MATERIALS AND METHODS

Data analysis: Preprocessing: The remote sensing data corresponding to the study region were downloaded, geo-referenced, rectified and cropped pertaining to the administrative boundary with 3km-5km buffer depending on data availability was considered. Landsat ETM+ bands of 2010 were corrected for the SLC-off by using image enhancement techniques, followed by nearest-neighbor interpolation. Data used for the analysis are listed in Table 1.

Table 1: Data used and the purpose

Land Cover Analysis: Among different land cover indices, NDVI - Normalised Difference Vegetation Index was found appropriate and NDVI was computed to understand the changes of land cover. NDVI is the most common measurement used for measuring vegetation cover. It ranges from values -1 to +1 depending on the earth features.

Land use analysis: The method involves i) generation of False Colour Composite (FCC) of remote sensing data (bands – green, red and NIR). This helped in locating heterogeneous patches in the landscape ii) selection of training polygons covering 15% of the study area and uniformly distributed over the entire study area, iii) loading these training polygons co-ordinates into pre-calibrated GPS, vi) collection of the corresponding attribute data (land use types) for these polygons from the field, iv) Supplementing this information with Google Earth. Land use classification was done using supervised pattern classifier - Gaussian maximum likelihood algorithm based on various classification decisions using probability and cost functions [20]. Land use was computed using the temporal data through open source GIS: GRASS- Geographic Resource Analysis Support System (http://ces.iisc.ernet.in/grass). Four major types of land use classes considered were built-up, vegetation, cultivation area (since major portion is under cultivation), and water body. 60% of the derived signatures (training polygons) were used for classification and the rest for validation. Statistical assessment of classifier performance based on the performance of spectral classification considering reference pixels is done which include computation of kappa (κ) statistics.

Density gradient and zonal analysis and computation of Shannon’s entropy: Further the classified spatial data is divided into four zones based on directions considering the central pixel (Central Business district) as Northwest (NW), Northeast (NE), Southwest (SW) and Southeast (SE) respectively. The growth of the urban areas was monitored in each zone separately through the computation of urban density for different periods. Each zone was further divided into incrementing concentric circles of 1km radius from the center of the city. The built up density ineach circle is monitored overtime using time series analysis. Landscape metrics were computed for each circle, zone wise using classified land use data at the landscape level with the help of FRAGSTATS [21]. To determine whether the growth of urban areas was compact or divergent the Shannon’s entropy [21,1] was computed direction wise for the study region. Shannon's entropy (Hn) given in equation 1, provide insights to the degree of spatial concentration or dispersion of geographical variables among ‘n’ concentric circles across Zones.

Hn=- ……………. (1)

Where Pi is the proportion of the built-up in the ith concentric circle. As per Shannon’s Entropy, if the distribution is maximally concentrated the lowest value zero will be obtained. Conversely, if it evenly distribution the value would be closer to log n indicating dispersed growth or sprawl.