Land degradation leads to alteration of ecological and economic
functions due to a decrease in productivity and quality of the
land. The aim of the present study was to assess land degradation
with the help of geospatial technology – remote sensing (RS) and
geographical information system (GIS) – in Bathinda district,
Punjab. The severity of land degradation was estimated
quantitatively by analyzing the physico-chemical parameters in the
laboratory to determine saline or salt-free soils and calcareous or sodic soils and
further correlating them with satellite-based studies. The pH
varied between 7.37 and 8.59, electrical conductivity (EC) between
1.97 and 8.78
Land degradation is the process that makes land unsuitable for human beings as well as for soil ecosystems (Kimpe and Warkentin, 1998), occurs in arid, semi-arid and sub-humid areas as a result of anthropogenic activities and climatic variations (Barbero-Sierra et al., 2015), and eventually subjects livelihoods and sustainable development to severe risks (Fleskens and Stringer, 2014). Land use and land cover (LULC) change is a prime issue for scientists concerned with global environmental change (Muñoz-Rojas et al., 2015; Ochoa et al., 2016). Land use activities have a considerable influence on the people, posing serious consequences for social, economic and ecological aspects of human society (Burchinal, 1989; FAO, 1997). The alteration in ecological and economic functions due to the decrease in the productivity and quality of the land (Hill et al., 2005) can lead to decline in the biological productivity of land due to climate change and human activities (Zhang et al., 2014). Land degradation poses a great threat to food security and damages the environmental safety of land as well as influences the sustainable development of society and economy (Zhao et al., 2013). Degradation can lead to exhaustion of other natural resources in both developed and developing countries and affect arid, dry and even sub-humid areas (Omuto et al., 2014; Stringer and Harris, 2014). Soil degradation takes place not only as a result of interaction between physico-chemical and biological factors comprising topography, soil properties and climatic features (Brevik et al., 2015; Taguas et al., 2015) but also includes human factors and land use management practices (Khaledian et al., 2017; Camprubi et al., 2015; Costa et al., 2015; Cerda et al., 2016). Inappropriate land use practices have been attributed as one of the major causes of land degradation by various researchers (Biro et al., 2013; De Souza et al., 2013; Pallavicini et al., 2015; Mohawesh et al., 2015). A major form of inadequate land use is the one leading to environmental land use conflicts that develop on soils used for activities not in compliance with the natural potential of the soil. This is characterized by a deviation between the actual and natural uses set by land capability. Environmental land use conflicts and their consequences for land degradation have been recently studied by various authors (Pacheco et al., 2014, 2016; Valera et al., 2016; Valle Junior et al., 2014a, b, 2015). The resilience and stability of landscape are affected to a great extent by the soil system, which in turn is affected by the inherent balance between inputs and nutrient loss and carbon (Amundson et al., 2015).
Land degradation is a severe problem due to which 1.5 billion people are
threatened (Nachtergaele et al., 2010) and about 1.9 billion hectares of land
and 250 million people are affected worldwide (Low, 2013). There is an
increasing trend in severity of degradation, covering most of the world's
land area, which includes 30 % forests, 20 % cultivated areas and
10 % grasslands undergoing degradation (Bai et al., 2008). According to
Barrett and Hollington (2006), approximately 10 to 20 million people live on
land affected by salts with poor productivity and under alarming threats of
ecosystem destruction. Every year, approximately 6 million hectares of
agricultural land becomes unproductive due to various processes of soil
degradation (Asio et al., 2009). A target of zero net land degradation at
Rio
According to an estimation of the Indian Council of Agricultural Research
(ICAR, 2010), about 120.40 million hectares (out of 328.73 million hectares)
of land in the country is affected by land degradation. In the state of
Punjab, 2.33 % (1172.84
LULC change analysis is very important for assessment and recognition of potential impacts and the development of planning strategies and land management practices (Leh et al., 2013). The dynamics of LULC not only include the damage by agriculture but also degradation of historic values and conservation functions (Ohta and Nakagoshi, 2011). As a result of LULC changes, the landscape suffers from serious problems such as soil erosion, flood and drought, reducing productivity of land or land degradation (Firdaus et al., 2014). LULC affects soil's physical and chemical properties in such a way that areas going through continuous cultivation but lacking appropriate management practices have low fertility levels due to over-utilization (Majule, 2003). These changes in LULC associated with intensive agricultural practices in semi-arid areas and even in high-rainfall areas (Misana et al., 2003) have a significant impact on soil chemical degradation (Maitima et al., 2009). Being a critical problem in many developing countries (Ananada and Herath, 2003), land degradation has serious consequences on social, economical and environmental aspects of society. Information on LULC changes and their impacts is essential for management of land as it avails key environmental data and information for resource management. Thus, it becomes very important for decision and policy-makers to propose remedial analysis of LULC change and land degradation.
