Table of Contents

Candidate’s Declaration ............................................................................................................ I

Abstract ................................................................................................................................. II
Acknowledgement .................................................................................................................. III
Table of Contents ................................................................................................................... IV
List of Figures......................................................................................................................... V
List of Tables ......................................................................................................................... VI
List of Abbreviations ............................................................................................................. VII
CHAPTER 1. INTRODUCTION............................................................................................ 1
1.1 Background and Significance of Study ....................................................................... 1
1.2 Purpose and Objective of Study ................................................................................. 1
1.3 Organization of Thesis .............................................................................................. 2
CHAPTER 2. STUDY AREA ................................................................................................. 4
2.1 A brief overview (Geographical characteristics) of the study area .................................. 4
2.2 Hydrological characteristics of the study area .............................................................. 5
2.3 Geology and Soil properties of the study area .............................................................. 6
CHAPTER 3. LITERATURE REVIEW ................................................................................. 8
3.1 GRACE Background ................................................................................................ 8
3.2 GLDAS Overview.................................................................................................... 9
3.3 Satellite-based Hydrological Research: Advances and Applications ............................... 9
3.4 Drought Monitoring Background ............................................................................. 12
3.5 Drought in India..................................................................................................... 13
3.6 Drought Monitoring Techniques: A Comprehensive Review of Satellite-Derived Indices 13
CHAPTER 4. DATASETS AND METHODOLOGY ............................................................. 14
4.1 Data Processing and Descriptions............................................................................. 15
4.2 Retrieval of Groundwater Storage Change................................................................. 15
4.3 Estimation of GRACE Groundwater Drought Potential Index (GDPI) .......................... 16
4.4 Mann Kendall Trend test ......................................................................................... 18
4.5 Standardized Precipitation Index .............................................................................. 19

IV
 List of Figures
Figure 1: Map of the Study Area ............................................................................................... 4
Figure 2: The Gravity Recovery and Climate Experiment (GRACE) mission is shown artistically. ... 8
Figure 3: Explains GRACE mission's observation of the time-varying gravitational field ................ 8

V
 List of Tables
Table 1: Salient features of the study area ................................................................................... 5
Table 2: Sub-basin-wise watersheds ........................................................................................... 5
Table 3: The study makes use of the following datasets .............................................................. 15
Table 4: SPI values ................................................................................................................ 19

VI
 List of Abbreviations
GRACE Gravity Recovery and Climate Experiment
IMD India Meteorological Department
GLDAS Global Land Data Assimilation System
TWS Terrestrial Water Storage
GWS Groundwater Storage
GDPI Groundwater Drought Potential Index
DPA Drought Potential Amount
SPI Standardized Precipitation Index

VII
VIII
 CHAPTER 1. INTRODUCTION

1.1 Background and Significance of Study

Groundwater serves as a crucial drinking water source, catering to the domestic needs of around
half of the world's population (Zektser & I. & Everett, n.d.). Groundwater provides water for
more than 80% of agricultural needs and about 60% of the drinking water in India (Sarkar et
al., 2020). This makes it highly valuable and in high demand. The Great Artesian Basin in
Australia is the world's largest aquifer, spanning 1.7 million square kilometres and containing
64.9 trillion cubic meters of water. Groundwater is indispensable in numerous facets, including
ensuring food security, maintaining water availability, and fostering economic development
for over 1.3 billion individuals in India (Dangar et al., 2021). The Ganga Basin ranks among
the most densely populated river basins globally, hosting a population of 600 million people
with a population density of 550 persons per square kilometre (S. K. Jain et al., 2009). The
Ganga Plain is the largest groundwater reserve, upon which the entire population of India
heavily relies for various developmental endeavours, including domestic, agricultural, and
industrial purposes (Bates et al., n.d.; Bhardwaj et al., 2010).
The study focuses on the sub-surface hydrogeological investigation conducted in the Gangetic
Plain of India to estimate changes in regional groundwater levels using remote sensing
technology. The Gangetic Plain is a significant groundwater-bearing region, surrounded by
thick strata of alluvial deposits and rich aquifer systems. The study aims to address the
increasing impact of climate change on water resources and develop sustainable management
strategies. The significance of the study lies in its potential to provide vital insights for
understanding future scenarios, planning sustainable management strategies, and aiding in the
development of adaptation and mitigation measures. The integration of satellite-based
observation techniques, such as GRACE and GLDAS, holds a promising future for addressing
water resource management challenges and developing an early warning system for water
scarcity. This study is crucial for groundwater management, conservation, and understanding
the complex interactions within the hydrological cycle.

1.2 Purpose and Objective of Study

The study aims to conduct a sub-surface hydrogeological investigation in the Gangetic Plain
of India, focusing on the region's groundwater resources, aquifer characteristics, and overall
hydrogeological conditions. The study aims to estimate the changes in subsurface groundwater
levels for the pre-monsoon and post-monsoon period of January 2003 to December 2022 using
Gravity Recovery and Climate Experiment (GRACE) and Global Land Data Assimilation
System (GLDAS) data. The primary objective is to understand the dynamics of groundwater
storage, depletion, problems, and strategies for mitigation in the Ganga basin and all its 19 sub-
basins.
The study also aims to evaluate the dynamics of GRACE groundwater and its drought potential
in the Ganga basin and its subbasin of India. The study's objectives are to reveal groundwater
storage (GWS) dynamics in the study area using satellite- and model-based observations,
separate the relative contributions of climate and anthropogenic activities to GWS changes,

