You are on page 1of 17

Received: 15 December 2020 Revised: 23 March 2021 Accepted: 3 April 2021

DOI: 10.1002/ldr.3965

RESEARCH ARTICLE

Monitoring land sensitivity to desertification using the ESAI


approach and evaluation of the key indicators:
A spatio-temporal study in India

Joy Rajbanshi1 | Sharmistha Das2

1
Department of Geography, Vivekananda
College for Women, University of Calcutta, Abstract
Kolkata, India Desertification is one of the primary environmental problems that negatively affect
2
Department of Geography, Presidency
agricultural production leading to poverty, hunger, and economic instability in a coun-
University, Kolkata, India
try like India. A holistic study is therefore extremely valuable rather than localized
Correspondence
ones for addressing this issue and its' further consequences. In this study, the envi-
Joy Rajbanshi, Department of Geography,
Vivekananda College for Women, University ronmentally sensitive area index (ESAI) has been used to identify the areas vulnerable
of Calcutta, Kolkata, India.
to desertification from 1992 to 2015 in India. In addition to this, the random forest
Email: jrgeog_rs@caluniv.ac.in
(RF) model was used to identify the drivers with major influence over the phenome-
non. The result revealed that the desertification process has extended from the west
and spread over the southern region in the Country. Among the states of India, Rajas-
than and Ladakh possess the highest mean ESAI values (1.5–1.7), where 87.61% and
83.83% of land, respectively, are critically degraded due to desertification. According
to the Mann-Kendall mutation point detection test, the years of 1996, 2009, and
2014 were identified as the abrupt change of desertification processes over the
Country. The most severe form of degradation is observed in the period between
1997 and 2009, where the highly sensitive (Critical-3) area increased up to 37.44%.
Based on variable importance for the RF-model, drought resistance has the highest
importance among all the other inputs that is, 46% followed by erosion protection
with 20%. Our study also points out that 33.76% of the land of the Country is still
classed under a very high critical zone and without any immediate action; it may
result in irreversible loss.

KEYWORDS
change-year detection, desertification, environmental sensitivity area index, random forest

1 | I N T RO DU CT I O N degradation (Darbalaeva, Mikheeva, & Zhamyanova, 2020;


Safriel, 2009). Currently, desertification has been occurring all over
Desertification is commonly a slow-paced and often unnoticed pro- the world affecting 110 countries with higher concentrations in
cess that eventually poses a severe threat for the arid, semiarid, and mostly the developing countries in Africa, Asia, and South America
sub-humid regions across the world (Akbari, Jafari, Memarian, & (Hu, Han, & Zhang, 2020; Kannan, 2014). This is expected to raise
Gholami, 2020; Wang, Pan, Wang, Shen, & Lu, 2013) with its' influ- poverty and inequality resulting in rapid deterioration of the environ-
ence on the water availability (Emadodin, Reinsch, & Taube, 2019; ment along with societies and their political and economic structure
Safriel, 2009) and hence, retarding land productivity which in due (UNCCD, 1994). India has not escaped this fate either. According to
course, results in environmental as well as socio-economic Singh, Santra, Kumar, Panwar, &Meghwal (2016) and Kar &

Land Degrad Dev. 2021;1–17. wileyonlinelibrary.com/journal/ldr © 2021 John Wiley & Sons, Ltd. 1
2 RAJBANSHI AND DAS

Kumar (2020), around 12%–14% of land in India is under an arid cli- However, the latter work are highly localized and scattered in distribu-
matic zone containing both hot (Thar Desert) and cold regions tion andconsideronly a few factors (mostly anthropogenic activities)
(Ladakh), which are experiencing degradation of land and vegetation responsible for controlling the desertification processed so fail to pro-
in varying degrees due to climate change and encroaching desertifica- vide a broad view of the entire Country. There is a serious lack of
tion processes. However, Dasgupta, Sastry, Dhinwa, Rathore, & Nat- studies performed over the entire region of India limits our knowledge
hawat (2013) and Christian & Dhinwa (2018) are of the opinion that about the true nature of degradation. In order to understand the pro-
this process is highly influenced by human activities. gression of the risk of desertification in India, a holistic study is
Scholars all over the world have identified key factors responsible required. Our work, aimed to produce an ESAI that considered a range
for this global peril such as natural vegetation and climatic factors of indicators encompassing both natural and man-made to identify
(Bezerra et al., 2020; Sun, Gao, & Li, 2014; Zhang & Huisingh, 2018). the areas vulnerable to desertification according to the degrees of risk
Several studies have stressed the importance of the normalized differ- they are exposed to. Our study also seeks to rank the drivers and to
ence vegetation index (NDVI), albedo and net primary productivity identify those with major influence over the phenomenon. To our
(NPP) as the most direct indicators of desertification (Jiang, Jiapaer, knowledge, our work is the first to successfully investigate the extent
Bao, Li, & Guo, 2019; Vorovencii, 2017), but these take under consid- of the desertification process in the entire subcontinent of India using
eration only a few of the factors operating or merely individual ones modern and more reliable methods.
(Kundu, Patel, Saha, & Dutta, 2017; Singh et al., 2016; Sun
et al., 2014) when identifying the areas affected by desertification,
underestimating the possibility of a graver scenario that may arise due 2 | M A T E R I A L S A N D M ET H O D S
to the combined effects of those individual factors. Therefore,
present-day researches focus on studying and expanding the range of 2.1 | Study area
key factors (both natural and anthropogenic) and investigating the
repercussions of their combined influence. How land is influenced by India has an area of 3,287,263 km2, with the Indian Ocean in the
these factors gave rise to the concept of environmental sensitivity south, the Arabian Sea in the southwest, and the Bay of Bengal in
which was adopted as a major element in the MEDALUS the southeast; the north is bordered by the Himalayan Mountains and
(Mediterranean desertification and land use) Project initiated by the the Thar Desert in the west. The western region of the Country has
Europeans exploring environment and climate research in 1999 an arid and semiarid climate, the north and northeast has a cool cli-
(Kosmas, Kirby, & Geeson, 1999; Kosmas et al., 2014; Lee mate , while the rest of the Country is humid tropical or humid sub-
et al., 2019). The scholars investigating the desertification process in tropical. The general temperature of our study region is around 10 C
the Mediterranean region have mainly adopted the MEDALUS in winter and 40 C in summer. The Indian political territory covers the
model as it offers the researchers an ability to consider a variety of Andaman and Nicobar Islands and Lakshadweep, but due to difficulty
climate, vegetation and soil, as well as anthropogenic, indicators in obtaining data and fitting it into our work, these parts have been
which enable a holistic approach in assessing desertification (Budak excluded in our study, and instead, only the main landmass of the sub-
et al., 2018; Jafari & Bakhshandehmehr, 2013; Ouachoua & continent has been considered (Figure 1). The locational setting along
Karkouri, 2020; Uzuner & Dengiz, 2020; Zakerinejad & with climate change and human interferences have resulted in severe
Masoudi, 2019). However, the scoring framework was later updated degradation and desertification countrywide (Dharumarajan
by Ferrara et al. (2020) to allow the method to be adjusted for other et al., 2018). India is home to the second-largest population in the
climatic and social situations. Our work attempts to analyze the world and with primary activities as the backbone of its' economy and
desertification scenario in India and also pursues the ESAI approach society it is indeed a grave situation.
adopting this updated scoring system.
Before our study, few attempts have been carried out to inspect
the desertification process in India, (Christian & Dhinwa, 2018; 2.2 | ESAI approach
Dasgupta et al., 2013; Kundu et al., 2017; Moharana et al., 2016) and
those are mostly concentrated in different parts of Rajasthan where The present work follows the ESAI framework, which takes into
desertification is most pronounced. Some 75% of lands in western account several indicators of desertification sensitivity as major qual-
Rajasthan and around 10%–35% lands in the eastern parts are cur- ity indices. For this work, a total of 14 indicators were taken under
rently being highly degraded and are prone to water erosion. Saji & four major quality indices namely soil, climate, vegetation, and land
Jayakumar (2017), Dharumarajan, Bishop, Hegde, & Singh (2018) and management (Tables 1–4). These indicators were selected according
Kumar, Babu, Rajasekhar, & Ramachandra (2020) have portrayed a to their importance in analyzing/ identifying the process and severity
few individual scenarios from the southern states in their work, or of land degradation in the region. The indicators were given scores
have depicted the spread of desertification from the western states. ranging from 1 to 2 based on their respective roles in influencing the
The Ladakh region has recently caught the attention of researchers process of desertification and reflecting the sensitivity of the land to
because it presents cases of severe desertification (see Kar & those processes, with score of 1 for lowest sensitivity and a score of
Kumar, 2020; Tewari, Pareek, Raghuvanshi, Kumar, & Roy, 2016). 2 for highest. The process of scoring the indicators was performed by
RAJBANSHI AND DAS 3