In comparison to other states, the land area of Punjab under different wasteland categories is less (State of Environment Punjab, 2007) but land degradation assessment and monitoring is essential to improve understanding and assistance in decision-making processes. Hence, this research uses both remote sensing methods and physico-chemical analysis of soil for assessing the severity of land degradation. Being a serious problem worldwide, the management of degraded land can be carried out using a more appropriate approach by evaluating the spatial variability in soil properties, including chemical properties of degraded land, and mapping such variations (Cambardella et al., 1994). For example, soil pH is an important property determining the availability and toxicity of nutrients. A better consideration of the spatial variability in soil properties like pH, EC and organic carbon would facilitate advanced agricultural and environmental management practices by identification of proper sites for management (Cambardella et al., 1994). Digital soil mapping is critical when quantifying the relationship between soil types and their associated surrounding environment. Mapping soil characteristics allows the spatial distribution evaluation of physico-chemical properties and presentation in a form that users can deal with efficiently and interpret in a better way (Sheng, 2010).
In addition to land degradation assessment and land use cover dynamics, this research also included the study of spatial variability in soil pH, electrical conductivity (EC) and alkalinity using the digital soil mapping interpolation method in agricultural and non agricultural soils of the Bathinda district in Punjab, India. The results obtained by both approaches have been correlated to get a better picture of the extent of land degradation in Bathinda, a semi-arid town in northwestern India.
Overall, there is a dearth of literature regarding land degradation
studies for the Bathinda region of Punjab, India. Further, sufficient
quantitative data on current land use practices are lacking. Whatever
data exist are scattered and are not easily available to researchers,
planners and policy-makers, hindering the complete assessment of land
degradation problems. So far, no study in which
multispectral Landsat satellite images (Landsat_7 ETM to perform spatial–temporal analysis of selected soil
parameters such as pH, EC and alkalinity
as well as the nature of soil (saline or salt-free and
calcareous or sodic). to study the spatial variability in selected soil
parameters through digital soil mapping using the inverse distance
weighting (IDW) interpolation method. to integrate satellite data and field-based soil data to
assess severity of land degradation quantitatively in the Bathinda
district, Punjab. to carry out change detection analysis of temporal
satellite datasets of Landsat 7 ETM to perform correlation analysis between soil parameters and
satellite data for 2014.
Bathinda is one of the historical and important cotton-producing towns of
northwestern India, situated in the Malwa region of southern Punjab. The
total area extent of the district is about 336 725
However, one of the major environmental concerns of the Bathinda district is
soil degradation including soil erosion, waterlogging, sand dunes and
salinization. Salinization is one of the most serious problems and occurs naturally in irrigated water and soil,
particularly in semi-arid and arid regions. (Sheng, 2010). The groundwater irrigation practiced predominantly
in the Bathinda district is one of the main factors of soil salinization,
which in turn leads to land degradation (El Baroudy, 2011). Additional salts
may be introduced through agricultural fertilizers. Salt-affected soils are
widespread, especially in arid, semi-arid and some sub-humid regions. The two
main constraints that are present in irrigated agricultural lands include
salinity and waterlogging (Koshal, 2012), both widespread in the study area.
