DE is the expected mean distance for the features given in a random pattern:
0.5
De = (4)
√n/A
In the above equation, di equals the distance between feature i and its nearest neighboring feature. While n corresponds to the
total number of features, A is the area of a minimum enclosing rectangle around all features or a user-specified Area Value.
The average nearest neighbor z-score for the statistic is calculated as:
Do−De
z= (5)
SE
Where:
0.26136
SE = 1 + (6)
√n2 /A
The pattern exhibits clustering if the average nearest neighbor index (ratio) is less than 1. However, if the index is greater than
1, the trend is toward dispersion.
Given the z-score of -359.279371 in table 5, there is a less than 1% probability that this clustered pattern could result froma
random distribution. This means that the distribution of the wells subject of this study is clustered and could be considered a
good representative for groundwater analysis and prediction in the study area [18].
During the analysis phase of this project, a regression equation was used in order to assess the relationship between the
collected water levels at the wells and their locations. In our case, the dependent variable to be modeledis the water level, while
the independent or explanatory variables are Easting and Northings (Coordinates). Every independent variable is associated
with a regression coefficient that describes the relationship's strength between the variable and the dependent one. The
regression equation used is described belowwhere Y is the dependent variable (Water Level), the Xs are the explanatory
variables (Northings and Eastings), and the βs are regression coefficients:
The Residuals are the unexplained portion of the dependent variable (well water level), represented in the regression equation
as the random error term ε.
According to Average Nearest Neighbor summary results, the clustered distribution of the wells in Dubai Emirate suggests that
the Dubai wells distribution is clustered and not random, which means that groundwater parameters for a specific well with
unknown values can be estimated and predicted based on the nearest wells with known values (z score = -359.279371). Fig. 8
demonstrates the category of Dubai wells distribution as clustered.
The measurements of water levels in groundwater wellsprovide the most fundamental indicator of the status of thisresource.
They are critical to meaningful evaluations of the quantity and quality of groundwater and its interaction with surfacewater.
Therefore, scatter plotting isused in this study as another way to visualize the relationship between the spatial location and the
water level as essential groundwater parameters. As described in the methodology section, a regression equation had been
calculated, and the associated trend line and R² were plotted on the scatter plots (Fig.9). The trend line models the linear
relationship between Latitude, Longitude, and water levels, respectively, and the R² quantifies how well the data fits the model.
Fig.9. Scatter plotting of groundwater level vs.(a) Longitude and (b) Latitude
Fig.9 depictsthe relationship between the dependant variable, which is, in this case, the water level against the longitude and
latitude as explanatory variables.Overall, for both explanatory variables, R-squared = 0.01, which is 1%, and this means that
the model doesn't fit well for prediction if we take into consideration all wells as a frame for sampling; however, more than
50% of the wells fit thisregression model with more than 85%. These wells that fit the regression model are interactively
selected through the system and can be used as the main sampling frame (Fig 10).
(a) (b)
Fig.10. Scatter plotting and selection of best fitting wells to the regression model (85% fit)(a) Longitude and (b) Latitude
Evaluating the residualsis also very important since it can display problematic patterns in the residuals. Fig. 11 portrays these
residuals for better regression analysis of the results [20].
Overall, this histogram of residuals matchesapproximately the standard curve indicated above in blue.This supports that the
regression model is not biased [21].
In order to focus the study and the analysis results on the wells that match the regression model discussed in this paper, a
selection of 29,713 wells was made to create the mainframe of wells for groundwater analysis and strategic planning-related
tasks. These wells, along with the archived ones, were stored in the central ModeflowMap enterprise GIS database for further
analysis, dissemination, and evaluation. Fig. 12 shows an example of filtering the wells based on their type. These wells
highlighted in red represent the Dubai Tunnel wells used to monitor the groundwater resources in the proximity of the world's
biggest stormwater drainage tunnel.
Fig. 13 displays the differences between the calculated and actual water depths (retrieved from ModeflowMap database)
compared to the Dubai Municipality Datum (DMD for some boreholes where the model gives the acceptable results (b) and
where it doesn't (a) [22].
In total, 47,433 groundwater wells were registered in Dubai. Of the wells surveyed 43.9 percent were found operational. Most
of the wells belonged to agricultural areas, followed by forests and remote well fields. 141 wells were found in regions outside
redefined municipal boundaries. Depth to groundwater were measurements recorded for 24,312 wells showing that
approximately a third of the wells were dried-out with an average depth of 102 meters (m) below the surface and large
regional.
(b)
Fig.13. Modeflow Map Web Module- Dissemination of water level prediction vs. Actual
This effort helps establish a baseline for groundwater resources throughout the seasons to identify key performance indicators
for agricultural activities' future management plans.