F I G U R E 1 Location of the Indian landmass with states and union territories (UTs); (1) Arunachal Pradesh, (2) Assam, (3) Chandigarh,
(4) Karnataka, (5) Manipur, (6) Meghalaya, (7) Mizoram, (8) Nagaland, (9) Punjab, (10) Rajasthan, (11) Sikkim, (12) Tripura, (13) Uttarakhand,
(14) Telangana, (15) Bihar, (16) Kerala, (17) Madhya Pradesh, (18) Gujarat, (19) Odhisa, (20) Dadra and Nagar Haveli and Daman and Diu,
(21) Ladakh, (22) Jammu and Kashmir, (23) Chhattisgarh, (24) Delhi, (25) Goa, (26) Haryana, (27) Himachal Pradesh, (28) Jharkhand, (29) Tamil
Nadu, (30) Uttar Pradesh, (31) West Bengal, (32) Andhra Pradesh, (33) Puducherry, (34) Maharashtra [Colour figure can be viewed at
wileyonlinelibrary.com]

following the updated scoring system of the ESAI framework by Fer- 2.2.1 | Soil quality index
rara et al. (2020) which was upgraded to better consider the deserti-
fication scenario of the entire world. Assisted by the previous works The indicators used in this work under the soil quality index (SQI) are
of Kosmas et al. (1999, 2014), all the indices were then calculated slope, texture, salinity, drainage, and topsoil clay fraction (Table 1). For
from the geometric mean of the scored indicators under each index. acquiring the slope, shuttle radar topography mission (SRTM) digital
After that, the ESAI was calculated from the values of the four elevation model (DEM) was used. The soil data were collected from
indices. the Food and Agricultural Organization Harmonized World Soil
4 RAJBANSHI AND DAS

TABLE 1 Description of the indicators and scores of soil quality index (SQI)

Quality Ranges and


index Indicators characteristics Scores Description Data source References
SQI Texture clay, loam, sandy clay 1 Increase in the amount of Food and Agricultural Ferrara et al. (2020)
loam, sandy loam, sand in soil may lead to Organization
loamy sand loss of nutrients and Harmonized World Soil
silty clay loam, silty 1.2 can therefore, add to Database
loam, sandy clay the desertification
process
clay (heavy), silty 1.6
clay, clay (light), silt
sand 2
Electrical <2 1 Abnormal amount of salt Food and Agricultural Kosmas et al. (2014)
conductivity 2<4 1.3 in soil can affect the Organization
(dS m1) natural vegetation and Harmonized World Soil
4<8 1.5 thus lead to Database
8<15 1.8 desertification
>=15 2
Topsoil clay >= 25 1 The topsoil layer holds Food and Agricultural Jiang, Jiapaer, Bao, Li, &
fraction 10<25 1.3 the most part of Organization Guo (2019), Kosmas
(in %) organic matter and Harmonized World Soil et al. (1999)
< 10 2 other nutrients, the loss Database
of which leads erosion
and land degradation
Drainage well 1 One of the major features Food and Agricultural Ferrara et al. (2020)
moderately well 1.2 of desertification is Organization
acute deficiency of Harmonized world Soil
imperfectly 1.4 water Database
somewhat excessive 1.7
poor; very poor; 2
excessive
Slope (in %) <3 1 Slope has a great EARTH EXPLORER, Ferrara et al. (2020)
3<6 1.1 influence on the USGS
process of erosion
6 < 12 1.2
12 < 18 1.3
18 < 24 1.4
24 < 30 1.5
30 < 36 1.7
> = 36 2

Database with a resolution of 30 arc-second (http://www.fao.org/ Han, Sun, & Liu, 2009) and natural vegetation cover over land
soils-portal/data-hub/soil-maps-and-databases/harmonized-world- (Higginbottom & Symeonakis, 2014; Kéfi et al., 2007). In this work,
soil-databasev12/en/). After collecting all the data, scoring was per- the climate quality index (CQI) consists of three indicators namely
formed based on the role of each indicator in causing land sensitivity precipitation, aridity index, and the slope aspect (Table 2). The
to desertification. The SQI was obtained by calculating the geometric gridded precipitation (0.25  0.25 ) and temperature (1  1  ) data
mean of all the five indicators [Equation (1)]: were collected from the Indian Meteorological Department (IMD).
The aspect was obtained from the SRTM DEM. The aridity index
SQI ¼ ðtexture  salinity  drainage  top soil clay fraction  slopeÞ1=5 : was calculated from the precipitation and potential evapotranspi-
ð1Þ ration following the method adopted by the United Nations Envi-
ronment Programme (UNEP) (Boschetto, Mohamed, &
Arrigotti, 2010).

2.2.2 | Climate quality index AI ¼ P=PET, ð2Þ

Climatic indicators are very important in the identification and Where: AI is Aridity Index, P is average annual precipitation and PET is
analysis of the desertification process (Sivakumar, 2007; Xue, Guo, potential evapotranspiration.
RAJBANSHI AND DAS 5

TABLE 2 Description of the indicators and scores of climate quality index (CQI)

Quality Ranges and


index Indicators characteristics Scores Description Data source References
CQI Precipitation > = 650 1 Precipitation is very important Indian Ferrara et al. (2020)
(mm yr1) 570<650 1.05 for maintaining the water Meteorological
balance and growth of biomass Department (IMD)
490<570 1.15 in soil, the lack or irregularity
440<490 1.25 of which may lead to
390<440 1.35 desertification

345<390 1.5
310<345 1.65
280<310 1.8
< 280 2
Aridity index > = 1.00 1 Aridity points toward the Indian Ferrara et al. (2020)
0.75<1.00 1.05 dryness of the climate that Meteorological
influence the vegetation and Department (IMD)
0.65<0.75 1.15 water content directly
0.50<0.65 1.25
0.35<0.50 1.35
0.20<0.35 1.45
0.10<0.20 1.55
0.03<0.10 1.75
< 0.03 2
Slope aspect NW – NE 1 Slope aspect determines the EARTH EXPLORER, Jiang, Jiapaer, Bao, Li, &
SW - SE 2 amount of insolation received USGS Guo (2019), Kosmas
by a parcel of land and thus et al. (1999)
influences the climate

The PET has been calculated from the equation given by Har- having a 300 m resolution (https://e4ftl01.cr.usgs.gov/MEASURES/
greaves & Samani (1985) as follows: VCF5KYR.001/) were used to calculate the first three indicators. For
natural vegetation cover, the continuous fields version 1 (VCF5KYR)
PET ¼ 0:0135  KPET  RA  ðTmax  Tmin Þ0:5  ðTavg þ 17:8Þ, ð3Þ data product was used which is derived from the long-term records of
AVHRR (advanced very high-resolution radiometer) observations
Where: KPET is the calibration coefficient (0.17), RA is the extrater- (https://lpdaac.usgs.gov/products/vcf5kyrv001/). After scoring each
restrial radiation (MJ m2 day1), Tmax and Tmin are the maximum of the indicators according to their role in land sensitivity, the VQI
and minimum air temperature (  C) and Tavg is the average air tem- was obtained using Equation (5):
perature ( C).
Afterward, the CQI was calculated using Equation (4): VQI ¼ ðdrought resistance  erosion protection  fire risk

natural vegetation coverÞ /4 :


1
ð5Þ
CQI ¼ ðprecipitation  aridity index  slope aspectÞ1=3 : ð4Þ

2.2.3 | Vegetation quality index 2.2.4 | Management quality index

Vegetation cover plays a major role in the evaluation of the progress Desertification is considered as the outcome of anthropogenic exploi-
of the desertification process as it has the capacity to counter the pro- tation of resources as much as it is thought as the result of natural
cess (Abdollahi, 2019; Sarparast, Ownegh, & Sepehr, 2020). In this _
atmospheric phenomena (Imamoglu & Dengiz, 2019). Therefore, criti-
work, four indicators were assessed for calculating the vegetation cal consideration of the anthropogenic activities must be taken into
quality index (VQI), that is, drought resistance, erosion protection, fire account while investigating desertification (Avni, Porat, Plakht, &
risk, and natural vegetation cover (Table 3). The CCI (climate-change Avni, 2006; Jiang et al., 2019). Hence, two important indicators have
initiative) Land Cover products of the European Space Agency (ESA) been used for calculating the management quality index (MQI), which
6 RAJBANSHI AND DAS

TABLE 3 Description of the indicators and scores of vegetation quality index (VQI)

Quality
index Indicators Scores Data source and references
VQI Land use land cover Drought Fire Erosion Data collected from CCI
resistance risk protection (Climate Change Initiative)
Cropland rainfed 1.5 1.4 1.7 Land Cover products,
European Space Agency
Herbaceous cover 1.6 1.4 1.8 (ESA). Scoring system
Tree or shrub cover 1.3 1.2 1.6 adopted from Ferrara et al.
Cropland irrigated or post-flooding (herbaceous cover) (<50%) 1.4 1.4 1.4 (2020)

