Page 1 of 1
 Anshu Gupta Centre for Remote Sensing & GIS, NIT, Bhopal, India anshugupta20002001@gmail.com  Vivek Dey Civil Engineering Department, Indian Institute of Technology, Kanpur, India vivekde@gmail.com Alok Choudhary Head of Image Processing, M.P. Council of Science & Technology, Bhopal, India alok@yahoo.com  Fig. 1: Location of wards under the study Environmental quality assessment is essential for urban development. Rapid urbanisation has made it all the more essential now than before. But there is dearth of appropriate techniques to assess urban environment quality (UEQA). Here is a technique that is feasible, flexible and valid. UEQA requires environmental information in the form of air quality, noise quality, topography as slope and aspect, vegetation quantity and quality of greenness, water quality, soil quality etc. This work deals only with the first four qualities. Environmental information has the obvious spatial character that can be addressed by geographical information system (GIS). For example, air quality may vary for different land-use classes. Population density, as a socio-economic factor involved in urban environmental quality evaluation (UEQE), also changes in the different spatial unit. Other environmental factors such as noise and green coverage also have spatial character. So, while evaluating the UEQ, GIS provides a powerful tool to represent environmental information in support of environmental evaluation (Dai, Lee et al. 2001). An important feature of GIS is its ability to generate new information by integrating diverse datasets. The purpose of environmental evaluation in this study is to represent environmental quality in the form of maps which combines all information of each of the environmental factors. STUDY AREA Bhopal, the capital of Madhya Pradesh, is one of the fastest growing cities of India (Bhopal City Development Plan under JNNURM). The city, also known as "the city of lakes", is losing its beauty under the increasing pressures of urbanisation. DATA AND MATERIAL Spatial Data: Remote sensing satellite data used are - CARTOSAT- 1 PAN (2.5 m resolution) of Bhopal municipal corporation area, dated January 2007 and IRS P-6 LISS IV MX (5.8 m resolution) of Bhopal municipal corporation area, dated January 2007. High resolution images (2.5 m) provide more details about the spatial features. However, multispectral images provide more land cover information than panchromatic images, as each spectral waveband provides specific information about land cover features. Fusion of multispectral and high spatial resolution panchromatic data enhances the understanding of both spatial and spectral resolution of the feature and also enhances the accuracy and visual interpretation (Jensen J. R.). Annual mean concentration levels of air and noise pollution (ward wise and along four major traffic corridors) for the year 2006 have been collected from Madhya Pradesh State Pollution Control Board, LEA Associates South Asian Pvt. Ltd. and Egis BCEOM India Pvt. Ltd. Data obtained pertain to spatially well distributed locations. These locations are considered as sample locations and data of the complete area was obtained by spatial interpolation IDW using GIS. METHODOLOGY Environmental information for UEQE assessment is broken into smaller components or indicators. Air quality, noise quality, topography, slope and aspects, vegetation as NDVI, demography of the study area and land use have been evaluated as per their contribution towards urban environmental pollution (UEP). The contribution of traffic has been given a special attention in evaluating UEP. Road buffer (Kwang Hoon Chi, No-Wook Park, 2002) has been created along major traffic corridors to consider the enhanced effect of traffic towards air and noise pollution. Each of the smaller components has been shown at the topmost level in Figure 2. The combination of indicators was carried out using the analytical hierarchical process and fuzzy weights which involves the opinion of experts in urban pollution board, urban development, meteorology and urban road and traffic development (John M., Martin Hale (2001)). Fuzzy inference network has been established as shown in Figure 2 to incorporate all the environmental information in a logical manner. Boolean inference network is similar but the difference lies in the combination strategy. Instead of fuzzy weights, Boolean weights of 0 and 1 are considered and Boolean algebraic sum and Boolean OR are used to evaluate the final quality map (Figure 4). Threshold for the four evaluation classes is determined by trial and error. Exact classes of few of the locations, used to evaluate interpolated data, are evaluated by experts. Threshold, which classifies the training location correctly with maximum accuracy, is chosen as threshold for both the Boolean and fuzzy approach.  