1
and establish a new index to evaluate groundwater drought comprehensively. The study also
includes analysing drought characteristics in India at sub-basin scales using observed data,
specifically focusing on the spatiotemporal changes in the GRACE-GDPI drought indices. The
study aims to evaluate trends in data, identify regions with significant trends, and determine
the magnitude of change using statistical tests. Additionally, the study seeks to characterize
drought events based on the identified indices and understand whether drought is due to a
deficit in precipitation, terrestrial water storage, or both with the help of calculated SPI. The
goal is to contribute to understanding drought at the sub-basin spatial scale and provide
valuable insights for drought management and mitigation efforts. Additionally, the study seeks
to assess the impact of climate change on groundwater resources and develop sustainable water
management strategies.
The study aims to provide clear knowledge of water resource dynamics, better explain the
driving mechanisms of groundwater change, and improve water resource management and
understanding of groundwater operation mechanisms.
Further, the objectives of this study are:
 To examine changes in groundwater storage derived from GRACE/FO data within the
Ganga Basin and its 19 sub-basins, focusing on longer-term decadal trends and
observations specific to the region.
 To quantify groundwater recharge and loss phases by integrating GRACE/FO-derived
Total Water Storage (TWS) with other terrestrial water components over the past two
decades.
 Evaluate the applicability and reliability of GDPI and SPI as practical tools for
assessing groundwater drought and precipitation variability in the Ganga Basin and its
19 sub-basins, thereby providing valuable insights for enhancing drought management
and mitigation strategies.

1.3 Organization of Thesis

The thesis work is organized into five chapters as follows:
Chapter 1 provides an overview of the groundwater dynamics, followed by an explanation of
the drought. It also explains why this study is so vital, and the objectives of the present thesis,
followed by the scope of the study, are also defined in this chapter.
Chapter 2: The study area is discussed in this chapter.
Chapter 3 delves into satellite-based hydrological research, detailing the GRACE mission's
significance, including its methodologies and solutions for estimating Terrestrial Water
Storage Anomaly (TWSA). It also explores the Global Land Data Assimilation System
(GLDAS) project and reviews recent studies in satellite-based hydrological research,
emphasizing their methodologies and implications for water resource management.
Chapter 4 describes the data acquired and methodology used in estimating the groundwater
storage change, estimation of change in volume and a GPA, GDPI and SPI calculation for the
study area.

2
Chapter 5 discusses the conclusion of the whole study, followed by a study of the future scope
of this research.

3
 CHAPTER 2. STUDY AREA

2.1 A brief overview (Geographical characteristics) of the study area

Ganga basin: The Ganga basin encompasses vast expanses of land across India, Tibet (China),
Nepal, and Bangladesh, covering a total area of 10,86,000 square kilometres. Predominantly
situated within India, it constitutes the largest river basin in the country, spanning an area of
8,61,452 square kilometres, which is roughly 26.3% of India's total geographical area. The
basin extends across multiple states in India, including Uttar Pradesh, Madhya Pradesh,
Rajasthan, Bihar, West Bengal, Uttarakhand, Jharkhand, Haryana, Chhattisgarh, Himachal
Pradesh, and Delhi. The Ganga River, originating from the Gangotri glacier near Gomukh in
Uttarakhand, descends through the valley, gathering momentum as it merges with various
tributaries such as the Bhagirathi, Alaknanda, Mandakini, Dhuli Ganga, and Pindar. With a
total length of 2,525 kilometres, the Ganga River is vital in sustaining life and livelihoods
across the basin. The socio-economic significance of the Ganga basin is profound, with a rich
history of human civilization dating back centuries. Considered one of the holiest rivers in the
world, the Ganga holds immense cultural and spiritual significance for the people of India. The
basin supports a large population, accounting for approximately 43% of India's population,
with extensive agricultural lands, water bodies, and dense urban settlements (Dr. J R Sharma
& Er. Yogesh Paithankar, n.d.; GOI Ministry of Water Resources & Ganga Basin, n.d.).

Figure 1: Map of the Study Area

4
Salient features of the Ganga Basin are tabulated below:
Table 1: Salient features of the study area
No. Features Specification
1 Basin Extent 73° 2’ to 89° 5’ E
21° 6’ to 31° 21’ N
2 Area (Sq.km) 10,86,000 (Total)
8,61,452 (Within India)
3 Length of Ganga River (km) 2525
4 States in the basin Uttar Pradesh (28.02 %)
Madhya Pradesh (21.02 %)
Rajasthan (13.06 %)
Bihar (10.86%)
West Bengal (8.3 %)
Uttarakhand (6.15 %)
Jharkhand (5.85 %)
Haryana (3.99 %)
Chhattisgarh (2.08 %)
Himachal Pradesh (0.5 %)
Delhi (0.17 %)
5 Districts (Census 2011) 252
6 Average Annual Rainfall (mm) 1059.74
7 Average Maximum Temperature (oC) 32.05
8 Average Minimum Temperature (o C) 18.44°
9 Number of Sub Basins 19
10 Number of Watersheds 980

2.2 Hydrological characteristics of the study area

In the India-WRIS project, a semi-automated approach delineates hydrological units, including
basins, sub-basins, and watersheds. The Ganga Basin is subdivided into 19 sub-basins, each
representing a distinct hydrological unit within the larger basin. Hydrological units are further
subdivided into watersheds to facilitate water resources management at a micro level. A
watershed is a natural hydrological entity encompassing a specific land area from which rainfall
flows into a defined drainage system. The size of a watershed is determined by the size of the
stream and its boundaries. Within the Ganga Basin, the 19 sub-basins are classified into 980
watersheds, each representing a distinct tributary system. This hierarchical delineation of
hydrological units enables comprehensive assessment and management of water resources
within the Ganga Basin, facilitating effective planning and decision-making for sustainable
water use and conservation(Dr. J R Sharma & Er. Yogesh Paithankar, n.d.; GOI Ministry of
Water Resources & Ganga Basin, n.d.).
Table 2: Sub-basin-wise watersheds
Sub Basin Area No. of
(Sq.km.) Watersheds
Above Ramganga Confluence Sub Basin 39104.61 51
Banas Sub Basin 51651.51 64