Mosaic cropland (>50%)/natural vegetation (tree shrub 1.5 1.5 1.5


herbaceous cover) (>50%)/cropland (<50%)
Mosaic natural vegetation (tree shrub evergreen) 1.5 1.4 1.4
Tree cover broadleaved evergreen closed to open (>15%) 1 1.3 1
Tree cover broadleaved deciduous closed to open (>15%) 1.1 1.4 1.1
Tree cover broadleaved deciduous closed (>40%) 1 1.3 1
Tree cover needle leaved evergreen closed to open (>15%) 1.1 1.5 1.1
Tree cover needle leaved evergreen closed (>40%) 1.1 1.3 1
Tree cover needle leaved deciduous open (15–40%) 1.2 1.2 1.3
Tree cover needle leaved deciduous closed to open (>15%) 1.2 1.3 1.3
Tree cover needle leaved deciduous closed (>40%) 1.3 1.3 1.5
Tree cover mixed leaf type (broadleaved and needle leaved) 1 1.1 1
Mosaic tree and shrub (>50%)/herbaceous cover (<50%) 1.4 1.5 1.2
Mosaic herbaceous cover (>50%)/tree and shrub (<50%) 1.3 1.4 1.3
Shrubland 1.5 1.4 1.6
Shrubland evergreen 1.2 1.6 1.2
Shrubland deciduous 1.4 1.4 1.6
Grassland 1.6 1.3 1.4
Sparse vegetation (tree shrub) 1.6 1.3 1.7
Tree cover flooded fresh/saline/brackish water 1 1.1 1.2
Shrub or herbaceous cover flooded 1.1 1.1 1.2
Bare areas 2 1 2
Consolidated bare areas 2 1 2
Unconsolidated bare areas 2 1 2
Degrading vegetation qualities cause drought, fire, and erosion which accelerate the progress of desertification processes
Natural vegetation cover (in %) Score Data source and references
> 40 1 Data collected from LP DAAC,
10–40 1.8 USGS. Scoring system
adopted from Jiang, Jiapaer,
< 10 2 Bao, Li, & Guo (2019),
Kosmas et al. (1999)
Natural vegetation can retard the process of desertification by adding biomass to the soil

are land use potency and policy implementation (Table 4). Land use 2.2.5 | Environmentally sensitive area index
potency provides an idea of the degrees at which land is vulnerable to
desertification while the policy implementation gives a concept on the All the data collected from various sources for the period from 1992
remedies adopted for resisting desertification conditions. The MQI to 2015, were resampled and used as indicators for different indices.
was calculated by the following Equation (6): With all the four indices that is, SQI, CQI, VQI, and MQI, a composite
index was formulated that is the ESAI. This uses the geometric mean
MQI ¼ ðland use potency  policy implementationÞ½ : ð6Þ of the four indices as shown in Equation (7).
RAJBANSHI AND DAS 7

TABLE 4 Description of the indicators and scores of management quality index (MQI)

Quality
index Indicators Score Data source and references
MQI Land use land cover Land use Data collected from CCI (Climate-Change
potency Initiative) Land Cover products, European
Cropland rainfed 1.7 Space Agency (ESA). Scoring system
adopted from Ferrara et al. (2020)
Herbaceous cover 1.7
Tree or shrub cover 1.4
Cropland irrigated or post-flooding (herbaceous cover) (<50%) 1.6
Mosaic cropland (>50%)/natural vegetation (tree shrub 1.5
herbaceous cover) (>50%)/cropland (<50%)
Mosaic natural vegetation (tree shrub evergreen) 1.5
Tree cover broadleaved evergreen closed to open (>15%) 1.1
Tree cover broadleaved deciduous closed to open (>15%) 1.2
Tree cover broadleaved deciduous closed (>40%) 1
Tree cover needle leaved evergreen closed to open (>15%) 1.1
Tree cover needle leaved evergreen closed (>40%) 1.1
Tree cover needle leaved deciduous open (15–40%) 1.2
Tree cover needle leaved deciduous closed to open (>15%) 1.1
Tree cover needle leaved deciduous closed (>40%) 1.5
Tree cover mixed leaf type (broadleaved and needle leaved) 1
Mosaic tree and shrub (>50%)/herbaceous cover (<50%) 1.2
Mosaic herbaceous cover (>50%)/tree and shrub (<50%) 1.5
Shrubland 1.7
Shrubland evergreen 1.5
Shrubland deciduous 1.6
Grassland 1.5
Sparse vegetation (tree shrub) 1.8
Tree cover flooded fresh/saline/brackish water 1.2
Shrub or herbaceous cover flooded 1.3
Bare areas 2
Consolidated bare areas 2
Unconsolidated bare areas 2
Land use influences the land to great extents determining its' sensitivity toward degradation and desertification
Policy implementation Score Data Source and References
Permanently irrigated land, broadleaved forest 1 Data collected from CCI (Climate-Change
Cropland with natural vegetation 1.5 Initiative) Land Cover products, European
Space Agency (ESA). Scoring system
Rainfed cropland, natural vegetation, forest, beaches dunes, sands 2 adopted from Jiang, Jiapaer, Bao, Li, &
Guo (2019), Kosmas et al. (1999)
Proper policy implementation can protect the land from further degradation along with reviving it

ESAI ¼ ðSQI  CQI  VQI  MQIÞ /4 :


1
ð7Þ 2 (1.27–1.32), fragile 3 (1.33–1.37), and (iv) critical 1 (1.38–1.41),
critical 2 (1.42–1.53), critical 3 (>1.53).
Having followed the scoring system proposed by Ferrara
et al. (2020) and Kosmas et al. (2014), the ESAI values ranged
between 1 (least sensitive) and 2 (most sensitive) which were later 2.3 | Linear regression analysis
classified into four classes according to the vulnerability of the lands
toward the risk of desertification. The classes are (i) non-affected With all the data collected for a 24-years' time span (1992–2015), a
(<1.17), (ii) potential (1.17–1.22), (iii) fragile 1 (1.23–1.26), fragile trend was monitored for each index by incorporating the linear
8 RAJBANSHI AND DAS


regression analysis (Jiang, Jiapaer, Bao, Li, & Guo, 2019). SQI, CQI, UBk ¼ UFk
: ð14Þ
VQI, MQI, and ESAI, all these five indices were met with this k ¼nþ1k
approach. Equation (8) was used to obtain the slope of the trend of
each index. When the UBk and UFk curves cut the two significant lines, the point
of intersection denotes the break or interval in time from where a
P P P
n ni¼1 i  Xi  ni¼1 i  i¼1
n Xi change is about to take place. This point is known as the mutation
Slope ¼   , ð8Þ
iP
¼1 P 2
n i2  i¼1 point or mutation year.
n i
n

Where: Slope denotes the trend of change of each pixel of the 2.5 | Identification of the important indicators
dependent variables, X is the yearly value of the indicator for the ith
year, i is the consecutive year, and n is the total time period in years. The random forest (RF) technique was applied for identifying the
We have applied the linear regression on each pixel in the Country major indicators affecting the desertification process in the study area.
that is, a total of 339,880 pixels. From these, a number of points RF is an enhanced classifying tool that is a collection of several tree-
were used to measure the robustness of the model. The statistical structured classifiers inspired by the structure of a forest, and ana-
validity of linear regression model verified by: (a) ANOVA of the lyses a dataset consisting of independent random vectors
model tested, (b) significance of each variable included in the model, (Breiman, 2001). This method has been successfully used in soil degra-
(c) a mean of residues, (d) normality of unstandardized residues by dation analysis by Taghizadeh-Mehrjardi et al. (2020) and Yu, Wang,
Shapiro–Wilk test, (e) the existence of potential outliers by Cook's Wang, Ren, & Zhang (2019). To perform this model, 3,000 random
distance. points were selected from the entire study area and the dataset was
divided into 75% and 25% for training and testing purposes
(Figure S1). The evaluation of the model was performed by calculating
2.4 | Mann-Kendall mutation detection test the confusion matrix and the receiver operating characteristic (ROC)
curve, which plotted the sensitivity (false positive rate) against the
To understand the nature of the desertification process over time, it is specificity (false negative rate). It is also known as the success rate
important to know if there is an abrupt change in the occurrence of curve (calculated from the training dataset) and the prediction
the process that marks the evolution of one state to another rate curve (calculated from the testing/validation dataset) depending
(Thangavel et al., 2018; Wang, 2020; Wang, Zheng, & Zang, 2012). In upon the dataset being used and represents how well the explanatory
this work, the period of 24 years (1992–2015) has been subjected to variables explain the desertification process in the study area. The
the Mann-Kendall (MK) mutation point detection test to identify the ROC curve has also been verified using the area under the curve
different time breaks, considering the estimated ESAI. The test has (AUC) statistics the values of which vary from 0.5 to 1.0. The value
been performed as follows [Equations (9)–(14)]: closer to 1 indicate the more accurate and robustness of the model
(Park & Kim, 2019).
X
n1 X
n
S¼ signðxi  xk Þ i ¼ 1, 2,…, n, ð9Þ
k¼1 i¼kþ1
3 | RE SU LT S
8 9
< 1 if xi  xk > 0 >
> =
signðxi  xk Þ ¼ 0 if xi  xk ¼ 0 , ð10Þ 3.1 | Spatial pattern of the quality indices
>
: >
;
1 if xi  xk < 0
The CQI map (Figure 2a) shows high values in west Rajasthan (1.5–
Where: S is the MK test statistic and xi is the data point for the 1.9) and in parts of Punjab, Haryana, and Gujarat while low CQI values
ith year. can be observed over the eastern, southeastern, and central India, the
The statistics for the variables are estimated by the following: western coast and in parts of Ladakh, Jammu and Kashmir, Himachal
Pradesh and Uttarakhand (1.0–1.2). A general negative change in the
jSk  EðSk Þj CQI can be observed (Figure 2e) all over the Country, mostly in parts
UFk ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi k ¼ 1,2, …, n, ð11Þ
VarðSk Þ of Punjab, Rajasthan, Gujarat, and Karnataka (0.016 to 0.004 yr
1
). A prominent positive change can be observed over Ladakh
EðSk Þ ¼ kðk  1Þ=4, ð12Þ (0.007–0.018 yr1) with p < 0.05 proving a significant change while
at some places (eastern Ladakh) with insignificant change (p > 0.05)
VarSk ¼ kðk  1Þð2k þ 5Þ=72: ð13Þ was observed. The VQI map (Figure 2b) exhibits the lowest values in
Arunachal Pradesh, Nagaland, Manipur, Mizoram, parts of Assam, Sik-
The reversed statistics for the variables can be obtained by the kim, Tripura, Meghalaya, Uttarakhand and along the Malabar Coas
following: region (1.00 and 1.17). However, moderate VQI values can be
RAJBANSHI AND DAS 9