Fig. 2: Fuzzy Inference Network Each component's effect in enhancing urban environmental pollution is evaluated in a spatial raster layer format through GIS. Considering the raster data structure, every indicator is considered as the individual layer in the fuzzy overlay operation. The value of each cell is the score of the indicator considered (Chi, Park and Chung, 2002). The implementation process of fuzzy multi-criteria evaluation in GIS through fuzzy inference network includes three phases. Firstly, every bottom indicator of each component is overlaid based on fuzzy operation, also called intermediate hypothesis. For example, in air pollution criteria, the criterion consists of four indicators (SO2, NO2, SPM and CO) in wards as well as along the four major traffic corridors. The first phase of evaluation is to overlay these four indicators based on fuzzy operation (fuzzy algebraic sum). That is to say, it is a bottom to top approach. Secondly, the fuzzy operation is carried out to overlay air pollution in wards and air pollution along road (Fuzzy OR). Finally, the final  Table 1: Area Distribution hypothesis performing the fuzzy overlay operation of environment pollution and physical environment component to get the final quality map (fuzzy GAMMA). 'GAMMA' operator has been used while applying FUZZY LOGIC technique to obtain the final output map. FUZZY ALGEBRAIC SUM and FUZZY OR are used as intermediate hypothesis. The final criteria map by BOOLEAN theory has been processed by using "ARITHMETIC SUM". The results so obtained have been classified under four categories of pollution. These evaluation classes are Low, Moderate, High and Critical.  (Fig. 3: Classified Final Quality Map by Fuzzy Approach) (Fig. 4: Final Quality Map by Conventional Approach) DISCUSSION AND ANALYSIS The major difference between the two maps is on the roadside pollution. From the map using fuzzy approach, it is clearly seen that the areas along the major traffic corridors are in 'cyan', indicating that these areas are highly polluted. This is not at all seen in the map using conventional approach. With Fuzzy approach, the information on pollution is retained and clearly reflected in the evaluation result for environmental pollution, whereas this information is lost with the conventional (Boolean) approach during the process of evaluation. This is because the fuzzy approach employs a set of logically evaluated weights to determine in what degree the component belongs to one evaluation class.  Table 2: Study Area Covered in Each Class Ambiguity resolution is more in fuzzy approach because of the continuous range of values whereas in Boolean approach, it is a discrete integer value. This clearly suggests that fuzzy approach will give more information about the pollution level than the Boolean approach which is also evident from the higher percentage of area obtained by fuzzy approach in critical class. VALIDATION Validation of the obtained results is carried out by a field visit to a sample location other than the sample locations used for map evaluation. Urban environmental quality at this sample location is evaluated by the panel of experts from urban pollution board, urban development, meteorology and urban road and traffic development. Some of the major anomalies in Boolean and fuzzy results are illustrated along with the reasons behind the results.  Fig. 5: Pollution along major roads As encircled in Figure 6, the area labelled as A (Karariya Sajidabad), shows different behaviour in fuzzy and Boolean approach. The anomaly in the observed area is that Boolean approach is showing the area in the High zone while Fuzzy approach categorises it in Low and Moderate class. The reason for the above difference is the land use. Upon a field visit to the area, agricultural farms (vegetation) are found. The difference is also obtained around the area shown in the rectangular box B (Nakkar Khana). Now observe the area covered under rectangle 'B'. The area has well structured road network (VVIP road) due to which the traffic runs smoothly. But after just around 0.5 - 1.0 km, the area, Peer Gate, comes in Critical class.  Fig. 6: Critical Condition in Old City of Bhopal The Boolean approach is highly influenced by this area and shows the area around VVIP road in critical class where as in fuzzy approach, the quality of environment decreases gradually from high to critical class which reveals the continuous behaviour of fuzzy approach. CONCLUSION Inference network provides a flexible format in which at every stage, any new component can be added to the network while continuing to maintain the logical format. Evaluated final quality map with the available data has established the feasibility of the inference network in urban environmental quality assessment. Further, the integration of fuzzy logic and GIS through inference network in evaluating urban environmental quality gives better results than the Boolean approach. The ability of fuzzy approach to quantify the ambiguity of complexity of urban environmental quality has been established on comparison with the Boolean approach by using the error matrix. Error matrix shows the overall accuracy of 55 % of Boolean approach with respect to the fuzzy approach. This concludes that conventional Boolean approach is insufficient in explaining the urban environmental quality. Page 1 of 1
|