Land is an area of the earth’s surface, the characteristics of which embrace all reasonably stable or predictable cyclic attributes of biosphere. The atmosphere, soil hydrology, plant and animal populations and the result of past and present human activity exert a significant role on future uses of land by human beings.
Land-use and land cover change (LULCC), also known as human induced land transformation, is a generic term for the human modification of earth’s terrestrial surface (Richards, 1994). Though humans have been modifying land to obtain food and other essentials for thousands of years, current rates, extents and intensities of LULCC are far greater than ever in history, driving unprecedented changes in ecosystems and environmental processes at local, regional and global scales. These changes encompass the greatest environmental concern of human populations today including climate change, biodiversity loss and pollution of water, soil and air. Monitoring and mediating the negative consequences of LULCC while sustaining the production of essential resources has therefore become a major priority of researchers and policy makers around the world.
A geographic information system (GIS) is designed to work with geographic or spatial data. The unique feature of GIS is its ability to link non-geographic data with geographic data. In other words, GIS is both a database system with specific capabilities for spatially referenced data as well as a set of operations for working with the data. GIS technology integrates common database operations such as query and statistical analysis with the visualization and geographic analysis benefits offered by maps (Jensen et al., 1996). GIS is computer based, capable of digitally reproducing and analyzing the features and events occurring on the earth’s surface. Since a large part of data generated today has a geographical reference, it becomes imperative to underline the importance of a system which can represent the given data adequately.
Remote sensing is the science (and to some extent, art) of acquiring information about the earth’s surface without actually being in contact. By recording reflected or emitted energy and processing, analyzing and applying that information, remote sensing offers a handy tool that can be used for procuring information about terrestrial objects from a platform placed at a sufficiently high altitude (ibid).
LULCC, defined as the assemblage of biotic and abiotic components on the Earth’s surface, is one of the most crucial properties of the earth system (Turner et al., 1994). Land cover plays a major role in the carbon cycle, acting as both a source and a sink of carbon. In particular, the rates of deforestation, forestation and regrowth of plant play a significant role in the release and sequestering of carbon and consequently affect atmospheric CO2 concentration and the strength of the greenhouse effect. Finally, land cover also reflects the availability of food, fuel, timber, fibre, and shelter resources for human populations and serves as a critical indicator of other ecosystem services such as biodiversity. Information on land cover is fundamental to many national/global applications including watershed management and agricultural productivity. Thus, the need to monitor land cover is derived from multiple intersecting drivers including the physical climate, ecosystem health and societal needs.
Land cover mapping is a product of the development of aerial photography and remote sensing technology because of the benefits it offers (wide area coverage, frequent revisits, multispectral, multisource and storage in digital format to facilitate subsequent updating and compatibility with GIS technology) that proved to be a very practical and economic means for an accurate classification of land cover Colwell and Weber, 1981). Remote sensing offers an important means of detecting and analyzing temporal changes and since the early 1970s satellite data has been commonly used for change detection studies (Jensen et al., 1996). The satellite images provide a digital mosaic of the spatial arrangement of land cover and vegetation types amenable to computer processing. The use of remote sensing and GIS technologies can greatly facilitate the process of collection, analysis and presentation of resource data. Satellite images or aerial photographs are useful for both the visual assessment of natural resources dynamics occurring at a particular time and space as well as well as the quantitative evaluation of LULCC. The changes in LULCC due to natural and human activities can be observed using the past and the current remotely sensed data, which helps monitor and determine the impact on the ecosystem (Roy and Roy, 2010).
The remote sensing and GIS approach for land cover dynamics is now a widely accepted tool because of its ability to examine spatially referenced objects over time however, the GIS overlaying forest change detection technique has been found to be superior (Roy and Tomar, 2001). Satellite remote sensing provides a synoptic view of forests and their condition n real-time basis, playing a pivotal role in generating information about forest cover, vegetation type and land use changes (Houghton and Woodwell, 1981).
LULCC are perhaps the most prominent as they occur at spatial and temporal scales immediately relevant to our daily existence. Technically, LULCC means quantitative changes in areal extent (increase or decrease) of a given type of land use and land cover respectively. The changes in land use in various spatial and temporal domains are the material expressions and also indicate environmental and human dynamics and their interactions mediated by land availability.