This study also depicts the results of standard spatial analytics used for groundwater, including data elevation modeling,
vegetation cover mapping,wells distribution…etc. Fig 14. represents the data elevation model generated from a total of
298,484control points with 100 meters resolution, which was used as the primary reference for calculating the well ground
depths versus the Dubai Municipality Datum (DMD), which is also referred to as Dubai Maritime Datum [22].
Conclusions
The drive of this project has been to illustratethe importance of systematic, long-term collectionof water-level data. Such data
are fundamental to understanding and resolving complex water resource issues commonly faced by hydrologists, water-supply
managers, engineers, regulatory agencies, and the public.
In order to ensure that adequate waterlevel data are being collected for present and anticipated future uses, observation-well
networks need to be evaluated periodically. Therefore, this study depicts the field survey results conducted in Dubai, which
covered 47,433 wells and boreholes in the Emirate.
This field survey was conducted using ModeflowMap; an innovative Enterprise GIS Solution developed to collect groundwater
quality parameters, including the water level. This tool was the primary solution used to collect groundwater data for 47,433
wells and boreholes in Dubai. It was designed as a preliminary effort to store the records relevant to the quality, quantity,
biological, and chemical properties, as well as the type and location of groundwater resources. Such data helped track and
Given the importance of the coordinated efforts between the concerned government departments to collect, analyze and share
spatial data related to groundwater resources, the task of deploying ModeflowMap as an integrated solution for collecting
critical groundwater wells information to enhance the simulation and the prediction of groundwater is of great value for
severalentities. Moreover, the results of this project support groundwater scientists in establishing a clear plan for
implementing the necessary framework of policies, data procedures, partnerships, standards, technology, and institutional
capabilities that collectively will comprise Dubai's groundwater Resources Hub. +
However, a commitment to long-term monitoring is needed to avoid data gaps resulting from an inadequate distribution of
wells or periods of no measurements in a hydrologic record.Disruptions in the hydrologic register can hinder the ability of
water-resources managers to make sound decisions. Wherever water-level data are not available, hydrologic information
needed to address critical groundwater problems may be impossible to obtain. Although ModeflowMapgroundwater databases
can be accessed by many entities, detailed and complete records of historical groundwater data are limited or unavailable.
Consequently, potentially valuable data reside in paper files where accessibility and utility are minimal. Finally, to improve the
collection and accessibility of water-level data, Dubai Municipality needs to examine ways to enhance interagency
coordination in constructing and maintaining observationwells networks, collecting waterlevel measurements, and sharing and
disseminating spatial data. Better interagency cooperation will definitely help ensure that data-collection efforts are sufficient
to address issues relevant to the greatest variety of the groundwater-resources challenges.
REFERENCES
1. FAO,"Coping with water scarcity in agriculture: a global framework for action in a changing climate", 2016.
2. Mervyn Piesse, "UAE and Israel Discuss Closer Co-operation in Food and Water Security", Global Food and Water Crises
Research Programm, 2020
3. Lauren Power, "Scarcity and Abundance: UAE Food and Water Security", Global Food and Water Crises Research
Programme, 2014.
4. L.El Hoummaidi, A. Larabi, "GIS application in hydrogeological studies (Case Study: Dubai and Abu DhabiEmirates)",
Turkish Journal of Computer and Mathematics Education, Volume 12, Number 10, pages 3893-3899, 2021
5. Mullah, M. -al, 2011. UAE State of Water Report. Second Arab Water Forum, Cairo, 20-23 November 2011.
6. ESCWA, "Water scarcity in the Arab world. Population and development report", Beirut, Lebanon,2003.
7. Nouh, M., "An Overview for the Water Resources of the United Arab Emirates", Proceedings of the 1st Technical Meeting
of Muslim Water Researchers Cooperation, Malaysia, 2008.
8. Natural Resources Conservation Section, "TECHNICAL GUIDANCE FOR GROUNDWATER PROTECTED AREAS
IN EMIRATE OF DUBAI", 2020.
9. Bob Lyner, "Digitization Process, Digital Imaging", Digital Preservation, 2016.
10. Starý, K.; Jelínek, Z.; Kumhálová, J.;Chyba, J.; Balážová, K.," Comparing RGB - based vegetation indices from UAV
imageries to estimate hops canopy area ", Agronomy Research, Volume 18, Number 4, 2020. DOI: 10.15159/ar.20.169
11. Christiansen, M.P.; Laursen, M.S.; Jørgensen, R.N.; Skovsen, S.; Gislum, R. "Designing and Testing a UAV Mapping
System for Agricultural Field Surveying". Sensors, Volume 17, page 2703, 2017. DOI: 10.3390/s17122703