The Bathinda region shows typical conditions of desertification and soil
salinization; hence, effective means of combating soil salinization and
desertification need to be pursued. From the study area, a total of 21 sites
(Fig. 1) were randomly selected for soil sample collection (at a depth of
15–25
Sampling sites of the study area (Bathinda district).
The least-clouded multispectral Landsat satellite images of the years 2000
and 2014 (Table 1) were procured from the United States Geological Survey
(USGS, 2015;
Details of the satellite images used for study.
Satellite image processing software, Earth Resources Data Analysis System (ERDAS) version 9.1 Imagine (Hexagon Geospatial – formerly ERDAS, Inc.) and ArcGIS 10.1 (Environmental Systems Research Institute (ESRI) product) were used to process, classify, analyse and display the satellite images.
The false colour composite (FCC) of multi-temporal Landsat satellite
images of the years 2000 and 2014 was generated on a
Three chemical parameters – pH, EC and alkalinity – were analysed for 21 soil samples in the study area as represented in Table 2.
Analysis of soil parameters.
The pH was determined in accordance with the procedure of IS: 2720,
part 1-1983 in which a pH meter (Oakton PC2700) was used to record the pH
in an extract of
The measurement of EC was carried out in accordance
with IS: 14 767-2000 using an EC meter (Systronic water analyser
371). An extract of soil sample or supernatant liquid of
Alkalinity of the soil samples was measured with the titration method,
using 0.05 N
Digital soil mapping can be used for the prediction of individual soil properties in large areas over space, generating maps in digital format in a rapid, effective, efficient and low-cost manner (Sheng, 2010). Severity of land degradation was shown in quantitative terms within the GIS environment using statistical analysis in ArcGIS. There are several interpolation methods such as Kriging (Sheng, 2010) and IDW that estimate cell values by averaging the values of sample data points in the neighbourhood of each processing cell. For the preparation of soil maps in the present study, IDW was used to estimate the variables pH, EC and alkalinity over space. The IDW method is a non-geostatistical interpolation method based on the fact that the local impact of a variable gradually disappears with the increase in distance (Liu, 2016). The methodology incorporates soil scientific knowledge and provides a reliable logical framework to the mapping of continuous surfaces in a quantitative approach (Mora-Vallejo, 2008).
Pearson's correlation analysis was carried out in the present study to
analyse the relationship between digital number (DN) values of satellite data and
physico-chemical parameters of soil. The correlation was not based on
time series of digital number (DN) or on average values; rather, the
DN values corresponding to geographical coordinates of sampling sites
were considered for the present study. The Pearson's correlation
coefficient determines the strength of a linear
association or relationship between two data variables. Statistical
significance of the correlation coefficients between DN and pH and between EC and pH,
The FCC image was prepared for the 2000
Landsat ETM
FCC of Landsat ETM
Followed by preparation of FCCs and visual interpretation, LULC maps of Bathinda were prepared for 2000 and 2014 as given in Fig. 3a and b using the Iterative Self-Organizing Data Analysis Technique (ISODATA).
LULC map of Bathinda using unsupervised classification for
After unsupervised classification, in which an insight was gained about spectral variability in classes, supervised classification of FCC images of both the temporal datasets of the Landsat image was carried out for 2000 and 2014 as given in Fig. 4a and b using the MLC algorithm. The agricultural lands with crops and without crops were assigned dark green and light green, respectively. The blue colour was assigned to the water bodies while the red and yellow colours were given to settlements and trees/forest cover, respectively.
LULC map (supervised classification) of Bathinda for
Accuracy assessment is often carried out to quantify the reliability of a classified image using a reference dataset. Accuracy is a measure of agreement between standard information at a given location and the information at same location on the classified image. Kappa is a measure of the difference between the observed agreement of two maps as reported by overall accuracy and the agreement that might be contributed solely by chance matching of two maps. The overall classification accuracy of the Landsat image (supervised classification) of 2000 was 96.48 % and the overall kappa statistic was 0.947 (Table 3a). Similarly, for the Landsat image (supervised classification) of 2014, accuracy was 97.66 % with a kappa statistic of 0.964 (Table 3b). These values indicate that the results were appreciably better than random of the values contained in an error matrix (Jensen, 1996).