perceived over most of the Country (1.57–1.69). Almost the entire are covered by the states such as Ladakh, Rajasthan, Maharashtra,
Country shows an increasing trend of vegetation quality with a yearly Uttar Pradesh, Jharkhand, West Bengal, the Malabar Coast, and the
change rate of 0.006–0.03 (Figure 2f). Statistically significant changes northeastern states along with a few scattered areas of Odisha and
can be observed over the Gangetic Plains, the Malabar Coast, and the Chhattisgarh. The statistics used to validate the regression provided
northeast part of the Country with a rate varying from 0.001 to 0.03 excellent and statistically significant results that successfully explain
(p < 0.05). In the states of Jharkhand, Chhattisgarh, Madhya Pradesh, the robustness and the accuracy of the model application and its out-
Odisha, Andhra Pradesh, Telangana, Arunachal Pradesh and comes. An example of such validation is given in Figure S2.
Uttarakhand we found lower MQI values in a scattered manner, rang-
ing from 1.00 to 1.16 (Figure 2c). Whereas, most of the peninsular
India, Rajasthan, Gujarat, Jharkhand, Odisha, Chhattisgarh, and parts 3.3 | Mutation points of ESAI and transitions of
of Uttar Pradesh, Ladakh, Assam, and Haryana show high quality of different desertification risk category
management with the mean MQI values ranging within 1.8–2.0. The
entire Country shows an overall moderate to a degrading rate of The MK-test shows that between the critical lines (±1.96) the two
change management quality, with a varying rate of 0.008 to curves UF and UB intersect at three points, years 1996, 2009, and
1
0.004 yr (Figure 2g). These areas show a statistically significant rate 2014 denoting abrupt changes (Figure 4) and therefore the entire time
of change in MQI with p < 0.05. The mean SQI map (Figure 2d) illus- span has been divided into three periods that is, 1992–1996, 1997–
trates that the northern states like Jammu and Kashmir, Ladakh and 2009, and 2010–2015. The spatial distribution of these three muta-
Himachal Pradesh and the eastern region consisting of Nagaland, tion years was acquired by assessing each of the pixels for detecting
Manipur, Mizoram, and parts of Assam have the highest values of SQI changes (Figure 5a–c) along with the statistical outcomes of each of
ranging from 1.49 to 1.74, closely followed by Odisha and Chhattis- the states depicting the percentage of areas experiencing the changes
garh and parts of Jharkhand, Andhra Pradesh and Madhya Pradesh over these mutation years (Figure 5d). During the span of 1992–
(1.49–1.63). Lowest SQI values can be observed mostly across Punjab, 1996, more prominent changes can be observed in the northern
Haryana, parts of Bihar, Tamil Nadu, Karnataka, Andhra Pradesh, and western part of the Country. In the west, Gujarat, Rajasthan,
Gujarat, Chhattisgarh, West Bengal, Madhya Pradesh, Himachal and Maharashtra had 43.39%; 36.57%, and 32.38% of their lands
Pradesh, Ladakh, and parts of Jammu and Kashmir, Rajasthan, altered, respectively, while Chandigarh, Jammu and Kashmir, and
Uttarakhand, Uttar Pradesh, and Odisha (1.0–1.2). It is noticeable that Ladakh in the north showed 80%; 50.39% and 43.81% of their
the mountainous and Plateau regions of Ladakh, Himachal Pradesh, lands, respectively, experiencing changes over the period of 1992–
Jammu and Kashmir, Chhattisgarh with their varying topography con- 1996. The 1997–2009 period showed a roughly similar trend of
tain both high and low SQI values. changes in lands by desertification, however, in this period, more
significant changes have been observed over almost the entire
Country with the exceptions of Ladakh, Rajasthan and few parts in
3.2 | Spatial pattern of the ESAI the northeast. Most of the states suffered desertification changes
over nearly 25%–35% of their lands. The changes observed over
The ESAI analysis for 1992–2015 (Figure 3) provides a comprehensi- 2010–2015 are significantly lower than the previous years. Yet,
ble perception of land degradation in India and the varying degrees of distinct effects of desertification can now be observed over the
sensitivity to desertification in various parts of the Country. The north-eastern states such as Nagaland, Arunachal Pradesh, Mani-
states in the northeast show low-to-moderate sensitivity according to pur and Assam where apparently 10%–17% of lands have suffered
the mean ESAI results (1.05–1.28). These are the states where better negative changes.
soil and vegetation quality could be observed previously. The lowest In the next step, the four different ESAI maps were created for
mean ESAI values can be identified in Arunachal Pradesh (1.05–1.4). the years 1992, 1996, 2009, 2015 (Figure 6a–d) featuring eight fragil-
During 2010–2015, the State had 25.16% (Table S1) of its' total land ity classes from Non-affected by desertification to Critical-3. During
under non-affected category, which was the highest among all the 1992–1996, the total amount of fragile land was about
2
states. Along with the northeast, other states of the Gangetic basin 1,148,660 km (34.94% of the total land) encompassing the Gangetic
and the western coast, especially the Malabar Coast present a low to Plains and the northeast, while 1,969,910 km2 (59.93%) was critically
moderate mean ESAI values (1.05–1.34) denoting a low sensitivity. sensitive to desertification affecting mostly the western and southern
On-the-other-hand, Rajasthan and Ladakh give the highest mean ESAI states which worsened in 1997–2009 as the fragile lands started to
values (1.5–1.7), where 87.61% and 83.83% of land, respectively, are transform into Critical categories (Figure 6b,c and Figure 7). Degrees
critically degraded, followed by most of the southern states with land of critical sensitivity also worsened during this time as 143,970 km2
critically degraded up to 55%–74%. A negative trend of the ESAI can was added to the Critical-3 category, the majority of which came from
be observed in Rajasthan, Gujarat, Punjab, Kerala, Tamil Nadu, and the transition of lands from the Critical-2 category (Figure 7). The Crit-
1
Karnataka (0.01 to 0.004 yr ). However, the eastern part of Ladakh ical lands are mostly concentrated in Rajasthan in 1992 under both
displayed a positive change in the ESAI qualities throughout 1992– Critical-2 (56.22%) and Critical-3 (26.28%). This degradation can be
2015 (0.002–0.02 yr1). The areas with significant change (p < 0.05) observed to be continuously spreading toward Gujarat and
10 RAJBANSHI AND DAS

F I G U R E 2 Spatial distribution
and change of the annual ranges of
the quality indices (climate quality
index, vegetation quality index,
management quality index and soil
quality index) over the period of
1992–2015: (a) CQI Mean; (b) VQI
Mean; (c) MQI Mean; (d) SQI Mean;
(e) CQI Slope; (f) VQI Slope and
(g) MQI Slope. The statistically
significant changes (p < 0.05) are
shown as black dots [Colour figure
can be viewed at
wileyonlinelibrary.com]

Maharashtra as more than 65% area in these states becomes Critical However, in 2015, the lands again transform back to fragile lands as a
lands. By 2009, most of Rajasthan is under highly critical category reduction in the extent of Critical land is observed (Figures 6d and 7).
with a total of 84.97% of land under Critical-2 and the Critical-3. After 2009, a decrease of 84,650 km2 was observed in Critical-3,
RAJBANSHI AND DAS 11