5
 Bhagirathi and others (Ganga Lower) Sub Basin 64038.97 75
Chambal Lower Sub Basin 10941.26 14
Chambal Upper Sub Basin 25546.57 30
Damodar Sub Basin 41965.49 60
Gandak and others Sub Basin 56260.43 76
Ghaghara Confluence to Gomti confluence Sub Basin 58634.18 36
Ghaghara Sub Basin 26254.06 76
Gomti Sub Basin 29865.21 41
Kali Sindh and others up to Confluence with Parbati Sub 48492.61 64
Basin
Kosi Sub Basin 18413.58 19
Ramganga Sub Basin 30839.69 40
Sone Sub Basin 65110.05 83
Tons Sub Basin 16905.74 23
Upstream of Gomti confluence to Muzaffarnagar Sub Basin 29061.37 40
Yamuna Lower Sub Basin 124867.19 98
Yamuna Middle Sub Basin 34586.39 43
Yamuna Upper Sub Basin 35798.19 47

2.3 Geology and Soil properties of the study area

The Ganga basin exhibits diverse geological and soil characteristics influenced by its
topography, climate, and geological history. Understanding these properties is crucial for
assessing the region's hydrological behaviour and agricultural potential.
Geological Composition: The Ganga basin encompasses three major geological regions: the
Himalayan Young Fold Mountains in the north, the Gangetic Plain, and the Central Indian
highlands in the south. These regions differ in geological structure, with the Himalayas
consisting of young-fold mountains and the Gangetic Plain characterized by extensive alluvial
deposits. The Deccan plateau in the south comprises residual soils resulting from the
weathering of ancient rocks.
Soil Diversity: The basin hosts various soils, ranging from alluvial soils in the plains to lateritic
soils in the undulating tracts along the Chottanagpur highlands. Major soil types include:
Alluvial Soil Covers over 52% of the basin, characterized by rich soil nutrients deposited by
the Ganga and its tributaries. Usar and Bhur soil (Variants of alluvial soils) found in the Upper
Ganga plains, with varying drainage conditions and nutrient compositions. Bhat Soil
(Calcareous soil) found in the lower Gandak valley, known for its fertility and high
productivity. Laterite and Red Soil found in regions such as Chottanagpur highlands and
Maldah-Dinajpur districts, characterized by low water-holding capacity and moderate to severe
erodibility.
Terrain and Slope: The basin's terrain ranges from steep slopes in the Himalayan foothills to
level plains in the Gangetic region. Parts of Uttarakhand and Madhya Pradesh have steep slopes
due to the Lesser Himalayas and Siwalik ranges, while lower regions like Jharkhand and
Chhattisgarh have gentle slopes. These variations influence soil erosion patterns and
agricultural productivity.

6
Agricultural Productivity: Soil productivity varies across the basin, with regions like Uttar
Pradesh exhibiting high productivity due to fertile soils and favourable climatic conditions.
However, areas with hilly terrain like Uttarakhand may have lower productivity. The basin
supports various crops, including paddy, wheat, millets, pulses, cotton, and jute, contributing
significantly to India's agricultural output (Dr. J R Sharma & Er. Yogesh Paithankar, n.d.; GOI
Ministry of Water Resources & Ganga Basin, n.d.).

7
 CHAPTER 3. LITERATURE REVIEW

3.1 GRACE Background

The GRACE experiment, short for Gravity Recovery and Climate Experiment, is used in
mapping the Earth's gravity potential. By analysing gravity maps generated by GRACE,
scientists gain insights into alterations in water storage on both regional and global scales, as
well as fluctuations in ice mass and shifts in sea levels attributed to the ocean's water expansion
(Frappart & Ramillien, 2018).

Overview: NASA and the German Aerospace Centre work together to run the GRACE
mission. It was launched on March 17, 2002, and terminated in 2017(figure 2). It comprised
two spacecraft orbiting in tandem between 450 and 500 km above the earth, about 200 km
apart. The basic idea behind gravity readings is depicted in a simplified manner in Fig. 2. The
two GRACE satellites flew in a tandem orbit over the Earth (first panel). In the scenario
depicted in Fig. 3, a peak is a positive mass anomaly. The mass anomaly's gravitational
attraction caused the lead satellite to be drawn towards the mountain as it moved in its direction.
Thus, the distance between the two satellites increased (second panel). The same mass anomaly
held back the lead satellite as the trailing satellite neared the mountain, and as they got closer
to one another, the gap between them shrunk (third panel). The trailing satellite was eventually
held back as it backed away from the peak, increasing their separation again (fourth panel).
Essentially, the measurement of the distance (range) and the rate of change of distance (range-
rate) between the two satellites is linked to the gravitational potential of Earth's
masses(Tangdamrongsub et al., 2016). The monthly examination of range and range-rate
measurements between the satellites yields temporal variations, commonly called anomalies,
in the Earth's gravity field (Girotto & Rodell, 2019).

