An Assessment of Land Use Land Cover Change in Central Highland

Conversion, degradation and fragmentation threaten the integrity of ecosystems worldwide (Uddin K et al. 2016). Farmland expansion, deforestation and urbanization are the major developmental activities in the Anthropocene epoch causing worldwide depletion of biological diversity at genetic, species and ecosystem levels (Pardini R et al. 2017; Jenkins 2003; Caswell and Cohen 1993; Turner 1987; Miller 1982; Connell 1978). Nowadays, biological species live in increasingly fragmented habitat islands embedded in a matrix of human civilization (Fearnside 2001, Sagan et al. 1979; Charney et al. 1975; Otterman 1974) where the change in land cover is evident in a heterogeneous cluster of ecosystem called a landscape. The landscape, as defined by Forman and Godron (1986); is a heterogeneous land area composed of a cluster of interacting ecosystems that are repeated in similar form throughout. Diverse land use activities like human settlements, agriculture, livestock raising, forest harvesting directly affects the land cover of an area and this is a global phenomenon today (Laurance 2008; Sanderson & Harris, 2000; Xiao et al. 1991).

Shifting in the land use patterns caused by social and economic issues; results in land cover change. Other than anthropogenic drivers land cover can be changed by natural events such as weather, flood, fire, climate fluctuations, and ecosystem dynamics (Townsend et al. 2008) which gets further reflected in the landscape structure over time (Castillo et al. 2015; Baker et al. 2015; Otero et al. 2015) in different spatial sizes and frequencies as well (Farina 2000). Apart from the information of existing land cover changes; it is indeed important to monitor environmental dynamics of the land under influence of increasing population which changes the shape and nature of the landscape over time. The ability to map landforms is an important aspect of any environmental or resource analysis and modelling effort. To ensure a reliable quality of results several important steps need to follow. Conventional ground methods of land use mapping are labour intensive, time-consuming and are done relatively infrequently (Prakash A & Gupta R. P. 2010; Rogan J. 2004). These maps soon become outdated with the passage of time, particularly in a rapidly changing environment. The traditional method of data collection of land use/cover through the survey is quite difficult (Rao D. P.; Gautam N.C.; Nagaraja R & Mohan P. R. 1996; Olorunfemi 1983).

Nowadays the developed techniques of satellite imagery processing by remote sensing and GIS analysis have cut down the cost and time to prepare land use/cover maps in regular intervals of time (Mc Garical et al. 2002). Remote sensing and GIS techniques have the capability to record a response which is based on many characteristics of the land surface, including the natural and artificial cover where the structure of a landscape is defined by the spatial pattern and represented by two components: composition and configuration (Cushman et al. 2010; Forman & Godron 1986). The interpretation of a satellite image is being processed using the element of tone, texture, pattern, shape, size, shadow, site and association to derive information about the land cover. This method successfully provides efficient information of temporal trends and spatial distribution of urban areas, which is required for understanding, modelling and projecting land changes (Lu D &Weng Q 2006).

The hyperspectral and multispectral images can provide information of inaccessible areas, vegetation patches, related to phenological types, gregarious formations and communities occurring in a unique environmental setup whereas the temporal images help in monitoring a landscape over time (Delcourt 1987; Olorunfemi 1983). The images also provide a digital mosaic of the spatial arrangement of land cover and vegetation types amenable to computer processing (Chuvieco E 1999). The wide variety of land cover changes that occur on the landscape between certain periods of time can be monitored by different change detection methods. In many landscapes, recent synergistic combinations of natural disturbance and human-related changes have caused disruption and consequently, transform the prevailing landforms (Schlesinger et al. 1990; D’Antonio & Vitousek 1992). Therefore, the analyses of the landscape characteristics relative to type and frequency of disturbance can provide implications for land and wildlife management (Harris 1984; Shugart & Seagle 1985; Forman and Godron 1986; Merriam 1988; Turner and Gardner 1991; Callaway and Davis 1993; Turner et al. 1995).

