Land cover relates to the discernible Earth surface expressions, such as vegetation, soil, water or anthropogenic features, and thus describes the Earth’s physical state in terms of the natural environment and the man-made structures (Xavier Baulies and Gerard Szejwach, 1998). Essentially, land cover can have only one class or category at a given time and location, and can be mapped using suitable remote sensing data with spectral signatures. Land use is an expression of human uses of the landscape, e.g. for residential, commercial, or agricultural purposes, and has no spectral basis for its unique identification. Thus it can not be explicitly derived from image data, but only inferred by visual interpretation or assessed in the framework of object-based contextual analysis

Land cover changes induced by human and natural processes play a major role in global as well as at regional scale patterns of the climate and biogeochemistry of the Earth system. Humans have left an impressive mark on the world's land over the past several centuries. With the dramatic growth in world population, from roughly one billion in 1800 to well over five billion today, pressure on land have greatly increased. The need for greater food production has led to a massive increase in cropland. By the early 1990s, almost 40 percent of Earth's land surface had been converted to cropland and permanent pasture. This conversion has occurred largely at the expense of forests and grassland (Ramachandra and Shruthi, 2007). Variations in topography, vegetation cover, and other physical characteristics of the land surface influence surface-atmosphere fluxes of sensible heat, latent heat, and momentum of heated air particulates caused by conduction, convection and radiation, which in turn influence weather and climate. The land cover features can be classified using remotely sensed satellite imagery of different spatial, spectral and temporal resolutions. This would be of great importance for the management and planning of activities such as land-use development, natural resource exploitation and engineering projects.

Implementation of sound management strategies is a prerequisite for sustainable utilization of resources. The sustainable development of a region requires a synoptic ecosystem approach in the management of natural resources that relates to the dynamics of natural variability and the effects of human intervention on key indicators of biodiversity and productivity. A regional database on natural resources is needed to support the information requirements of planning at disaggregated levels for sustainable development. Natural resource inventory through land cover mapping helps in describing the quality, quantity, change, productivity and condition of bioresources in a given area. Also, it helps in identifying the reasons responsible for their inequality in distribution, availability and demand. This aids the regional planners to incorporate conservation measures for the resources as well as rehabilitation of degraded resources during policy interventions (Ramachandra, et al., 2004, 2006). The detection and quantitative assessment of vegetation is one of the major applications of remote sensing for environmental resource management and decision-making. Remote sensing technology has been useful in environmental studies involving monitoring of large areas and estimation of environmental parameters, which are often the principal indicators of change in highly variable heterogeneous landscape. It provides both reliable periodic observations and a synoptic view of the region concerned, allowing high frequency repetitive surveillance and identification of problematic sites and emergencies, respectively (James Campbell, 2002, Binaghi et al, 1999, Ramachandra 2007a). This helps in mapping and classification of land cover features, such as vegetation, soil, water and forests, and also in assessing the extent and diversity of vegetation. Geographical information system (GIS) helps in archiving, analysis and visualistaion of remotely sensed data along with other collateral data (spatial as well as statistical). Remote sensing data along with GIS and GPS (Global positioning system) help in land cover analyses (Ramachandra, 2007b).

The accuracy of the National land cover map using a probability sampling design incorporating three levels of stratification and two stages of selection assessed and reported for each of the four regions comprising the eastern United States for both Level I and II classifications. Overall accuracies for Levels I and II are 80% and 46% for New England, 82% and 62% for New York/New Jersey (NY/NJ), 70% and 43% for the Mid-Atlantic, and 83% and 66% for the Southeast (Stehman, et al, 2003). Evaluation of the performance of univariate and multivariate decision trees (DT) for land cover classification considering training data from two different geographical areas and two different sensors––multispectral Landsat ETM+ and hyper spectral DAIS indicate that the performance of the univariate DT is acceptably good in comparison with that of other classifiers, except with high-dimensional data. Classification accuracy increases linearly with training data set size to a limit of 300 pixels per class in this case. Multivariate DTs do not appear to perform better than univariate DTs. While boosting produces an increase in classification accuracy of between three percent and six percent, the use of attribute selection methods does not appear to be justified in terms of accuracy increases. However, neither the univariate DT nor the multivariate DT is performed with high-dimensional data (Mahesh Pal, et al, 2003). Land cover discrimination was investigated through the analyses of fine resolution spectra, convolved spectra (MODIS band passes), and vegetation indices-the normalized difference vegetation index (NDVI) and the enhanced vegetation index (EVI). At these data levels, the amount of data correctly classified into five major land cover types were 91% (full spectra), 78% (red and NIR), 75% (NDVI), and 71% (EVI). NDVI and EVI together were capable of correctly classifying 82% of the total data set (Ferreira, et al, 2004).

In India, spatial accounting and monitoring of land cover have been carried out since 2004 at a national level at 1:250,000 scale, using multi-temporal IRS AWiFS (Advanced Wide Field Sensor) with 4 bands (Green, Red, NIR and SWIR) at 56 m resolution images to address the spatial and temporal variability in cropping patterns. The analysis of vegetation and detection of changes in the vegetation pattern are keys to the natural resource assessment and monitoring. The monitoring of  land cover involves the computation of vegetation indices as a radiometric measure of the spatial and temporal patterns based on the spectral responses of various land features. Vegetation indices play an important role in the derivation of biophysical parameters, such as percentage of vegetation cover, biomass production, etc.  Their interest lies in the detection of changes in land use and the monitoring of the seasonal dynamics of vegetation on regional scale. Several indices have been developed for various applications and under quite specific conditions. However, the use of these indices to characterize land cover can be limited by various physical effects that affect the signal at the sensor, namely: atmospheric effects, effects of the optical properties of the bare soil subjacent to the vegetation cover, etc. These factors increase or decrease reflectance’s in the red and near infrared spectral bands and, consequently, limit the detection of land cover changes using vegetation indices, which causes errors in the interpretation and the analysis of the results.  Primary objective of this investigation is to explore and comparatively assess various techniques that could be adopted for an assessment of land cover in a dry arid region. This analysis was carried out for Kolar district in Karnataka State, India.