Figure 2: The Gravity Recovery and Climate Figure 3: Explains GRACE mission's observation of the
Experiment (GRACE) mission is shown artistically. time-varying gravitational field.
Source:
https://www.jpl.nasa.gov/spaceimages/images/largesi Source:(Girotto & Rodell, 2019)
8
ze/ PIA04235_hires.jpg
GRACE solutions: The classic spherical harmonic technique (SH) and the Mascon approach
are the two solutions used to calculate TWSA estimates. Mass concentration blocks (mascon)
are used to construct the mascon-based solutions. In this solution, Earth's surface is divided
into small, equal-sized regions (mascon), each with a unique mass value. Mascon solutions
have fewer leakage errors than SH solutions, which results in increasing signal amplitude. This
Mascon solution has a Higher spatial resolution and is suitable for regional studies, localised
mass variations, and specific geographic features. It can identify land and ocean areas at
regional and global scales. On the other hand, a typical spherical harmonic technique is used
to process the SH-based GRACE solutions; it represents the Earth's gravitational field as a
series of spherical harmonics functions. These Spherical harmonic solutions cannot
discriminate between land and sea areas. It has Lower spatial resolution than MASCON
solutions and is suitable for global-scale studies, capturing large-scale mass changes.
The GRACE TWSA solutions are released by three different processing centres: the
Geoforschungs Zentrum Potsdam (GFZ), the Centre for Space Research at the University of
Texas, Austin (CSR), and the Jet Propulsion Laboratory (JPL). These centres use different
approaches and parameters, resulting in different GRACE solutions(Cheng et al., 2013;
Mouyen, 2023; Scanlon et al., 2016; Wahr et al., 1998; Watkins et al., 2015).

3.2 GLDAS Overview

The Global Land Data Assimilation System (GLDAS) is a project funded by NASA's Energy
and Water Cycle Study (NEWS) Program, and it aims to generate land surface state and flux
products simulated by four land surface models: CLM, Mosaic, Noah, and VIC. These products
include soil moisture, surface temperature, evaporation, and sensible heat flux and are
accessible at the Hydrology Data and Information Services Center (HDISC), a component of
the NASA Goddard Earth Sciences Data and Information Services Center (GES DISC). The
dataset includes data products with 1.0-degree resolution from all four models, spanning 1979
to the present, and a 0.25-degree data product from the Noah model, spanning 2000 to the
present. The products are in Gridded Binary (GRIB) format and can be accessed through
various interfaces, allowing users to search, subset, and convert the data. Additionally,
advanced visualization, access, and analysis capabilities are being developed. The long-term
GLDAS data are used to develop the climatology of water cycle components and explore the
teleconnections of droughts and pluvials. This data is crucial for understanding the global
hydrospheric cycle and is valuable for climate simulation and weather forecasting (Kato
Beaudoing & Teng, n.d.).

3.3 Satellite-based Hydrological Research: Advances and Applications

After an extensive review of the relevant literature in this field, I have thoroughly examined
various papers to gain a comprehensive understanding of the work that has been undertaken.
My analysis and synthesis of these papers lead me to the following conclusions.
1. Jiang et al. (2023) describe how data from the Gravity Recovery and Climate Experiment
(GRACE) were used to create a Modified Flood Potential Index (MFPI) for flood
monitoring in the Yarlung Tsangpo-Brahmaputra River Basin (YBRB). The MFPI

9
 concentrates on the periodic and residual signals of Terrestrial Water Storage Anomaly
(TWSA) and eliminates the long-term trend of TWSA. It performs better than GRACE-
based flood indices in catching flood episodes and representing their spatiotemporal
processes. The MFPI is helpful in regions with notable TWSA tendencies and can
efficiently monitor large-scale flood disasters. The study addresses the shortcomings and
ambiguities of the MFPI and emphasises how crucial it is to consider how dams affect flood
monitoring.
2. Mohanasundaram et al. (2021) focus on calculating streamflow and baseflow in the basins
of the conterminous United States using GRACE mission records. The study used a
complementary relationship (CR) methodology to estimate evapotranspiration (ET) and a
water balance method to estimate runoff and baseflow. The CR-GRACE model is tested
using USGS runoff information and compared to Land Surface Model (LSM) projections.
The results show that the CR-GRACE model produces appropriate predictions of runoff
and baseflow with restricted parameterisation and datasets. Overall, this study confirms the
utility of GRACE data products for calculating streamflow and baseflow in water resource
management.
3. Sarkar et al. (2020) use GRACE satellite data to examine the interannual variability of
groundwater storage in water-stressed areas of India and the reasons behind it. The analysis
was conducted from January 2003 to December 2016, with separate analysis for different
seasons (pre-monsoon, monsoon, and post-monsoon). To calculate groundwater storage
change, they subtracted total soil moisture, canopy water storage, and snow water
equivalent data from the TWS data obtained from GRACE. The soil moisture data was
derived from the GLDAS NOAH model. The study also compared the GRACE-derived
groundwater storage change with in-situ well data provided by India's Central Ground
Water Board (CGWB).
4. Xie et al. (2022) combine GRACE and GRACE-FO satellite data to monitor extreme flood
periods in the Yangtze River basin. Based on precipitation, temperature, and soil moisture
storage anomaly (SMSA) data, the researchers created models to reconstruct and
temporally downscale estimates of terrestrial water storage anomaly (TWSA) at daily
scales. They used methodology involving a combination of satellite data, meteorological
observations, machine learning models, and statistical analysis to monitor and downscale
TWSA using they developed a new indicator, the normalised daily flood potential index
(NDFPI), to track severe flood periods monthly. These findings majorly affect flood risk
prevention and reduction in the Yangtze River basin.
5. Long et al. (2014) analyse the increasing frequency and intensity of droughts and floods on
the Yun-Gui Plateau in Southwest China using data from the Gravity Recovery and Climate
Experiment (GRACE) satellite and an artificial neural network (ANN) model. The ANN
model accurately predicted total water storage anomalies (TWSA) for the plateau and sub-
regions. The report shows the region's amplification of climate extremes, which poses
considerable concerns for water resource management. The established ANN technique
could be used for future TWSA predictions and monitoring of climatic extremes. The
research provides important insights for flood and drought monitoring in the Yun-Gui
Plateau and other areas.
6. Soni & Syed (2015) explored Terrestrial Water Storage (TWS) differences in significant
Indian river basins in the reviewed literature. Based on GRACE satellite data and
hydrologic observations, their study found decreasing groundwater in the Ganga basin and