F I G U R E 3 Spatial distribution and change of the annual ESAI for the period of 1992–2015: (a) ESAI Mean and (b) ESAI Slope. The statistically
significant changes (p < 0.05) are shown as black dots [Colour figure can be viewed at wileyonlinelibrary.com]

lands under the Critical category and this scenario does not present
much change from 1992 to 2015. In the Ladakh region, some of the
most severe forms of degradation can be observed in the period
between 1992 and 1996. The area covering about 60.25% of the
state was under Critical-2 category while 10.9% was under
Critical-3. In 1997–2009, the Critical-3 area hiked up to 37.44% and
in 2010–015, the figure was 33.76%. On the contrary, the North-
East region shows a constant high vegetation and soil quality
throughout 1992–2015. Yet, it can be noticed that a general rate of
degradation has still taken place in that region with around 49% land
of the entire northeastern states being under Fragile category and
29% being under Critical category.
F I G U R E 4 Mutation points over the period of 1992–2015
[Colour figure can be viewed at wileyonlinelibrary.com]

3.4 | Important variables for desertification and


most of which transformed into Critical-2 and Critical-1. Some of the validation of the RF model
Critical category lands transformed into Fragile categories as well,
especially the Fragile-2 category that gains around 36,660 km2 since The indicators/drivers were further evaluated through the RF model.
2009. Categories such as Non-affected and Potentially affected In this study, the drought resistance has the highest importance
have also increased to some extent during 2009–2015 showing a value among all the other inputs that is, 46% followed by erosion
slight improvement in the overall scenario. Yet, it can be said after protection with 20%, precipitation with 7%, policy implementation
observing the process over the entire period, that Critical-3 category and aridity being 6% (Figure 8a). The performance of the RF-model
land has undeniably increased since 1992 (Figure 7) and the amount was analyzed using the test dataset, which showed the producers
is around 64,530 km2 confirming the fact that degradation has accuracy varies from 67.74% in non-affected to 94.81% in the
indeed taken place. The degradation process has extended from the Critical-2 category with an overall accuracy of 82.13% (Table S2).
west and has spread over the southern region as well. Except for The ROC curve showed that the RF-model had an overall higher
Kerala, all the southern states such as Andhra Pradesh, Karnataka, AUC (0.97) explaining good accuracy in the spatial prediction of the
Tamil Nadu, Telangana, and Odisha has around 60% or more of their desertification process (Figure 8b).
12 RAJBANSHI AND DAS

F I G U R E 5 Spatial
distribution of mutation occurring
to the lands due to desertification
over three classified periods
(a) 1992–1996, (b) 1997–2009,
(c) 2010–2015, along with
(d) percentage of area mutated
over the three periods for
different states in India. The
legend is the same for (a–c) and
(d) [Colour figure can be viewed
at wileyonlinelibrary.com]

4 | DISCUSSION encroaching anthropogenic activities in 1997–2009 (Moharana


et al., 2016; Salunkhe et al., 2018; Yusuf, Ameta, Usman, Tukur, &
The present work attempts to explain the succession of the desertifi- Hamza, 2020). By 2009, 84.97% of land Rajasthan is observed to be
cation process in India through different time periods. The MK test under the highly Critical category which corroborated with the find-
presented the years 1996, 2009, and 2014 as the points of abrupt ings of Moharana et al. (2016) explaining 83.85% of the land in Rajas-
changes in the desertification process (Figure 4). The spatial distribu- than was experiencing high degrees of degradation during this time.
tion advocated that the severe desertification has occurred over the Studies denote that Maharashtra suffered so harsh levels of land deg-
country especially in the states of Rajasthan, Gujarat, Maharashtra, radation and groundwater scarcity that it resulted in a great number
Ladakh and Jammu and Kashmir with the effects extended toward the of farmer suicides reaching up to 60,750 reported cases during 1995–
adjacent states (Figure 5). Few states such as Chandigarh and 2013 with the rate of 9 per day (Chinnasamy, Hsu, &
Puducherry showed high percentages of change (80%, 64.29%, Agoramoorthy, 2019). Gujarat faced heat waves of moderate-to-
respectively) in 1992–1996 and the same occurred for Dadra and severe degrees during 1990–2010 along with drought periods during
Nagar Haveli and Daman and Diu in 1997–2009 (94.12% combined). 1999–2002 affecting the land qualities massively (Bandyopadhyay,
But due to low areal extent compared to other states, they do not Bhuiyan, & Saha, 2019). Except for Kerala and the Western Ghat
project a significant impact upon the broad view of changes perceived region (Krishnan, Murugaiya, Shanmugham, & Mariappan, 2015 June
over the whole Country. A major negative change in the land qualities 22-July 5), the other southern states have more than 50% of their
has been noted in the dry and arid western part of the Country, the lands under Critical categories, especially Critical-2. This semiarid
region that hosts most of the Great Thar Desert within it region faces land degradation mostly through water-driven soil ero-
(208,110 km2) (Kundu et al., 2017; Sur & Chauhan, 2019). Rajasthan, sion caused by the lack of vegetation cover, salinization, wind erosion,
Gujarat, Maharashtra, and parts of Punjab, Haryana, and Delhi suf- and anthropogenic activities (Chouhan, 2005; Shoba &
fered the negative implications of the extension of the desert (Kar & Ramakrishnan, 2016; Saji & Jayakumar, 2017; Kumar et al., 2020).
Kumar, 2020). Our study shows that these states have already faced However, during 2010–2015, improvement in the land qualities
degradation to great extents during 1992–1996. This situation wors- of the western and southern states could be observed (Figure 5). In
ened the land and vegetation qualities in those regions with the Rajasthan, Critical-3 lands decreased distinctly by 20.05% in 2010–
RAJBANSHI AND DAS 13

F I G U R E 6 The degrees of sensitivity of lands


toward desertification as per ESAI for the years
(a) 1992, (b) 1996, (c) 2009 and (d) 2015 [Colour
figure can be viewed at wileyonlinelibrary.com]

2015 than before. The reports of the Central Arid Zone Research
Institute (CAZRI) support our results indicating an improvement in the
land qualities in Rajasthan (Moharana et al., 2016). This could be due
to the implementation of close monitoring of the desertification haz-
ards and changes in government land policies performed afterward
such as the introduction of Rajasthan State Action Plan on Climate
Change (RAPCC) under Rajasthan State Climate Change Agenda
(RCCA) for 2010–2014 (TERI, 2010). Implementation of projects con-
cerning saline soil and water treatment, land reclamation, sustainable
land use management, agro-forestry, eco-restoration, and most impor-
tantly, adaptation strategies to the changing environment adopted in
these areas has provided satisfactory outcomes (Ponnuchamy
et al., 2013; Singh, 2009).
Among the northern states and UTs, Ladakh has been facing the
most severe form of desertification. The lands that were under
F I G U R E 7 Sankey diagram showing the ESAI sensitivity
transitions for the period of 1992–2015 [Colour figure can be viewed Critical-2 class during 1992–1996, transformed into Critical-3 by huge
at wileyonlinelibrary.com] amounts till 1997–2009 and continued throughout 2010–2015. The
14 RAJBANSHI AND DAS

throughout 1992–2015. However, the region is not entirely devoid of


degradation. In Assam, Sikkim, and Manipur, low-to-moderate deserti-
fication has taken place. Arunachal Pradesh, Tripura, Meghalaya,
Mizoram, and Nagaland are also to some extent under the degrada-
tion process due to increase in population pressure, land-use change,
shifting cultivation, petroleum extraction, and natural phenomenon
such as floods (Ravindranath et al., 2011; Prokop & Poręba, 2012;
Ritse, Basumatary, & Susan, 2020). However, the general trend of
environmental qualities in this region has improved especially in
Meghalaya, Mizoram, Sikkim, and Nagaland with practice of conserva-
tion agriculture, conservation of forests, and restoration of degraded
lands (Ghosh et al., 2010; Kumar et al., 2019; Sengupta, 2020).
The RF results show that the drought resistance, erosion protec-
tion, precipitation, policy implementation, and aridity are the indica-
tors that have the most significant role in controlling the sensitivity of
land toward the desertification process. Our quality index analysis
(Figure 2) and ESAI map (Figure 3) also supported this idea by illustrat-
ing that the states with high drought prone area (Andhra Pradesh,
Gujarat, Jharkhand, Karnataka, Maharashtra, Rajasthan) present higher
sensitivity toward desertification and vice versa. The drought condi-
tions not only influence the soil but also the hydrological cycle as well,
influencing the climate in turn and thus causing a severe case of
desertification (Kumari, Banerjee, Kumar, & Kumar, 2019). The intense
desertification in the states of Rajasthan, Gujarat, Ladakh is also the
result of lack of precipitation, aridity along with deforestation and