Accuracy assessment report.
To analyse changes between different land features for a period of 14 years, supervised images of both 2000 and 2014 were used as input images and the changes were highlighted as a 20 % increase (blue colour) and 20 % decrease in the final map (pink colour) depicting change detection (Fig. 5). The other classes included unclassified, unchanged, some increased and some decreased, which were pointed out in black. The increased portion predominantly indicated the expansion of settlements with little increase in vegetation. The decreased portion depicted the decrease in overall land area under agriculture (with or without crops).
Change detection map of the Bathinda district.
Table 4 describes the change detection in the total area
(
Change detection in different LULC (
Post classifications, changes in areas of different ground features in square
kilometres (
The pH, EC and alkalinity values of the soil samples collected from the sampling sites of the study area are given in Table 5.
Soil physico-chemical parameters.
All 21 soil samples collected from different locations of the study area were
mostly alkaline in nature. None of the sampling sites were neutral or acidic.
Soil pH ranged lowest in the military cantonment (pH
Phenolphthalein alkalinity was absent due to the absence of carbonate ions
(
Digital soil mapping was used for the prediction of spatial variability in individual soil properties in the study area, where maps could be generated in digital format in a rapid, effective, efficient, and low-cost manner (Sheng, 2010). Land degradation severity was shown as spatial distribution of pH, EC and alkalinity in quantitative terms via IDW interpolation methods using the statistical analysis tool in the ArcGIS software. Based on the pH values, a soil map for sampling sites was composed with the ArcGIS 10.1 software (Fig. 6a). Similarly, on the basis of EC and alkalinity values, soil maps were composed as shown in Fig. 6b and c, respectively, depicting the severity of land degradation in terms of salinity.
Regions like Gidderbaha, Mehma Sarja and Lakhi jungle to the west of the study area showed a large percentage of soils that were alkaline or sodic in nature (Fig. 6a). Regions like Poohla, the military cantonment and Patel Nagar were calcareous or saline in nature. None of the sampling sites in the study area were acidic.
Based on EC values, it could be observed from the map (Fig. 6b) that in the study area, the maximum area was occupied by slightly saline soils, and some of the regions like Rampura Phul were moderately saline. Only a few of the sites were salt-free.
Based on alkalinity, it was observed from the map (Fig. 6c) that most
of the study area was less alkaline, ranging between 0.07 and 0.131
(
Based on the results obtained from the soil analysis, an attempt was made to
establish correlation (Pearson's
Correlation (
No significant correlation was observed between soil physico-chemical parameters and visible (band 3) and SWIR (band 7) of the satellite image; hence, the values are not reported here. The NIR band (band 5), however, exhibited significant correlation with physico-chemical parameters compared to the rest of the bands, thus proving it to be a better indicator of soil quality.
Digital image classification helped in identifying, delineating and mapping LULC into a number of classes. Classes identified included water bodies, human settlements and built-up areas, agricultural land, trees/forest cover, and barren land. Multispectral data were used for classification and the categorization on a numerical basis related with the spectral pattern of the data for each and every pixel (Lillesand and Kiefer, 1994).
The change detection study deals with the comparison of aerial photographs or satellite images of a region taken at different time periods (Petit et al., 2001), performed on a temporal scale to access landscape change caused by anthropogenic activities on the land (Gibson and Power, 2000). In order to understand landscape patterns for proper land management and decision-making improvements, it is necessary to consider the changes and interactions between human activities and natural phenomena (Prakasam, 2010). Change detection by remote sensing has proven to be a cost- effective method for creating LULC inventories and monitoring land change over time (Coppin et al., 2004; Fry et al., 2011).
The results of the present study revealed that area under barren land
decreased from 68.98
Land degradation results in damage to the physical, chemical and biological soil properties, which leads to a decline in productivity (Chartres, 1987). Thus, considering physico-chemical parameters is essential to assess land degradation.