10
 rising TWS in the Godavari and Krishna basins. According to the study, these variations
were attributable to the El Nio-Southern Oscillation (ENSO) pattern and human-induced
variables. Their findings highlight the critical impact of climate variability and human
intervention on water resources, highlighting the importance of long-term management
techniques in Indian river basins.
7. Hosseini-Moghari et al. (2019) introduced a new approach called MTSDI to effectively
monitor drought using GRACE observations, especially for basins with high water
consumption. The results show that MTSDI performs better than traditional TSDI and has
good potential for use in drought monitoring systems. The traditional GRACE-based
drought index called the Total Storage Deficit Index (TSDI) incorrectly classifies the early
years as wet and later years as dry. This is because the GRACE signal is affected by human
water consumption, which shows a decreasing trend. To overcome this, they proposed the
MTSDI approach, which removes long-term and periodic behaviour from the GRACE
signal using signal decomposition. Then, the residuals of the signal are used instead of the
total signal to calculate the index.
8. Liu et al. (2022) discuss the accelerating depletion of groundwater in California's Central
Valley, particularly during a period of a megadrought. It emphasizes the importance of
effective groundwater management to ensure its availability during increasingly intense
droughts. The study uses nearly two decades of observations from NASA's GRACE
satellite missions to demonstrate the accelerating rate of groundwater depletion. It also
highlights the challenges and potential utility of GRACE/FO-derived groundwater storage
changes for supporting regional groundwater management efforts. The document also
addresses the need for comprehensive groundwater models considering human activities
such as groundwater pumping, irrigation, and land subsidence. Additionally, it discusses
the impact of climate change and changing hydrologic extremes on groundwater resources.
The study provides valuable insights into the critical role of groundwater in freshwater
supplies for human life, agricultural production, and ecosystem processes. It emphasizes
the need for sustainable groundwater management practices.
9. Rawat et al. (2022) analyse drought characteristics in India at sub-basin scales using
observed data, specifically focusing on the spatio-temporal changes in the GRACE-DSI
(GRACE-Drought Severity Index) and CCDI (Combined Climatologic Deviation Index)
drought indices. The study aims to evaluate trends in data, identify regions with significant
trends, and determine the magnitude of change using statistical tests. Additionally, the
study seeks to characterize drought events based on the identified indices and understand
whether drought is due to a deficit in precipitation, terrestrial water storage, or both. The
ultimate goal is to contribute to understanding drought at the sub-basin spatial scale and
provide valuable insights for drought management and mitigation efforts.
10. M. Liu et al. (2022) discuss the dynamics of groundwater storage in the Taihang Mountain
Region (TMR) of China from 2003 to 2016. It focuses on the relative contributions of
climate and human activities to changes in groundwater storage and the establishment of a
new index to evaluate groundwater drought. The study reveals that human activities, such
as revegetation, damming, and agricultural irrigation, have dominated groundwater storage
decreases, with a contribution rate of 96% relative to climate change. It also highlights the
increasing severity of groundwater scarcity in the TMR and emphasizes the urgent need for
water management interventions. The findings provide valuable insights into the complex
interactions between human activities, climate change, and groundwater dynamics, with

11
 implications for water resource management and understanding groundwater operation
mechanisms.

3.4 Drought Monitoring Background

Several decades of research have yielded different definitions for drought (Wilhite, 1996).
Although they are all connected, droughts can vary by length and affect local resources
differently. In general, a meteorological drought is an abnormal period with low or no
precipitation. Hydrological drought deals with the effects of precipitation working through
reservoirs, streamflow, and groundwater. Agricultural drought is how crops respond to
increased heat stress and lack of water availability in the soil. Finally, socioeconomic drought
is associated with the economic supply and demand of goods, such as water and agricultural
products, which are heavily affected by meteorological, hydrological, and agricultural droughts
(Donald A. Wilhite & Mickey Glantz, 1985). All these drought types have water in common;
thus, monitoring water is a pivotal aspect of study for these sectors (L.M. Tallaksen & Henny
Van Lanen, 2004).
Drought monitoring using objective and subjective assessments of weather, hydrology,
agriculture, and human responses is an important part of the goal of successfully mitigating
and responding to drought effects. The widespread use of remote sensing systems in acquiring
meteorological, hydrological, and vegetation health data allows multiple high spatial resolution
and multi-faceted resources to quantify and respond to drought.
The first quantitative drought indices appeared early in the 20th century as Munger’s and
Kincer’s indices (R. R. Heim, 2002). These indices measured the period of time without a
specific amount of precipitation. Because precipitation is a highly variable quantity, any fixed
amount of precipitation would not be sufficient for all regions. The 4 mid-20th century saw
more drought indices evolve to include more than just precipitation and specifically analyzed
variables necessary for agricultural and hydrological impacts (R. R. Heim, 2002). In 1965,
Palmer developed the Palmer Drought Severity Index (PDSI), which accounts for temperature
and precipitation in a water balance model (Wayne C. Palmer, 2006). This index effectively
identified long-term droughts and accounted for several previously ignored variables.
However, it lacked a high degree of comparability between regions and did not account for
snow or ice. The next largest innovation in drought monitoring was in 1993, with the creation
of the Standardized Precipitation Index (SPI), which determines precipitation surpluses or
deficits in anomalies from normal, allowing uniform calculation in different regions (Mckee et
al., 1993). While SPI deals very well with meteorological drought, it has limitations in
identifying hydrological and agricultural droughts (World Meteorological Organization 2012).
Climate-based drought indices are also based on weather station data (sometimes interpolated
to a uniform grid), which are far less dense in remote areas. In contrast, satellite-based indices
have continuous, equal coverage of the entire area of interest. Modelling and remote sensing
have recently become driving forces in drought monitoring with their ability to look at the
large-scale effects of drought. The Normalized Difference Vegetation Index (NDVI) was one
of the first to use remotely sensed imagery as a drought tool (Rouse et al., 1974).

