|
LAND SURFACE TEMPERATURE RESPONSES TO THE LAND COVER DYNAMICS IN WESTERN GHATS
(PDF)
T V Ramachandra, Srijith A H and Bharath S
Energy and Wetlands Research Group,
Centre for Ecological Sciences,
Indian Institute of Science - 560012
6. Results and Discussions
6.1. Spatio temporal LU change analysis: The LU maps have been prepared from MODIS EVI data for the years 2001, 2008, 2005 and 2016. The Enhanced Vegetation Index (EVI) is an optimised vegetation index designed to monitor vegetation cover in high biomass region due to its reduced atmospheric effects (Franklin et al., 2002). The LU maps at different time scale based on the states covering Western Ghats have been prepared. Table 6.1 describes the relative percentage of study area to area of state. Fig. 6.1 shows the area of state to study area in that state.
6.1.1 Time Series MODIS EVI based LU change Analysis from 2001 to 2016 – Kerala:
Kerala, located towards the Southern end of the Western Ghats supports a rich flora and fauna. The state in the last few decades has been undergoing rapid changes in the land use due to rise in commercial plantations. This has led to serious impacts on forest eco-system with shrinkage in the state’s forest cover and the loss of structural integrity of the remaining forest (Kumar et al., 2005). The forest cover of Kerala is largely spread over the Western Ghats bordering the state. Even though the state has a recorded forest cover of 11125.59 sq. km.(28.9% of the total area of the state), the actual forest cover is only 9400 sq. km. The state has a large area under plantation and the diminishing forest has impacted the state with rising temperature, soil erosion, silting of rivers, water scarcity, and drought conditions etc. Fig. 6.2 shows the spatial extent of change in land use during the period 2001 to 2016. Coastal regions are majorly dominated by dense commercial plantations by disturbing ecology.
Fig. 6.2: Spatial extent of LU changes in Kerala from 2001 to 2016.
The error matrix has been computed but has not been shown due to the large spatial and temporal nature of the data. The kappa has been ranging from 0.76 to 0.83 for different zones of Kerala. Table 6.2 shows the area under different land use classes with respect to the total area under study of the state. Fig. 6.3 shows area under different land use class in the study period from 2001 to 2016.
The error matrix has been computed but has not been shown due to the large spatial and temporal nature of the data. The kappa has been ranging from 0.76 to 0.83 for different zones of Kerala. Table 6.2 shows the area under different land use classes with respect to the total area under study of the state. Fig. 6.3 shows area under different land use class in the study period from 2001 to 2016.
The study has revealed there has been a decline of 1.99% of forest cover in the period between 2001 and 2016. During the same period plantations has increased by 1.67%. Table 6.4 describes the percentage change in area of different land use classes in the study period. Fig 6.3 shows the percentage change of area of different land use classes in the study period.
6.1.2. Time Series MODIS EVI based LU change Analysis from 2001 to 2016 – Tamil Nadu:
The hill ranges in South Eastern side of Western Ghats falls in the state of Tamil Nadu. Around 19.09% of the state area comes in the ranges of Western Ghats. Tamil Nadu is dominated by agrarian economy but with the rise in population there has been pressure on land. This has led to change in LU patterns with forest being cleared for plantations, settlements, industries, infrastructure etc. As per the current statistics the state is endowed with only 16.3% (12th Five Year Plan, Tamil Nadu) as compared to the target of 33% forest cover set by Govt. of India. Even though there has been decrease in agricultural land, the productivity has increased tending to maintain the balance. The decrease in forest cover has been a concern of threat and there have been serious afforestation and plantation drives all across the state (State of Environment, FSI, 2015). Commercial crops like rubber, teak, eucalyptus, palm, tea, coffee etc. have been planted in the recent years to maintain a green blanket across the state. Natural calamities like cyclone, flood, forest fire etc. have also led to tremendous loss of forest cover. Fig. 6.4 shows the spatial extent of change in LU during the period 2001 to 2016.
The accuracy of the classified maps has been computed by generating error matrix and the coefficient of kappa ranges from 0.79 to 0.85 for different zones of Tamil Nadu. Table 6.5 shows the area under different land use classes with respect to the total area under study of the state.
Table 6.6 shows the annual rate of land use change occurring during different periods 2001-08, 2008-16 and 2001-16.
The study has revealed that there has been an annual decline of 0.63% in forest cover between 2001 and 2016. During the period 2008 – 16 there has been a lot of afforestation drives all across Tamil Nadu, which has been reflected with an annual increase of 1.29% in plantation. Table 6.7 describes the percentage change of each land use category in the study period. The land use change matrix reveals the rate of change of each land use category. The study has revealed about 3.15% of forest cover has been lost between 2001 to 2016 which has been supplemented with the growth of plantations (2.69%) and built – up areas and agricultural lands (0.57%). Fig 6.5 show the percentage change of area in different land use classes in the study period.