Maps and measurements of land cover can be derived directly from remotely sensed data by a variety of analytical procedures, including statistical methods and human interpretation. Assessing the driving forces behind LULCC is necessary if past patterns are to be explained and used I forecasting future patterns. Driving forces of LULCC can include almost any factor that influences human activity including local culture (food preference, etc.), economics (demand for specific products, financial incentives), environmental conditions (soil quality, terrain, moisture availability), land policy quality, terrain, moisture availability), land policy and development programmes agricultural programmes, road building, zoning0 and feedbacks between these factors including past human activity on the land (land degradation, irrigation and roads).
Spatially explicit models of the social and environmental causes and consequences of LULCC are made possible by GIS and other computer-based techniques which can define and test relationships between environmental and social variables using a combination of existing data (census data, soil maps, LULC maps), observations on the ground (ecological measurements, household surveys and interviews with land managers) and data from remote sensing. These spatial models of LULCC drivers and their impacts can be used to establish cause and effect in LULCC observed in the past and are also extremely useful tools for land manages and policymakers offering for casts of future land use changes and their effects.
A remote sensing device records response which is based on many characteristics of the land surface including natural and artificial cover. An interpreter uses the element of tone, texture, pattern, shape, size, shadow, site and association to drive information about land cover.
Case Study: Neyyar Wildlife Sanctuary
A study has been carried out to ascertain the land use transformation in the Neyyar Wildlife Sanctuary in Kerala between 1973 and 2009. It is located between 77o8’ and 77o17’ east and between 8o29’ and 8o37’ north is spread over the southeast corner of the Western Ghats and covers a total area of 128 sq km. two suitable cloud-free images were used for this study. A Land sat Multi Spectral Scanner (MSS) image dated March 20, 1973 and IRS-IC Linear Imaging Self Scanner (LISS)-III satellite data of March 19, 2009 covering path and row101/68 was obtained from the National Remote Sensing Agency, Hyderabad. LANDSAT-MSS data with a spatial resolution of 80 m and spectral bands (B1 0.5-0.6 B2 0.6-0.7, B3 0.7-0.8, and B4 0.8-1.1 µm) and IRS-P6 LISS-III data with a spatial resolution of 24 m and spectral bands (B2 0.52-059, B3 0.62-068, B4 0.77-0.86, and B5 1.55-1.70 µm) were analyzed. The digital number (DN) values of the Landsat MSS and IRS P6 LISS III data were converted into radiance values using the corresponding satellite sensor parameters.
Unwanted artefacts like additive effects due to atmospheric scattering were removed through a set of pre-processing or cleaning up routines. First-order corrections were done using the dark pixel subtraction technique (Lilles and Kiefer, 1999). This technique assumes that there is a high probability that at least a few pixels within an image would be black (0 per cent reflectance). However, because of atmospheric scattering, the image system records a non-zero DN value at the supposedly dark shadowed pixel location. This represents the DN value that must be subtracted from the particular spectral band to remove the first-order scattering component.
Subsets of satellite images were first rectified for their inherent geometric errors using 1:25,000 topographic maps in World Geodetic System, 1984. Registration was carried out using distinctive features such as road intersections and stream confluences that are also clearly visible in the image. A first-degree rotation scaling and translation transformation function and nearest neighbor resembling methods were applied. These resampling methods use the nearest pixel without any interpolation to create the warped image (Richards, 1994). A total of 32 ground points were used for registration of Landsat MSS image subset with a rectification error of less than 1 pixel. The LISS III images were registered to the already registered Landsat MSS images through image-to-image registration technique with rectification errors of 0.12 and 0.08 pixels respectively with the help of toposheets. A very high level of accuracy in georeferencing of the images was possible because of the use of digital reference data that allowed zooming to the nearest possible point location (Gautam et al., 2003).
A hybrid approach combines the advantages of the automated and manual methods to produce a comprehensive land cover map. One hybrid approach is to use one of the automated classification methods to do an initial classification and then use manual methods to refine the classification and correct errors; refining the classes that did not get labelled correctly. A reasonably good classification can be obtained quickly with this approach.