12
3.5 Drought in India
Lack of moisture is a severe weather phenomenon known as drought. Drought brought on by
humans, agriculture, hydrology, and meteorology. In India, droughts can occur in areas with
heavy and infrequent rainfall. Drought is a potential hazard across roughly 68% of the country.
35% of the regions are at risk of drought if they receive rainfall between 750mm and 1125 mm,
and 33% if they receive less than 750 mm. The Himalayan region's water shortages are also
not unique. India has already experienced severe droughts that have affected the country's water
supplies, agricultural output, economic growth, and standard of living (A. K. Mishra & Singh,
2010) . Millions of people in India died from food shortages brought on by some of the country's
most severe droughts (V. Mishra et al., 2019). The Indo-Gangetic plain and Maharashtra region
experienced a drought in 2015, impacting crop productivity and water availability (V. Mishra
et al., 2016). Significant groundwater depletion occurred in the southern states and Indo-
Gangetic plains of India during the 2015–2016 drought, which affected over 330 million people
across 10 states (Prasad et al., 2016). During the droughts of 2004, 2009, and 2012, there was
a particularly acute shortage of terrestrial water storage, with the Ganga basin peaking in 2009
(Goldin, 2016). Thus, correctly analysing and executing appropriate mitigation strategies and
monitoring, assessing, and quantifying drought is imperative. Using the Integrated Drought
Vulnerability Index, V. K. Jain et al. (2015)measure vulnerability to drought.

3.6 Drought Monitoring Techniques: A Comprehensive Review of Satellite-

Derived Indices
Numerous drought monitoring indices have been established due to the complicated nature of
drought and its dependence on multiple hydrological parameters. Most of the drought indices
like Palmer Drought Severity Index (PDSI) (Wayne C. Palmer, 2006), Standardized
precipitation index (SPI) (Mckee et al., 1993), Standardized Precipitation Evapotranspiration
Index (SPEI) (Vicente-Serrano et al., 2010), Standardized Runoff Index (Shukla & Wood, 2008),
Standardized Drought Index(Nalbantis & Tsakiris, 2009), are based on model simulation or
observation of land surface states and fluxes while Normalized Difference Vegetation Index
(Rouse et al., 1973), Vegetation Drought Response Index (Brown et al., 2008) based on remote
sensing techniques but are restricted to show agricultural drought only. PDSI calculations have
limits for different climatic situations because they are based on the Mid-Western climate in
the United States. The multi-time scale features displayed by SPI and SPEI make it challenging
to select a suitable timeline for characterising droughts. Mckee et al. (1993) developed the
Standardised Precipitation Index to characterise droughts using precipitation data. Due to its
multi-time scale properties, SPI can characterise droughts at various time scales. A new
climatic drought index, the standardised Precipitation Evapotranspiration Index (SPEI), was
created by Vicente-Serrano et al. (2010). SPEI accurately characterises drought by using
temperature and precipitation data. Because temperature data is included in SPEI, drought
characteristics are depicted more accurately than SPI and are mathematically identical to SPI.
When (V. K. Jain et al., 2015) analysed several drought indices, including the Rainfall Decile
Drought Index, SPI, Effective Drought Index, China Z-Index, and Rainfall Departure, they
concluded that the EDI has a stronger correlation with the other indices across all-time scales.
With the advancement of the GRACE, direct observation of Terrestrial Water Storage (TWS)
or its deviation from the climatological mean can now serve as a metric for characterizing

13
drought. Widely used GRACE drought indices for this purpose include the GRACE-Drought
Severity Index (GRACE-DSI), Water Storage Deficit Index (WSDI), Combined Climatologic
Deviation Index (CCDI), GRACE-Based Groundwater Drought Index (GGDI) and
Groundwater drought potential index (GDPI).
GDPI utilizes both Precipitation and GRACE-TWSA data along with GLDAS data, while
GRACE_GGDI uses only GRACE-TWSA data and GLDAS data only (Kumar et al., 2021; M.
Liu et al., 2022). The GDPI encompasses all facets of drought occurrences, including
meteorological, agricultural, hydrological, and anthropogenic activities. While extensive
research has been conducted on drought analysis and characterization at the basin scale, studies
focusing on temporal changes and characterization at the sub-basin scale are limited. Therefore,
this study aims to investigate spatio-temporal changes in GRACE-GDPI and characterize
drought using drought indices at the sub-basin spatial scale.