6.1.3. Time Series MODIS EVI based LU change Analysis from 2001 to 2016 – Karnataka:
The Sahyadri range runs continuously from North to South in the Western part of the state Karnataka. The state has a lush forest cover which is under threat of degradation through natural and man-made factors. The state has about 22.61% its geographical area under forest (State of Environment, FSI, 2015). Over the last few decades, these forests are threatened due to increased deforestation, rise in commercial plantations, settlements along the forest, construction of dams etc. Around 22% of the geographic area of the state has been part of this study. In the last few years there has been a rampant rise in commercial plantations like teak, acacia, coconut, rubber, eucalyptus etc. leading to the loss of forest cover. Fig. 6.6 shows the spatial extent of LU change in the period from 2001 to 2016.
The accuracy of the classified map has been computed by generating error matrix and the coefficient of kappa ranges from 0.77 to 0.82 for different zones of Karnataka. Table 6.8 shows the area under different land use classes with respect to the total area under study of the state.
The study reveals there has been continuous decline in forest cover with increase in scrub vegetation and plantations which can be attributed to increase in human settlements, commercial plantations. Even though plantations provide a green cover they destroy the structural integrity of the forest. The study shows about 3.77% forest cover has been lost in the period between 2001 and 2016. The loss has been dominating in the regions adjoining coast and plains of Western Ghats in Karnataka. Table 6.10 describes the percentage change of area in each land use classes in the study period. Fig 6.7 shows the percentage change of area under different land use classes in the study period
7.1.4. Time Series MODIS EVI based LU change Analysis from 2001 to 2016 – Goa:
The Western Ghats cover about 50% of the geographic area of Goa. The state has a rich forest cover with 59.94% of its total geographical area under forest (State of Forest Report, FSI, 2015) which satisfies the ecological target of 33% of forest cover set by Govt. of India. The forests of Goa are typical of the vegetation of Western Ghats. Over the recent years, there has been decline in forest cover due to increase in mining, settlements, plantations etc. The plantations in Goa are coconut, teak, spices, acacia, eucalyptus, bamboo etc. Fig. 6.8 shows the spatial extent of land use change from 2001 to 2016. The accuracy of the classified map has been computed by generating error matrix and the coefficient of kappa ranges from 0.79 to 0.82 for different zones of Goa. Table 6.11 shows the area under different land use classes with respect to the total area under study of the state.
The annual rate of change of each land use class in different time periods 2001-08, 2008-16 and 2001-16 is shown in Table 6.12. The study reveals that there has been a strenuous rise in forest plantation in the period between 2008 and 2016 at the loss of forest and open spaces. The regions covered with scrub vegetation are also undergoing land use changes at 0.25% annually.
The study shows about 1.97% of forest cover has been lost between 2001 and 2016. This loss has been identified to have occurred in fringes adjoining settlements, plantations etc. Table 6.13 describes the percentage change in area of each land use classes in the study period. Fig 6.9 shows the percentage change of area under different land use classes in the study period
6.1.5. Time Series MODIS EVI based LU change Analysis from 2001 to 2016 – Maharashtra:
The Western Ghats run about 720 km continuously from North to South of the state making Maharashtra with the longest stretch of continuous range. About 20% of the geographic area of the state comes under the purview of this study. The forest cover of the state is estimated to be around 16.45% (State of Environment, FSI, 2015) which is estimated to have increased over the recent years. This can be attributed to rise in commercial plantation like eucalyptus, bamboo, acacia etc. Fig. 6.12 shows the spatial extent of changes in land use from 2001 to 2016. The accuracy of the classified map has been computed by generating error matrix and the coefficient of kappa ranges from 0.80 to 0.84 for different zones of Maharashtra. Table 6.14 shows the area under different land use classes with respect to the total area under study of the state.
The annual rate of change of each land use class in different time periods 2001-08, 2008-16 and 2001-16 is shown in Table 6.15. Even though dense evergreen forest cover is comparatively less in this region, the region has a rich deciduous forest cover which is declining at the rate of 1.12%. The study reveals that there has been a tremendous increase plantation in the period between 2008 and 2016. The regions covered with scrub vegetation are also undergoing land use change at the rate of 0.11% due to tremendous pressure from commercial plantation; agricultural farms etc. at the rate of 0.35% annually.