Analysis of the satellite sensor data has been carried out using various digital analytical procedures. For the classification of the satellite sensor data, a stack of maximum normalized difference vegetation index (NDVI) images were generated. The NDVI images were examined, mean and standard deviation values were calculated and a thresholding technique (Fung and Ellesworth, 1988) was applied to demarcate different forest types. The supervised maximum likelihood classification method was also used for the images; classification. Training sites were derived from the satellite images using reference maps. Based on the knowledge of the data and ground truthing, different land cover classes were identified in the study area. Parametric signatures were used to train a statistically based (mean and covariance matrix) classifier to define the classes. Training sites were digitised within Erdas Imagine (Erdas, 2010), using the AOI tools. After the signatures were defined, the image was classified using the maximum likelihood parametric rule. To deliver the appropriate support size of each category the required training set for each class was determined at least ten times the number of discriminating variables used in the classified ap. Maximum likelihood classifier assigns a pixel to a particular class based upon the covariance information and a substantially superior performance is expected from this method compared to other approaches (Richards, 1994). Landsat MSS and IRS LISS III data are classified in the same manner. The different forest classes are classified based on the standard forest classification scheme, following which the map underwent an accuracy assessment.
Accuracy assessment involves identifying a set of sample locations (ground verification points). The forest cover found in the field is then compared to that which was mapped in the image for the same location by means of error or confusion matrices. Accuracy assessment is done by suing four measures: overall accuracy, user’s accuracy, producer’s accuracy and Kappa coefficient. A total of 30 random points for each class were taken to determine the accuracy of the classification method. The overall map accuracy was calculated by dividing the total correct classified pixels (major diagonal of the error matrix) by the total number of pixels in the error matrix. Producer’s accuracy indicates the probability of reference pixel being correctly classified and is a measure of omission error. User’s accuracy is the probability of the classified pixel actually representing that category on the ground. User’s accuracy is a measure of commission error (Jensen, 1996). The Kappa Coefficient measures the proportional improvement of classification over purely random assignment to classes. This accuracy measure attempts to control for a chance agreement by incorporating the off-diagonal elements as a product of the row and column of the error matrix (Cohen, 1960). After accuracy assessment, all images were clumped and vectorised in Erdas Imagine 9.1 programme.
Classified and accuracy assessed satellite images are used for change detection analysis–the raster image converted into corresponding land cover polygon with the ESRI Arc GIS software. Arc GIS geographic analysis extension is used for change detection analysis and the 1973 land cover polygon is used in union with 2009 land cover image. A Boolean operation ‘AND’ was applied between the two binary land cover polygons to identify the unchanged areas in Arc GIS. Based on the change detection analysis, land cover change between the years 1973 and 2009 was generated and area statistics calculated. In the change table a positive value indicates that the area of land cover is increased with respect to the previous year and a negative value indicates that the land cover area is decreased compared to the previous land cover image. The detailed methodology is explained in. the land cover map for 1973 and 2009 are given.
Land cover change assessment for a period of 36 years helped identify the rate and characteristics of forest type transformations. Two major and divergent trends, positive and negative were observed–the former indicating that the area of forest type has increased while the latter marks a decrease. In Neyyar Wildlife Sanctuary, during 1973-2009 the rate of change of west coast tropical evergreen forest is -2.7 per cent, west coast semi-evergreen forest is -2.9 per cent, southern moist mixed deciduous forest is -3.0 per cent, southern hilltop tropical evergreen forest is -1.1 per cent while the southern dry mixed deciduous forest is 4.7 per cent, grassland is 1.3 per cent, water body is 0.4 per cent and encroachments/settlement is 1.0 per cent.
The land cover change map thus generated indicates that the forest type during the study period was degrading; meaning that the extent of west coast tropical evergreen forest and west coast semi evergreen forest was decreasing. The spatial location of land cover change indicated that most of these forest cover transformations have occurred on the fringes of the sanctuary or near the settlements inside the sanctuary.
The changes can be attributed to a number of causes, principally livelihood dependence, agricultural expansion and infrastructure development resulting from population growth in and around the area, tourism activities, forest fire and uncoordinated policies of different governmental agencies. A geographic understanding of land use change processes can be achieved by analyzing a temporal database for spatial patterns, rates of change and trends.
The present case study clearly chows the application of remote sensing and GIS technology for assessing human induced transformation in the forest ecosystem. It is very easy to identify the positive changes leading to regeneration of the ecosystem. The negative changes indicate the forest degradation status. GIS and remote sensing techniques are widely used for the sustainable and integrated management of natural resources like forests.