CHAPTER 4. DATASETS AND METHODOLOGY

14
4.1 Data Processing and Descriptions
This study has used several datasets gathered from ground-based and spaceborne sources, as
listed in Table 3. These datasets were chosen based on accuracy, dependability, and
availability.
Table 3: The study makes use of the following datasets
No. Dataset Source of Data Period Unit Spatial Temporal
Data Model Resolution Resolution
1 GRACE_CS NASA Earth CSR 2003- cm 0.5 ˚×0.5 ˚ Monthly
R_Mascon Data v1.0 2022
TWS (GRACE RL06
Mission,
GRACE-FO
Mission)
2 Soil Moisture Global Land NOAH 2003- kg/m2 0.25˚×0.25 ˚ Monthly
Storage Data v3.6 2022
(SMS) Assimilation
System
(GLDAS)
3 Canopy Global Land NOAH 2003- kg/m2 0.25˚×0.25 ˚ Monthly
Water Data v3.6 2022
Storage Assimilation
(CWS) System
(GLDAS)
4 Snow Global Land NOAH 2003- kg/m2 0.25˚×0.25 ˚ Monthly
Melting Data v3.6 2022
Water Assimilation
(SMW) System
(GLDAS)
5 Surface Global Land NOAH 2003- kg/m2 0.25˚×0.25 ˚ Monthly
Water Data v3.6 2022
Storage Assimilation
(SWS) System
(GLDAS)
6 Rainfall India Version 1985- mm 0.25˚×0.25 ˚ Monthly
Meteorologic 4.1 2022
al
Department
(IMD)

4.2 Retrieval of Groundwater Storage Change

This study used remote sensing and ground-based integrated observation to measure the
variation in groundwater storage across the study area. Based on a temporal mean baseline
spanning all months from 2004 to 2009, the Terrestrial water storage anomalies (TWSA) are
computed using CSR RL06 GRACE/GRACE-FO mascon solutions version 01, with a spatial
resolution of 0.5°×0.5°. Groundwater storage anomalies (GWSA) at any given time (t) were

15
determined by subtracting soil moisture storage anomalies (SMSA), snow melting water
anomalies (SMWA), surface water storage anomalies (SWSA) and canopy water storage
anomalies (CWSA) from the TWSA derived from GRACE observations (Kumar et al., 2021).
The calculation of soil moisture storage anomaly (SMSA) for the NOAH land surface model
was performed using the following equation:

𝑆𝑀𝑆𝐴 = 𝑆𝑀𝑆 − 𝑆𝑀𝑆 (1)

Here, SMSAt represents the soil moisture storage anomaly at time t, SMSt denotes the soil
moisture at time t, and 𝑆𝑀𝑆 signifies the average soil moisture with respect to the
baseline period from January 2004 to December 2009, which aligns with the GRACE terrestrial
water storage data.
Likewise, the snow melting water anomalies (SMWA) were determined for the NOAH land
surface model utilizing the following equation:

𝑆𝑀𝑊𝐴 = 𝑆𝑀𝑊 − 𝑆𝑀𝑊 (2)

Here, SMWAt represents the snow melting water anomaly at time t, SMWt denotes the snow
melting water at time t, and 𝑆𝑀𝑊 signifies the average snow melting water with
respect to the baseline period from January 2004 to December 2009, which aligns with the
GRACE terrestrial water storage data.
Likewise, the calculation of surface water storage anomaly (SWSA) for the NOAH land
surface model was performed using the following equation:

𝑆𝑊𝑆𝐴 = 𝑆𝑊𝑆 − 𝑆𝑊𝑆 (3)

Here, SWSAt represents the surface water storage anomaly at time t, SWS t denotes the surface
water storage at time t, and 𝑆𝑊𝑆 signifies the average surface water storage with
respect to the baseline period from January 2004 to December 2009, which aligns with the
GRACE terrestrial water storage data.
Likewise, the calculation of canopy water storage anomaly (CWSA) for the NOAH land
surface model was performed using the following equation:

𝐶𝑊𝑆𝐴 = 𝐶𝑊𝑆 − 𝐶𝑊𝑆 (4)

Here, CWSAt represents the canopy water storage anomaly at time t, CWSt denotes the canopy
water storage at time t, and 𝐶𝑊𝑆 signifies the average canopy water storage with
respect to the baseline period from January 2004 to December 2009, which aligns with the
GRACE terrestrial water storage data.
Finally, the GWSA is calculated as follows:
𝐺𝑊𝑆𝐴 = 𝑇𝑊𝑆𝐴 − 𝑆𝑀𝑆𝐴 − 𝑆𝑀𝑊𝐴 − 𝑆𝑊𝑆𝐴 − 𝐶𝑊𝑆𝐴 (5)

4.3 Estimation of GRACE Groundwater Drought Potential Index (GDPI)

Groundwater drought is a distinct type characterized by a sequence of events: initially, there is
a reduction in groundwater recharge, followed by declines in groundwater storage (GWS) and
discharge. Prior research has established and utilized groundwater drought indices based on

16
normalized values of deseasonalized groundwater storage anomalies (GWSA). However,
unlike the flood potential index, these indices reflect the extent of the GWS deficit and cannot
capture the intrinsic connection between GWS and precipitation (Thomas et al., 2017).
Expanding upon earlier investigations of Reager & Famiglietti (2009) and Thomas et al.( 2017),
who developed the GRACE-Groundwater Drought Index (GGDI), which does not incorporate
the precipitation for their estimation of the drought index, whereasM. Liu et al.(2022)
formulated a new groundwater drought potential index (GDPI). This GDPI is designed to
incorporate precipitation information into the evaluation of groundwater drought, offering a
comprehensive reassessment of groundwater drought potential.
The GDPI strives to standardize the magnitude of drought potential and quantify the likelihood
of groundwater drought occurrences. The step-by-step procedures are as follows:
1. Deseasonalized GWSA: This process involved deriving GWSAs from GRACE data for a
specific month and subtracting the average GWSAs for that month across the study period.
The goal was to create a continuous time series of GWSA that was unaffected by seasonal
variations.