F I G U R E 8 Analysis of drivers of desertification with Random unplanned cultivation (Gupta & Ansari, 2014; Memon, Vyas, &
Forest: (a) feature importance distribution of each driver, (b) receiver- Ft, 2016; Raghuvanshi, Moharana, Saxena, & Saha, 2020; Varghese &
operating characteristic (ROC) curve and area under the ROC curve Singh, 2016). Proper land-use planning has brought noticeable
(AUC) generated by Random Forest classifier using the test dataset changes in the western states as well as the north-east. Effective pol-
[Colour figure can be viewed at wileyonlinelibrary.com]
icy implementation has protected extensive parts of Rajasthan from
irreversible loss (Moharana et al., 2016; Singh, 2009; TERI, 2010). The
bare topography and harsh climatic conditions with high diurnal tem- northeast has also taken up land restoration and conservation prac-
perature range of this area coupled with anthropogenic activities have tices (Brahma et al., 2017; Ghosh et al., 2010) with a visible improve-
greatly influenced desertification in this region (Kar & Kumar, 2020). ment in land quality. The southern region is still struggling to reach
Being located upon the collision zone of the Indian Plate and Eurasian this goal yet making progress through small steps like performing con-
Plate, tectonic activities is another important factor in influencing the servation practices through religious belief (Ballullaya, Reshmi, Rajesh,
constant erosional activities in this area (Dietsch, Dortch, Reynhout, Manoj, & Lowman, 2019) and watershed management
Owen, & Caffee, 2015). Moreover, repeated occurrences of flash (Balasubramani, 2019). Nevertheless, the work still suffers from some
floods due to cloudbursts and snow melting also cause erosion in this minor limitations. The work is greatly dependent on high-resolution
region. In the past few decades, especially since 1995, there have satellite data; therefore, errors in the dataset may result in flawed out-
been several flash floods and in the last decade, the number has comes. This problem may have affected the accuracy of the results
increased due to heavy development and construction activities in gathered for the state of Ladakh. For performing the analysis in a
this region (Gupta, Khanna, & Majumdar, 2012; Hakhoo et al., 2019). national scale, few localized drivers such as physical setting and man-
In the east and southeast, Jharkhand and Odisha have more than agement systems at local levels may have been neglected as well.
50% of lands under critical categories. Throughout 1992–2015, these However, even with such minor shortcomings, the work still success-
states have been facing a steady rate of degradation with a slight fully provides a clear picture of the desertification-affected scenario
increase in the last decade due to anthropogenic activities such as of the Country.
deforestation and coal mining. Lack of effective mitigation plans
unlike the western states, has promoted continuous desertification in
these states (Kumar, Rani, Pandey, Majumdar, & Nathawat, 2010; 5 | CONC LU SIONS
Kumar, Sharma, Tara, & Muniyan, 2019; Mazumdar & Paul, 2017;
Shalini Tirkey, Pandey, & Nathawat, 2013). The north-eastern region, This work is an attempt to perform a holistic approach to the investi-
however, is showing a constant resistance toward desertification gation of land sensitivity to desertification in the subcontinent of
RAJBANSHI AND DAS 15

India for the time span of 1992–2015 using the ESAI approach. The of rural and urban neighbourhood communities. Land Use Policy, 89,
work explains with evidence how desertification has been affecting 104213. https://doi.org/10.1016/j.landusepol.2019.104213
Bandyopadhyay, N., Bhuiyan, C., & Saha, A. K. (2019). Progress in disaster
the western and southern regions of the Country and has been
science drought mitigation: Critical analysis and proposal for a new
encroaching toward the east. The process has also taken place in the drought policy with special reference to Gujarat (India). Progress in
Ladakh region in the north at an alarming rate, raising the call for close Disaster Science, 5, 100049. https://doi.org/10.1016/j.pdisas.2019.
attention. With proper mitigation practices, the west and the north- 100049
Bezerra, F. G. S., Aguiar, A. P. D., Alvala, R. C. S., Giarolla, A.,
east have been somewhat successful in recovering the quality of
Bezerra, K. R. A., Lima, P. V. P. S., … Arai, E. (2020). Analysis of areas
lands. But there is still a long way to go. According to our study, undergoing desertification, using EVI2 multi-temporal data based on
33.76% land of the Country is still under very highly critical zone and MODIS imagery as indicator. Ecological Indicators, 117, 106579.
without any immediate action; it may result in irreversible loss. The https://doi.org/10.1016/j.ecolind.2020.106579
Boschetto, R. G., Mohamed, R. M., & Arrigotti, J. (2010). Vulnerability to
work also points out that anthropogenic factors such as deforestation,
desertification in a sub-Saharan region: A first local assessment in five
mining, exploitation of resources, and unplanned land use are much villages of the southern region of Malawi. Italian Journal of Agronomy,
more dominant in influencing and aiding the desertification process in 5, 91–101. https://doi.org/10.4081/ija.2010.s3.91
this country. The ESAI approach adopted in this study thus has been a Brahma, B., Pathak, K., Lal, R., Kurmi, B., Das, M., Nath, P. C., … Das, A. K.
(2017). Ecosystem carbon sequestration through restoration of
very useful and effective method in the examination of the desertifi-
degraded lands in Northeast India. Land Degradation and Development,
cation process in India. The work successfully provides a clear picture
29(1), 15–25. https://doi.org/10.1002/ldr.2816
of the desertification affected scenario of the Country and therefore Breiman, L. E. O. (2001). Random Forests. 5–32.
attempts to make a valuable contribution to research conducted on Budak, M., Günal, H., Çelik, _I., Hakan, Y., Acir, N., & Acar, M. (2018). Envi-
this serious issue so that it may be a helpful base for the future ronmental sensitivity to desertification in northern Mesopotamia;
application of modified MEDALUS by using analytical hierarchy
researchers and for policy makers in their task of planning and seeking
process.
better and sustainable land use programmes. Chinnasamy, P., Hsu, M. J., & Agoramoorthy, G. (2019). Groundwater stor-
age trends and their link to farmer suicides in Maharashtra State, India.
CONF LICT OF IN TE RE ST Frontiers in Public Health, 7, 1–7. https://doi.org/10.3389/fpubh.2019.
00246
There is no conflict of interest that exists in the submission of this
Chouhan, T. S. (2005). Degree, extent and treatment of desertification
manuscript. hazards in India. Sociedade & Natureza, 1(1), 901–919. https://www.
redalyc.org/articulo.oa?id=321328500085.
DATA AVAI LAB ILITY S TATEMENT Christian, B. A., & Dhinwa, P. S. (2018). Long term monitoring and assess-
ment of deserti fi cation processes using medium & high resolution
The data that support the findings of this study are available from the
satellite data. Applied Geography, 97, 10–24. https://doi.org/10.1016/
corresponding author upon reasonable request. j.apgeog.2018.04.010
Darbalaeva, D., Mikheeva, A., & Zhamyanova, Y. (2020). The socio-
economic consequences of the desertification processes in Mongolia.
ORCID E3S Web of Conferences, 164, 11001. https://doi.org/10.1051/
Joy Rajbanshi https://orcid.org/0000-0001-7044-5476 e3sconf/202016411001
Dasgupta, A., Sastry, K. L. N., Dhinwa, P. S., Rathore, V. S., &
Nathawat, M. S. (2013). Identifying desertification risk areas using
RE FE R ENC E S fuzzy membership and geospatial technique – A case study, Kota Dis-
Abdollahi, A., Nezhad, M. P., & Pradhan, B. (2019, August). Investigation of trict, Rajasthan. Journal of Earth System Science, 122, 1107–1124.
the Vegetation Cover and the Vulnerability of the Mashhad Regions to https://doi.org/10.1007/s12040-013-0331-x
Desertification by Using MODIS Image and EVI [Paper presentation]. Dharumarajan, S., Bishop, T. F., Hegde, R., & Singh, S. K. (2018). Desertifi-
2019 IEEE International Conference on Cybernetics and Computational cation vulnerability index-an effective approach to assess desertifica-
Intelligence (CyberneticsCom) (pp. 46-49). IEEE. 10.1109/ tion processes: A case-study in Anantapur District, Andhra Pradesh,
CYBERNETICSCOM.2019.8875666 India. Land Degradation & Development, 29, 150–161. https://doi.org/
Akbari, M., Jafari, M., Memarian, H., & Gholami, A. (2020). Monitoring 10.1002/ldr.2850
desertification processes using ecological indicators and providing Dietsch, C., Dortch, J. M., Reynhout, S. A., Owen, L. A., & Caffee, M. W.
management programs in arid regions of Iran. Ecological Indicators, (2015). Very slow erosion rates and landscape preservation across the
111, 106011. https://doi.org/10.1016/j.ecolind.2019.106011 southwestern slope of the Ladakh Range, India. Earth Surface Processes
Avni, Y., Porat, N., Plakht, J., & Avni, G. (2006). Geomorphic changes lead- and Landforms, 40, 389–402. https://doi.org/10.1002/esp.3640
ing to natural desertification versus anthropogenic land conservation Emadodin, I., Reinsch, T., & Taube, F. (2019). Drought and desertification
in an arid environment, the Negev Highlands, Israel. Geomorphology, in Iran. Hydrology, 6(3), 66. https://doi.org/10.3390/
82, 177–200. https://doi.org/10.1016/j.geomorph.2006.05.002 hydrology6030066
Balasubramani, K., Gomathi, M., Bhaskaran, G., & Kumaraswamy, Ferrara, A., Kosmas, C., Salvati, L., Padula, A., Mancino, G., & Nolè, A.
K. (2019). GIS-based spatial multi-criteria approach for characteriza- (2020). Updating the MEDALUS-ESA framework for worldwide land
tion and prioritization of micro-watersheds: a case study of semi-arid degradation and desertification assessment. Land Degradation & Devel-
watershed, South India. Applied Geomatics, 11(3), 289–307. https: opment, 31, 1593–1607. https://doi.org/10.1002/ldr.3559
//doi.org/10.1007/s12518-019-00261-y Ghosh, P. K., Das, A., Saha, R., Kharkrang, E., Tripathi, A. K., Munda,
Ballullaya, U. P., Reshmi, K. S., Rajesh, T. P., Manoj, K., & Lowman, M. G. C., & Ngachan, S. V. (2010). Conservation agriculture towards
(2019). Land use policy stakeholder motivation for the conservation of achieving food security in North East India. Current Science, 99(7),
sacred groves in South India: An analysis of environmental perceptions 915–921. http://www.jstor.org/stable/24066068
16 RAJBANSHI AND DAS

