Assessment and Prediction of Particulate Matter Using In-Situ and Remote Sensing Data Sets

Gogikar, Priyanjali (2019) Assessment and Prediction of Particulate Matter Using In-Situ and Remote Sensing Data Sets. PhD thesis.

[img]PDF (Restriction upto 11/06/2022)
Restricted to Repository staff only

10Mb

Abstract

The particulate matter (PM) concentration forecast is a critical component to evaluate the air quality over any region of the world. The results are vital when it comes to health issues related to air pollution in developing countries like India. The present research work analyzes ambient air quality over two famous Indian cities, namely Agra and Rourkela. The two sites are contrasting: (i) Agra, a world heritage site renowned for Tajmahal, with no industries, present inside the city boundary and hence almost no local pollution generation from the industry sector, and (ii) Rourkela, an entire industrial city with more local pollution generation. In the present research work, Suspended PM (SPM) and Respirable Suspended PM (RSPM) were analyzed from 2011 to 2015 at four stations in Agra city, namely: Taj, Itmad-Ud-Daula, Rambagh and Nunhai and 2009-2014 over Rourkela considering three stations: Sonaparbat, Rourkela and Rajgangpur (for 2013-2014). The RSPM and SPM concentrations are above critical levels during the study period and were found to be in a crucial state of pollution with poor air quality. Source regions causing transboundary pollution over both Agra and Rourkela have been analyzed by using source apportionment techniques: Conditional probability function (CPF), conditional bivariate probability function (CBPF), Weighted Potential Source Contribution Function (WPSCF) and Weighted Concentration Weighted Trajectory (WCWT). The highly polluted national capital region (NCR) including the capital city Delhi, situated in the northwesterly direction, was identified as the primary contributor for Agra. For Rourkela, industries within the city along with vehicular exhaust are identified as the primary local sources. As in-situ measurements are sparsely distributed estimation of surface-level PM2.5 is carried out using both meteorology and satellite-derived AOD using regression analysis. The present research work also focused on the prediction of future concentrations of PM using the artificial neural networks (ANN), using four models of ANN: wavelet-based multi-layer perceptron neural network (WMLPNN), wavelet-based recurrent neural network (WRNN), multi-layer perceptron neural network (MLPNN) and recurrent neural network (RNN). The performance of these neural models is evaluated statistically. The proposed WMLPNN found to be providing reliable results by delivering aid in passing the alerts and notices for the improvement of air quality. The study exemplifies that the levels of PM are not depicting any decreasing trend over Agra even after taking mitigation measures inside the city by the Government of India. The study identified that transboundary pollution is playing a significant role in the elevation of PM concentrations over considered regions of study.

Item Type:Thesis (PhD)
Uncontrolled Keywords:RSPM, SPM, Wavelet Analysis, source apportionment, regression analysis, artificial neural networks.
Subjects:Engineering and Technology > Earth Science
Engineering and Technology > Atmospheric Science
Divisions: Engineering and Technology > Department of Earth and Atmospheric Sciences
ID Code:10114
Deposited By:Mr. Sanat Kumar Behera
Deposited On:11 Jun 2020 11:53
Last Modified:11 Jun 2020 11:53
Supervisor(s):Tyagi, Bhishma

Repository Staff Only: item control page