GWSA = 𝐺𝑊𝑆𝐴 − 𝐺𝑊𝑆𝐴 (6)

Where i= 1,2,3,……12 shows all 12 months, and n denotes the length of the total year of
the study period. In this study, n is 20 years (240 months) from January 2003 to December
2022. For the deseasonalized GWSA time series, they established the historical maximum
as the optimal GWS status.
2. Determining the highest groundwater recharge demand: The optimal groundwater
status was established by setting the historical maximum of the deseasonalized GWSA time
series. The discrepancy between the monthly deseasonalized GWSAs and this maximum
value represented the highest groundwater recharge demand, indicating the required
recharge to return groundwater drought to its optimal condition. here, the highest
groundwater recharge demand can be made by following equation:

GWCD = max (𝐺𝑊𝑆𝐴 , ) − 𝐺𝑊𝑆𝐴 , (7)

Where GWCDt denotes the highest groundwater charge demand in month t, signifying the
volume of water required to replenish the monthly GWSA to its historically optimal level.
GWSAdeseasonalized,t represents the deseasonalized GWSA in month t.
3. Assessing Precipitation Recharge Adequacy: If precipitation surpassed the highest
groundwater recharge demand, the likelihood of groundwater drought was nearly zero.
Conversely, insufficient precipitation compared to the demand indicated a lower potential
for groundwater recovery to the optimal level, increasing the risk of groundwater drought.
We assumed a consistent runoff pattern for precipitation, considering it the ideal source for
groundwater recharge. To assess if the optimal recharge potential could surpass the highest
demand in any given month, we formulated the drought potential calculation as follows:

DPA = 𝐺𝑊𝐶𝐷 − 𝑃 (8)

17
 where DPAt represents the drought potential amount in month t, and Pt indicates the
precipitation in month t. A negative value for DPAt (DPAt <0) suggests minimal potential
for groundwater drought occurrence. Conversely, a positive DPA t (DPAt >0) indicates a
higher likelihood of groundwater drought, significantly when it exceeds zero, signifying
lower precipitation relative to the groundwater recharge demand.
4. Standardization of Drought Potential Amount for GDPI Calculation: In summary, the
GDPI assesses the likelihood of groundwater drought based on whether precipitation fulfils
the groundwater recharge demand, leading to the recovery of GWSAs to their historical
maximum level. Normalization of GDPI Based on Drought Potential Amount as follows:
𝐷𝑃𝐴
𝐺𝐷𝑃𝐼 = (9)
min(𝐷𝑃𝐴 )

GDPIt represents the GDPI derived from GRACE data for month t.

Interpreting GDPI Values: A GDPI<0 indicates lower precipitation than the groundwater
recharge demand, leading to greater groundwater depletion and an increased likelihood of
groundwater drought. Conversely, a GDPI > 0 suggests minimal potential for groundwater
drought as precipitation levels are sufficient to restore groundwater to its historically
optimal state.

4.4 Mann Kendall Trend test

The Mann-Kendall test, a commonly used nonparametric method for trend analysis, is widely
employed worldwide. Nevertheless, it is crucial to consider that the persistence of
hydrometeorological datasets may impact the results of this test. Hamed & Rao, (1998)
introduced the Mann-Kendall test to address autocorrelation, offering a more consistent and
robust method for identifying trends in time series data.
A trend is a consistent and systematic movement in a time series through which values are, on
average, increasing or decreasing. Mann-Kendall tests the presence of rising, falling or no trend
in any series.
𝑆= ∑ ∑ 𝑠𝑔𝑛(𝑥 − 𝑥 ) (10)

1 𝑖𝑓 𝑥 > 𝑥
sgn (𝑥 − 𝑥 ) = 0 𝑖𝑓 𝑥 = 𝑥 (11)
−1 𝑖𝑓 𝑥 < 𝑥

For n≥ 8, S is normally distributed as

( )( )
Mean, 𝐸(𝑆) = 0 and Variance, 𝑉𝑎𝑟(𝑆) = (12)
/
(𝑠 − 1)/(𝑣𝑎𝑟(𝑠)) 𝑠>0
0 𝑠=0
𝑍= (13)
𝑠<0
( )

18
Z is standard normally distributed with zero mean and unit variance, + ve S denotes an upward
trend and - ve S downward trend (Hamed & Rao, 1998).

4.5 Standardized Precipitation Index

The Standardized Precipitation Index (SPI) was developed by American scientists McKee,
Doesken, and Kleist in 1993 to provide a simple, statistically relevant, and flexible drought
index. It requires only input precipitation data and can analyse wet and dry periods effectively.
While ideally, at least 20-30 years of monthly data is needed for optimal results, the program
can handle missing data, albeit with reduced confidence. Climatologists prefer complete data
sets, but data may be incomplete, necessitating estimation techniques. Users must consider the
trade-off between missing data tolerance and the accuracy of SPI calculations and analyses,
aiming to minimize the use of estimated data for more reliable results.
The Standardized Precipitation Index (SPI) quantifies precipitation deficits across multiple
timescales, reflecting their impact on water resources. Originally calculated for timescales of
3, 6, 12, 24, and 48 months, the SPI normalizes precipitation data to a zero mean, uniformly
representing wetter and drier climates. Negative SPI values indicate less than median
precipitation, while positive values indicate greater than median precipitation. Drought
intensities are classified based on SPI values, with drought events defined as continuous
negative SPI values reaching an intensity of -1.0 or less. Each drought event is characterized
by its duration and intensity, with the sum of negative SPI values representing its magnitude.
The SPI's standardization enables the assessment of drought rarity and the probability of
precipitation required to end it, facilitating comparisons of historical and current droughts
across different climatic and geographic locations.

Table 4: SPI values

+2.0 extremely wet

1.5 to 1.99 very wet

1.0 to 1.49 moderately wet

-0.99 to +0.99 near normal

-1.0 to -1.49 moderately dry

-1.5 to -1.99 severely dry

-2.00 and less extremely dry

19
CHAPTER 5. RESULTS

20