The percentage change in area of each land use class in the study period is described in Table 6.16. There has been a loss of about 1.61% of forest cover between 2001 and 2016 which has been highest when compared with other land use classes. There has been an increase of 0.25% in plantations between the period 2001 and 2016. Even though the region has been facing severe crop loss, there has been rise in agricultural class (Others) which has been attributed to the rise in cultivation of cereals, horticulture, cotton, vegetables, fruits etc. Fig. 6.12 shows the percentage change of area under different land use classes in the study period.
6.1.6. Time Series MODIS EVI based LU change Analysis from 2001 to 2016 – Gujarat: The Western Ghats cover about 2% of the geographic area of the State. This region is dominated with open forests and scrub vegetation. Dense forest cover types like evergreen, moist deciduous etc. are not found in this region. The forest cover of this region is estimated to be around 11.46% to the total geographic area of the state which falls well below the ecological target of 33% forest cover as set by the Govt. of India. Fig. 6.13 shows the spatial extent of change in land use from 2001 to 2016.
The accuracy of the classified map has been computed by generating error matrix and the coefficient of kappa ranges from 0.83 to 0.87 for different zones of Gujarat. Table 6.17 shows the area under different land use classes with respect to the total area under study of the state.
The study reveals an increase in area under plantations and agriculture. The annual rate of change of each land use class in different time periods 2001-08, 2008-16 and 2001-16 is shown in Table 6.18. The annual rate of increase of Others class is around 0.22% which has been attributed to the rise in horticulture, fruits (grape, pomegranate), jatropa etc. The region is facing a rise in cultivation of plantation crops like bamboo, date palm, eucalyptus etc. which has been identified with increase in area under plantation.
The study has revealed a decline of 1.78% of scrub vegetation in the period from 2001 to 2016. The spatial maps reveal increase in plantation has occurred by clearing of forest. Fig. 6.14 shows the percentage change in area of different land use class in the study period.
6.1.7. Time Series MODIS EVI based LU change Analysis from 2001 to 2016 – Western Ghats:
The land use of Western Ghats is compiled by aggregating the land use maps based on the agro-climatic and administrative boundaries. There have been scenarios where confusions are found to exist between different land use classes which eventually led the analysis to be carried out on a smaller scale. The confusions existing during analysis have been overcome by carrying out multiple clustering and reclassification of the image. Fig. 6.14 shows the spatial extent of land use change in Western Ghats between 2001 and 2016. Table 6.20 describes the area under different land use classes with respect to the total area under study. It may be noted there has been decrease in area under forest which indicates there is a decline in forest. The forest loss can be attributed to rise in plantation, agricultural farms, settlements etc.
Table 6.21 describes the annual rate of change of each land use class. The annual rate of decline of forest is 0.5% in the period between 2001 and 2016. During the same period the annual rate of increase in plantation is 0.77% in the entire Western Ghats. There has also been annual increase of 0.29% in Others class, which signifies the increase in area under built up, agricultural farms etc.
The percentage change in each land use during the study period is described in Table 6.22. The study reveals about 1.61% forest cover has been lost in the period 2001-16. Even though there has been increase of 0.59% of scrub vegetation in the period 2001-08, there has been a tremendous decline of about 1.33% in the period 2008-16. This has been attributed with scrub vegetation being turned into agricultural farms, settlements i.e. Others class. The study also reveals the entire belt of Western Ghats is under tremendous pressure due to increasing population, changing land use. There has not been any significant change in water-bodies.
6.2. Spatio temporal LST Analysis:
The temperature maps have been prepared from remote sensed images of LST data. Detailed documentation regarding compositing, preparation of data is available in NASA MODIS website. (MODIS 1999)
6.2.1. Time Series MODIS LST based LST change Analysis from 2001 to 2016 – Western Ghats:
LST is a crucial variable for environment and climate studies. This section describes the temporal dynamics of land surface temperature. The study has revealed an increase in mean temperature over the entire study area. Fig. 6.16 shows the temporal dynamics of Land Surface Temperature from 2001 to 2016. The study reveals LST values directly dependent on vegetation cover. The regions with thick vegetation cover like the states of Kerala, Karnataka, and Goa show a relatively lesser LST when compared with states like Gujarat, Tamil Nadu where vegetation cover is sparse. It has also been observed that LST varies with land use of the region i.e. water-bodies exhibit a lower LST during day time and a higher LST is observed in ‘Others’ class (i.e. built up, open spaces). There have been prominent differences in LST between different vegetation classes i.e. plantations and forest with plantations showing a relatively higher temperature than forest cover. The coast of Kerala dominated with plantations as depicted in land use map (Fig. 6.16) shows a relatively higher temperature than the hills of the same region which found to be similar in other regions of the study area. The plains of the study area show a relatively higher temperature when compared with coast and hills which may be attributed to the presence of open spaces and agricultural farms. The coastal side of Western Ghats show a relatively lesser temperature due to presence of sea (Ado., 1992).