In this study, most of the soil samples were calcareous or saline, while some
of the other samples were alkaline or sodic
(
Based on the EC values given by Ghassemi et al. (1995), we found that
some parts of the study area were slightly saline, some moderately
saline and a few areas were salt-free. The saline nature of the soil
revealed that these contain sodium as soluble salts (usually as
The study of different soil physico-chemical properties was helpful in the assessment of land degradation (Raina, 1999). Our results showed that soil is affected by inappropriate activities taking place during the past decades, including loss of soil fertility, erosion, soil salinization etc. Due to land degradation, vulnerable populations and fragile ecosystems are affected, with irreversible results (Bisaro et al., 2014). The consequences of land degradation may lead to vegetation loss, soil degradation, and pollution of soil, water and air, which need to be addressed to curb further degradation (Novara et al., 2013; Batjes et al., 2014; Olang et al., 2014; Srinivasarao et al., 2014). Since the whole state of Punjab is intensively cultivated with 80 % of water resources being used for irrigation, the irrigation and overdrafting are some of the main causes of salinization or intrusion of various salts into the soil system (Tiwana, 2007). Groundwater irrigation practised predominantly in the Bathinda district is one of the main factors of soil salinization, which in turn leads to land degradation (El Baroudy, 2011). Salt remains in the soil when water is taken up by plants or lost to evaporation, causing soil salinity (Slinger and Tenison, 2007). The zones receiving low rainfall, with shallow water table depth and hot and dry moisture regions in the irrigated areas of the old alluvial plains are found to be primarily affected by salt soils (Manua and Sharma, 2005).
With the help of the IDW interpolation method, the evaluation of the spatial variability in soil properties (pH, EC and alkalinity) and mapping these variations was carried out by generating soil maps for the precise determination of fluctuations in soil behaviour. Such evaluations would help for optimum fertilizer recommendation by appropriate use of nutrients that contribute in enhancing crop quantity and quality, while being environmentally sustainable (Miransari and Mackenzie, 2010).
Although the degree of salinity was not too high that the soil quality could be affected, some remedial measures must be adopted to prevent any problem that might affect the quality of the land as well as yields in the near future. In such regions where irrigation is a common practice, remote sensing can be used as a valuable tool for obtaining relevant data on soil salinity (Al Khaier, 2003). Symeonakis et al. (2016) proposed that land degradation is a dynamic process and should not be based on static datasets, rather its assessment should incorporate various temporal datasets while devising parameters as indicators of environmental sensitivity. Thus, according to Behmanesh et al. (2016), various criteria for mapping environmentally sensitive areas were climate, vegetation, soil, groundwater and socio-economic characteristics of land over different time periods.
According to our results, the correlation coefficients between soil salinity (EC) and related DN values using Landsat data were helpful in calculating and ascertaining the significant relation between satellite data and soil salinity. The salt-affected soils in arid regions show a high reflectance, especially when a salt crust (whitish colour) is formed. (Alavi-Panah and Goossens, 2001). Furthermore, Mehrjardi et al. (2008) proved that the correlation between DNs of satellite images and soil salinity could be an efficient parameter for assessing the land degradation by preparing soil salinity maps from remotely sensed data.
We found that the area under human settlements and built-up areas expanded by
about 10.56 % between 2000 and 2014. The land under agricultural
practices decreased from 3002.23 to 2399.79
The research data are available upon request to the corresponding author.
The authors declare that they have no conflict of interest.
The authors express their gratitude to the National Remote Sensing Centre, Hyderabad, for providing IRS P6 LISS-III data to carry out the present study. The study was supported by the Central University of Punjab, Bathinda, under the UGC fellowship for the M. Phil.-Ph. D. course for research scholars (CUPB/MPH-PHD/SEES/EVS/2013-14/17). Edited by: Miriam Muñoz-Rojas Reviewed by: Fernando Pacheco, Sunday Adeniyi, Ferhat Türkmen, and two anonymous referees