The wide variety of land cover changes that occur on the landscape between certain periods of time can be examined by different change detection methods. A wide range of methods are obtainable in remote sensing to analyze these changes, emphasizing different aspects of landscape studies such as land cover conversion, change in vegetation growth, change in the landscape configuration and composition, etc. (Boriana D 2007; Boriana P and Rogan J 2006). In this study of assessing the landscape structure, principal component analysis of MODIS images was followed by the hybrid classification approach involving both the supervised and unsupervised classification methods and Markov analysis. The accuracy assessment of the classification was accomplished using kappa statistics (Gómez D & Montero J 2011). MODIS (Moderate Resolution Imaging Spectroradiometer) is a payload scientific instrument which collects data in 250m resolution in a 16-day composite. It collects data in 36 spectral bands and because of the 250m resolution, the data covers a large area in a single image (modis.gsfc.nasa.gov). Principal Components Analysis (PCA) invented in 1901 by Karl Pearson and mostly used as a tool in exploratory data analysis to find the most important variables (a combination of them) that explain most of the variance in the data. So, when there is lots of data to be analyzed, PCA can make the task a lot easier. PCA also helps to construct predictive models (Anh T & Magi S 2009). A hybrid classification method was implemented in the analysis. Use of both supervised and unsupervised classification method is ordinarily termed as a hybrid classification (Omo-Irabor O.O.; Oduyemi K 2007). The purpose for using the combination was the large size of the landscape which consists of many landforms. Single classification methods sometimes merge the similar reflectance pixels, and thus misclassify the landform. The hybrid classification method prevents this error. Accuracy assessment is a must do step before implementing any classification methods to know the sources of errors (Powell et al. 2004, Congalton R. G. and Plourde L 2002; Congalton and Green 1999, 1993).

The parcel of land can only be in one state at a given time moving successively from one state to the other with a probability which depends only on the current state and not the previous states (Wijanarto A. B. 2006). Analyzed using Markov chain analysis, the probability of moving from one state to the other is called a transition probability which is captured in a transition probability matrix whose elements are non-negative and the row elements sum up to 1 (Camacho O.M.T. et al. 2015; Yang X et al. 2012 and Arsanjani J.J 2011). This study was carried out in the Central Indian Highland in the state of Gujrat, Rajasthan and Madhya Pradesh. The landscape is a significant tiger habitat having 11 tiger reserves interconnected with each other through corridors (Qureshi et. al 2014). There are 5 established corridors within the landscape (Ranthambore-Kuno-Madhav National Park, Bandhavgarh-Achanakmar, Bandhavgarh-Sanjay-Dubri-Guru Ghasi Das, Kanha-Achanakmar, Kanha-Pench).

The landscape supports ~40% of the total tiger population (Jhala et al. 2011). The result showed the presence of 257 (213-301) tigers in Madhya Pradesh covering 13, 333 sq. km of tiger habitat and 36 (35-37) individuals covering 637 sq. km area in Rajasthan (National Tiger Conservation Authority, 2011). As the area falls under two biogeographic provinces and the river beds of Narmada, this landscape is rich in agricultural productions as well (Eaton R. M 2005; Hugh C 1911). Incidentally, the landscape is the second largest belt of minerals in the country (Prakash, A & Gupta, R. P., 2010) where non-ferrous minerals, uranium, mica, beryllium, aquamarine, petroleum, gypsum and emerald are present in Rajasthan and Gujarat (Vagholikar et al. 2003; Ghose 2003; Swer and Singh, 2004). This also triggers the mining interests (Narain et al. 2005) and falls in the core industrial development zone (Tripathi J.G., 2017). The total population of Rajasthan, Madhya Pradesh and Gujarat as per census 2001 was 5, 65, 07,188; 6, 03, 48,023; 5, 06, 71, 017 with a decadal growth rate 28.4%, 24.3% and 22.7% respectively (Census of India, 2011). In 2011 the total population of these three states gone up to 6,86,21,012; 7, 25,97,565; 6,03,83,628; with a decadal growth rate 21.45, 20.3% and 19.2% (Census India 2011).

All the three states have population decadal growth more than the average of India’s decadal population growth in a decade. The area is home to largest scheduled tribe population and among the poorest in the country. High population density, increased need for livelihood facility and economic demand mostly from the mining sector changed the land use and land cover (LULC) of the area (Malaviya, S et al., 2010). The study covering 19 tiger reserves in the landscape showed how the development of urbanization and agricultural activity continued to shrink the tiger habitat over several years (Banerjee 2017). Yadav et al; 2012 examined the condition of Nawegaon-Nagzira corridor where the shrinkage of overall forest cover and water bodies was reported, and the result showed that there has been a decrease in forest and an increase in urban and agriculture. With the agricultural intensification, the constant reduction in forest cover will impact the movement of tigers in this landscape. The outcome of these studies indicated the importance of this landscape in term of wild species conservation and management. The spatial and temporal change in the land use/cover of this landscape was not assessed before. In this study, we analyzed two different bioprovinces, with different vegetation types and physical parameters. The study focused on two objectives: 1) to assess the land use pattern, and 2) to assess the land use/cover dynamics over time.