6.2.2. Validation of LST with Ground data:
The LST values have been compared and validated with ground data obtained from NASA climatology grid. The validation has been carried out by taking 13 representative locations all across the study area. The places taken up for validation are shown in Fig. 6.17.
The maximum, minimum and mean temperature of satellite data is compared with maximum, minimum and mean temperature of ground data. The ground data is the air temperature of the region at a height of 2m from ground, whereas satellite data gives the land surface temperature. The LST of MODIS during day time is of 2 to 3°C is higher than the air temperature and at night time vice versa (Colombi et. al., 2007). Fig. 6.18 shows the comparison between ground temperature and satellite temperature.
6.3. STATISTICAL ANALYSIS:
6.3.1. COEFFICIENT OF CORRELATION (r):
A coefficient of correlation is a statistical measure to quantify relationship and dependence between two variables (Schielzeth, 2010). The coefficient of correlation lies between -1 and 1. If a positive relationship exists between the variables, the coefficient is 1 and in a negative relationship, the coefficient is -1.
A coefficient of correlation (r) of 0.8 to 1 indicates a very strong relationship, 0.6 to 0.8 indicates a strong relationship, 0.4 to 0.59 indicates a substantial relationship and 0 to 0.4 negligible or low relationship. In this study, the coefficient of correlation is determined between rate of change in forest, rate of change in agriculture, built up and change in temperature. Table 6.23 describes the relationship between different variables and the coefficient of correlation.
Relationship between rate of change in forest cover and change in temperature:
The study has shown forest cover has a direct impact on the temperature of the surroundings. During the period 2001 to 2008, there has been a decline in forest by about 0.82% and in 2008-16 the decline in forest cover has been 1.67%. The study between rate of change in forest cover and change in temperature has shown a strong correlation of 0.83, 0.72 and 0.78 between the variables indicating declining forest cover has direct relationship with increase in temperature of the region. Fig. 6.19 shows the plot between rate of change in forest and change in temperature. The rate of change in forest cover and change in temperature show a second order polynomial relationship across all the locations of Western Ghats which indicates decline in forest, increases the temperature.
Relationship between %change in agricultural land, built-up and change in temperature:
The study has shown an increase in built up and agricultural land has direct impact on rise in temperature. During the period 2001-08, the rise in agricultural and built-up was only 0.09% whereas in the period 2008-16, the rise had been 1.03% indicating a huge area of the land has been converted to built-up and agricultural farms. The coefficient of correlation for the periods has shown a positive relationship indicating rise in agricultural lands and built-up rises the temperature of the region. The rate of increase in built up and agricultural lands have shown a second order polynomial relationship across all the locations of Western Ghats indicating the increase in built up and agricultural lands, increases the temperature.
6.3.2. Multiple Variable Analyses:
Stepwise regression of multi variables were carried out to find out the probable relationship of LST with different independent variables (changes in LU categories). The dependent variable in this study is the change in temperature which is dependent on change in forest cover, agricultural farms and built up areas. The multiple variable analyses helps in understanding the overall phenomena through statistical modelling of the relative contribution of each of the variables to the process. This is usually done by modelling the present rate of changes to understand the dynamics of change and simulate for future.
In this study, the variables are rate of change in forest cover (x1) and rate of change in agriculture and built-up(x2) have been modelled to understand the change of temperature (y) for the periods: 2001-08, 2008-16 and 2001-16.
The analysis of the variables have shown a high degree of relationship between rate of change in forest cover, rate of change in built up and agricultural etc. with the changes in temperature (LST). The multiple variable analyses have shown strong correlation during different periods of study. For the study period 2001 to 2008, it shows a correlation of 0.79 indicating a strong relationship between the variables. This shows that decline in the forest cover and increase in built-up and agricultural areas have a direct effect on the rise of temperature. In the period 2008-16, the correlation is 0.71, indicating a strong relationship between the variables. The year 2016 was hottest year of the century, which eventually led to higher rise in temperature values than predicted. This has led to reducing the relationship indicating there could be other parameters associated with rise in temperature other than changing land use. Even though there could be other parameters for increase of temperature in the region, the dominant factors are reduction in forest cover and increase in area under built-up and settlements. The period 2001-16 has an overall correlation of 0.71 indicating a strong relationship between the dependents and independents. The relationship between the variables has been formulated into an equation, to understand the dynamics of the period. Table 6.24 describes the variables and the relationship between them.
|