Gupta, P., Khanna, A., & Majumdar, S. (2012). Disaster management of Kumar, P., Rani, M., Pandey, P. C., Majumdar, A., & Nathawat, M. S. (2010).
flash floods in Leh (Ladakh): A case study. Indian Journal of Community Monitoring of deforestation and forest degradation using remote
Medicine, 37(3), 185. https://doi.org/10.4103/0970-0218.99928 sensing and GIS: A case study of Ranchi in Jharkhand (India). Report
Gupta, V., & Ansari, A. A. (2014). Geomorphic portrait of the Little Rann of and Opinion, 2(4), 14–20.
Kutch. Arabian Journal of Geosciences, 7, 527–536. https://doi.org/10. Kumar, R., Sharma, S., Tara, K., & Muniyan, S. (2019). Environmental issues
1007/s12517-012-0743-y of coal mines and its allied industries: Suggestive Measures for Root
Hakhoo, N., Bhat, G. M., Pandita, S., Hussain, G., Haq, A. U., Hafiz, M., … Level Remedies. In P. K. Singh, A. K. Singh, D. Kumbhakar, S. Singh, M.
Thusu, B. (2019). Natural hazards their drivers, mechanisms and Kumar (Eds.) : International Conference and Exhibition on Energy &
impacts in the Shyok-Nubra Valley, NW Himalaya, India. International Environment: Challenges & Opportunities ENCO 2019, Volume-I
Journal of Disaster Risk Reduction, 35, 101094. https://doi.org/10. (332-339). Central Institute of Mining and Fuel Research.
1016/j.ijdrr.2019.101094 Kumari, R., Banerjee, A., Kumar, R., & Kumar, A. Saikia, P., & Khan,
Hargreaves, G. H., & Samani, Z. A. (1985). Reference crop evapotranspira- M. L. (2019). Deforestation in India: Consequences and Sustainable
tion from temperature. 96–99. Solutions. In M. N. Suratman, Z. A. Latif, G. De Oliveira,
Higginbottom, T. P., & Symeonakis, E. (2014). Assessing land degradation N. Brunsell, Y. Shimabukuro & C. A. C. Dos Santos (Eds.). Forest
and desertification using vegetation index data: Current frameworks Degradation Around the World (53-70). London: IntechOpen.
and future directions. Remote Sensing, 6, 9552–9575. https://doi.org/ 10.5772/intechopen.85804
10.3390/rs6109552 Kundu, A., Patel, N. R., Saha, S. K., & Dutta, D. (2017). Desertification in
Hu, Y., Han, Y., & Zhang, Y. (2020). Land desertification and its influencing western Rajasthan (India): Efficiency and residual trend methods. Natu-
factors in Kazakhstan. Journal of Arid Environments, 180, 104203. ral Hazards, 86(1), 297–313. https://doi.org/10.1007/s11069-016-
https://doi.org/10.1016/j.jaridenv.2020.104203 2689-y
_Imamoglu, A., & Dengiz, O. (2019). Evaluation of soil quality index to Lee, E. J., Piao, D., Song, C., Kim, J., Lim, C., Kim, E., … Lim, C. (2019).
assess the influence of soil degradation and desertification process in Assessing environmentally sensitive land to desertification using
sub-arid terrestrial ecosystem. Rendiconti Lincei. Scienze Fisiche e Nat- MEDALUS method in Mongolia. Forest Science and Technology, 15(4),
urali, 30(4), 723–734. https://doi.org/10.1007/s12210-019-00833-5 210–220. https://doi.org/10.1080/21580103.2019.1667880
Jafari, R., & Bakhshandehmehr, L. (2013). Quantitative mapping and Mazumdar, J., & Paul, S. K. (2017). A spatially explicit method for identifi-
assessment of environmentally sensitive areas to desertification in cation of vulnerable hotspots of Odisha, India from potential cyclones.
Central Iran. Land Degradation & Development, 27(2), 108–119. International Journal of Disaster Risk Reduction, 27, 391–405. https://
Jiang, L., Jiapaer, G., Bao, A., Kurban, A., Guo, H., Zheng, G., & Maeyer, P. D. doi.org/10.1016/j.ijdrr.2017.11.001
(2019). Monitoring the long-term desertification process and assessing Memon, N., Vyas, A., & FT, D. (2016). Detection and analysis of desertifi-
the relative roles of its drivers in Central Asia. Ecological Indicators, 104, cation sensitive areas in Sabarkantha District of Gujarat. International
195–208. https://doi.org/10.1016/j.ecolind.2019.04.067 Journal of Engineering Research and Technology, 4(23), 1–5.
Jiang, L., Jiapaer, G., Bao, A., Li, Y., & Guo, H. (2019). Assessing land degra- Moharana, P. C., Santra, P., Singh, D. V., Kumar, S., Goyal, R. K.,
dation and quantifying its drivers in the Amudarya River delta. Ecologi- Machiwal, D., & Yadav, O. P. (2016). ICAR-central arid zone research
cal Indicators, 107, 105595. https://doi.org/10.1016/j.ecolind.2019. institute, Jodhpur: Erosion processes and desertification in the Thar
105595 Desert of India. Proceedings of the Indian National Science Academy, 82
Kannan, A. (2014). United Nations Convention to Combat Desertification: (3), 1117–1140. https://doi.org/10.16943/ptinsa/2016/48507
Issues and Challenges. E-International Relaitions, United Nations Con- Ouachoua, R., & Karkouri, J. A. (2020). Assessing environmental sensitivity
vention to Combat Desertification, 1–15.UN areas to desertification using MEDALUS model in Ziz-Rheris water-
Kar, A., & Kumar, A. (2020). Evolution of arid landscape in India and likely shed, Morocco. International Journal of Scientific Research in Multi-
impact of future climate change. Episodes, 43, 511–523. https://doi. disciplinary Studies, 6(8), 18–26.
org/10.18814/epiiugs/2020/020033 Park, S., & Kim, J. (2019). Landslide susceptibility mapping based on ran-
Kéfi, S., Rietkerk, M., Alados, C. L., Pueyo, Y., Papanastasis, V. P., dom forest and boosted regression tree models, and a comparison of
ElAich, A., & De Ruiter, P. C. (2007). Spatial vegetation patterns and their performance. Applied Sciences, 9, 1–19. https://doi.org/10.3390/
imminent desertification in Mediterranean arid ecosystems. Nature, app9050942
449(7159), 213–217. https://doi.org/10.1038/nature06111 Ponnuchamy, R., Pragasam, A., Aravajy, S., Patel, P., Das, L., & Anupama, K.
Kosmas, C., Kirkby, M. J., & Geeson, N. (Eds.). (1999). The Medalus Project: (2013). A floristic study on herbs and climbing plants at Puducherry,
Mediterranean desertification and land use: Manual on key indicators of South India: An approach to biodiversity conservation and regenera-
desertification and mapping environmentally sensitive areas to desertifica- tion through eco-restoration. Check List, 9(3), 555–600. https://doi.
tion. Directorate-General for Research and Innovation (European org/10.15560/9.3.555
Commission): Brussels Prokop, P., & Poręba, G. J. (2012). Soil erosion associated with an
Kosmas, C., Kairis, O., Karavitis, C., Ritsema, C., Salvati, L., Acikalin, S., … upland farming system under population pressure in Northeast
Ziogas, A. (2014). Evaluation and selection of indicators for land degra- India. Land Degradation & Development, 23(4), 310–321.
dation and desertification monitoring: Methodological approach. Envi- https://doi.org/10.1002/ldr.2147
ronmental Management, 54, 951–970. https://doi.org/10.1007/ Raghuvanshi, M. S., Moharana, P. C., Saxena, A., & Saha, D. (2020). Pasture
s00267-013-0109-6 and land degradation by weedy invasion on frost heaves in
Krishnan, V., Murugaiya, R., Shanmugham, R., & Mariappan, M. (2015 June Changthang-Ladakh: Retrospect and Prospects. Food and Scientific
22-July 5). Assessing the impact of natural factors on desertification in Reports, 1(4), 46–49.
Tamilnadu, India using integrated remote sensing[Paper presentation]. Ravindranath, N. H., Rao, S., Sharma, N., Nair, M., Rao, A. S., Malaviya, S.,
1st International Electronic Conference on Remote Sensing, MDPI … Krishna, N. (2011). Climate change vulnerability profiles for North
2015, Basel, Switzerland. https://doi.org/10.3390/ecrs-1-d013 East India. Current Science, 101(3), 384–394. http://www.jstor.org/
Kumar, B. P., Babu, K. R., Rajasekhar, M., & Ramachandra, M. (2020). Iden- stable/24078517.
tification of land degradation hotspots in semiarid region of Anantapur Ritse, V., Basumatary, H., & Susan, A. (2020). Monitoring land use land
district, Southern India, using geospatial modeling approaches. Model- cover changes in the eastern Himalayan landscape of Nagaland,
ing Earth Systems and Environment, 6, 1841–1852. https://doi.org/10. Northeast India. Environmental Monitoring and Assessment, 192(711),
1007/s40808-020-00794-x 1–17. https://doi.org/10.1007/s10661-020-08674-8
RAJBANSHI AND DAS 17

Safriel, U. N. (2009). Status of desertification in the Mediterranean region. network modeling of CO2 efflux in soils: Impact of land use change in
In Water scarcity, land degradation and desertification in the Mediterra- subtropical India (Meghalaya). Ecological Indicators, 93, 129–141.
nean region (pp. 33–73). Dordrecht: Springer. https://doi.org/10.1016/j.ecolind.2018.04.077
Saji, G., & Jayakumar, S. (2017). Modeling the present and future desertifi- UNCCD. (1994). United Nations Convention to Combat Desertification in
cation risk state: A case study in Kolli Hill, Eastern Ghats of Tamil those countries experiencing serious drought and/or desertification,
Nadu, India. Proceedings of the National Academy of Sciences, India particularly in Africa. Paris: UNCCD
Section A: Physical Sciences, 87(4), 951–960. https://doi.org/10.1007/ Uzuner, Ç., & Dengiz, O. (2020). Desertification risk assessment in Turkey
s40010-017-0462-6 based on environmentally sensitive areas. Ecological Indicators, 114,
Salunkhe, S. S., Bera, A. K., Rao, S. S., Venkataraman, V. R., Raj, U., & 106295. https://doi.org/10.1016/j.ecolind.2020.106295
Murthy, Y. K. (2018). Evaluation of indicators for desertification risk Varghese, N., & Singh, N. P. (2016). Linkages between land use changes,
assessment in part of Thar Desert region of Rajasthan using geospatial desertification and human development in the Thar Desert Region of
techniques. Journal of Earth System Science, 127(8), 1–24. https://doi. India. Land Use Policy, 51, 18–25. https://doi.org/10.1016/j.
org/10.1007/s12040-018-1016-2 landusepol.2015.11.001
Sarparast, M., Ownegh, M., & Sepehr, A. (2020). Evaluating the impacts of Vorovencii, I. (2017). Applying the change vector analysis technique to
combating-action programs on desertification hazard trends: A case study assess the desertification risk in the south-west of Romania in the
of Taybad-Bakharz region, northeastern Iran. Environmental and Sustain- period 1984–2011. Environmental Monitoring and Assessment, 189, 1–
ability Indicators, 7, 100043. https://doi.org/10.1016/j.indic.2020.100043 18. https://doi.org/10.1007/s10661-017-6234-6
Sengupta, M. (2020). Conserving and commercialisingforests: Tribal Wang, F., Pan, X., Wang, D., Shen, C., & Lu, Q. (2013). Combating desertifi-
women and subjectivity in Bagafa Forest of Tripura (Northeast India) cation in China: Past, present and future. Land Use Policy, 31, 311–
conserving and commercialising forests: Tribal. Gender, Place and Cul- 313. https://doi.org/10.1016/j.landusepol.2012.07.010
ture, 28, 1–22. https://doi.org/10.1080/0966369X.2020.1734539 Wang, J. (2020). Determining the most accurate program for the Mann-
Shalini Tirkey, A., Pandey, A. C., & Nathawat, M. S. (2013). Use of satellite Kendall method in detecting climate mutation. Theoretical and Applied Cli-
data, GIS and RUSLE for estimation of average annual soil loss in matology, 142, 847–854. https://doi.org/10.1007/s00704-020-03333-x
Daltonganj watershed of Jharkhand (India). Journal of Remote Sensing Wang, S. Q., Zheng, X. Q., & Zang, X. B. (2012). Accuracy assessments of
Technology, 1(1), 20–30. https://doi.org/10.18005/JRST0101004 land use change simulation based on Markov-cellular automata model.
Shoba, P., & Ramakrishnan, S. S. (2016). Modeling the contributing factors Procedia Environmental Sciences, 13(2011), 1238–1245. https://doi.
of desertification and evaluating their relationships to the soil degrada- org/10.1016/j.proenv.2012.01.117
tion process through geomatic techniques. Solid Earth, 7, 341–354. Xue, X., Guo, J., Han, B., Sun, Q., & Liu, L. (2009). The effect of climate
https://doi.org/10.5194/se-7-341-2016 warming and permafrost thaw on desertification in the Qinghai–
Singh, A., Santra, P., Kumar, M., Panwar, N., & Meghwal, P. R. (2016). Spa- Tibetan Plateau. Geomorphology, 108(3–4), 182–190. https://doi.org/
tial assessment of soil organic carbon and physicochemical properties 10.1016/j.geomorph.2009.01.004
in a horticultural orchard at arid zone of India using geostatistical Yu, H., Wang, L., Wang, Z., Ren, C., & Zhang, B. (2019). Using Landsat OLI
approaches. Environmental Monitoring and Assessment, 188, 529. and random forest to assess grassland degradation with aboveground
https://doi.org/10.1007/s10661-016-5522-x net primary production and electrical conductivity data. ISPRS Interna-
Singh, G. (2009). Salinity-related desertification and management strate- tional Journal of Geo-Information, 8(11), 511. https://doi.
gies: Indian experience. Land Degradation & Development, 20, 367– org/10.3390/ijgi8110511
385. https://doi.org/10.1002/ldr.933 Yusuf, T. U., Ameta, S. K., Usman, A., Tukur, A., & Hamza, Y. G. (2020).
Sivakumar, M. V. K. (2007). Interactions between climate and desertifica- Desertification in Western Rajasthan (India): Causes, effects and miti-
tion. Agricultural and Forest Meteorology, 142, 143–155. https://doi. gation measures. Asian Journal of Geological Research, 3(4), 26–36.
org/10.1016/j.agrformet.2006.03.025 https://journalajoger.com/index.php/AJOGER/article/view/30115.
Sun, B., Gao, Z., & Li, Z. (2014). Dynamic and dry/wet variation of climate Zakerinejad, R., & Masoudi, M. (2019). Quantitative mapping of desertifi-
in the potential extent of desertification in China during 1981–2010. cation risk using the modified MEDALUS model: A case study in the
Environmental Earth Sciences, 73, 3717–3729. https://doi.org/10. Mazayejan plain, Southwest Iran. AUC GEOGRAPHICA, 54, 232–239.
1007/s12665-014-3659-x https://doi.org/10.14712/23361980.2019.20
Sur, K., & Chauhan, P. (2019). Dynamic trend of land Zhang, Z., & Huisingh, D. (2018). Combating desertification in China: Mon-
degradation/restoration along Indira Gandhi Canal command area in itoring, control, management and revegetation. Journal of Cleaner Pro-
Jaisalmer District, Rajasthan, India: A case study. Environmental Earth duction, 182, 765–775. https://doi.org/10.1016/j.jclepro.2018.01.233
Sciences, 78, 1–11. https://doi.org/10.1007/s12665-019-8488-5
Taghizadeh-Mehrjardi, R., Mahdianpari, M., Mohammadimanesh, F.,
Behrens, T., Toomanian, N., Scholten, T., & Schmidt, K. (2020). Multi- SUPPORTING INF ORMATION
task convolutional neural networks outperformed random forest for Additional supporting information may be found online in the
mapping soil particle size fractions in central Iran. Geoderma, 376,
Supporting Information section at the end of this article.
114552. https://doi.org/10.1016/j.geoderma.2020.114552
TERI (2010). Rajasthan State Action Plan on Climate Change. New Delhi,
India: The Energy and Resources Institute.
Tewari, J. C., Pareek, K., Raghuvanshi, M. S., Kumar, P., & Roy, M. M. How to cite this article: Rajbanshi J, Das S. Monitoring land
(2016). Fodder production system-a major challenge in cold arid region sensitivity to desertification using the ESAI approach and
of Ladakh, India. MOJ Ecology & Environmental Sciences, 1(1), 1–7. evaluation of the key indicators: A spatio-temporal study in
https://doi.org/10.15406/mojes.2016.01.00005
India. Land Degrad Dev. 2021;1–17. https://doi.org/10.1002/
Thangavel, R., Kanchikerimath, M., Sudharsanam, A., Ayyanadar, A.,
Karunanithi, R., Deshmukh, N. A., & Vanao, N. S. (2018). Evaluating ldr.3965
organic carbon fractions, temperature sensitivity and artificial neural

You might also like