Modelling anthropogenic pm2.5 over India in the present and future climate

25  Download (0)

Full text









©Indian Institute of Technology Delhi (IITD), New Delhi, 2021







Centre for Atmospheric Sciences


In fulfilment of the requirements of the degree of DOCTOR OF PHILOSOPHY

to the




ii Certificate

This is to certify that the thesis entitled “Modelling anthropogenic PM2.5 over India in the present and future climate” is being submitted by Mr. Abhishek Kumar Upadhyay for the award of the degree of Doctor of Philosophy, is a record of the original bonafide research work carried out by him. He has worked under my guidance and supervision and has fulfilled the requirements for the submission of this thesis. The results presented in this thesis have not been submitted in part or full to any University or Institution for award of any degree/diploma

Prof. Sagnik Dey Prof. Pramila Goyal Associate Professor, Professor,

Centre for Atmospheric Sciences, Centre for Atmospheric Sciences,

Indian Institute of Technology Delhi, Indian Institute of Technology Delhi Hauz Khas, New Delhi-110016, India Hauz Khas, New Delhi-110016, India




The successful completion of any work that we take up in life is always a team effort. This thesis is no different. I want to express my sincere gratitude to my thesis supervisors, Prof.

Sagnik Dey, and Prof. Pramila Goyal, for providing me an opportunity to pursue Ph.D. under their supervision at the Centre for Atmospheric Sciences, IIT Delhi. I am very grateful for their valuable guidance, constructive criticism, and constant encouragement. I got full benefit from their subject expertise and academic experience that made my Ph.D. tenure enlightening, cheerful and motivating. They have always given me the freedom to pursue my interests and provided me with insightful suggestions and support in developing independent thinking and research skills.

I would like to extend my sincere gratitude to my SRC members for generously sharing their knowledge and time. I would also like to thank the Head of Centre, my CRC chairperson, and all the faculties at CAS, IIT Delhi, for providing all necessary facilities and a great environment to learn and grow at the Centre. They are always available for discussion, suggestions, and other academic and non-academic help during my Ph.D. tenure. I am very thankful to all my faculty for their comprehensive and educative courses. I also like to thank Prof. S. K. Dash, CAS, IIT Delhi, and Dr. Rajesh Kumar, NCAR, USA, for their active collaborations and valuable suggestions in my research work.

I want to acknowledge High-performance computing (HPC) facilities at IITD for providing the required computing facilities for model simulation and storage. I also like to acknowledge computing and storage facilities at CAS, IIT Delhi for hassle-free computation and analysis during my Ph.D. I thank NCAR for the open-source chemical transport model (WRF-Chem), the input meteorological data (FNL), geographical data, and boundary conditions. I thank JRC for



developing and providing EDGAR-HTAP emission inventory and IIASA to develop and provide ECLIPSE emission inventory.

I want to thank my friends from IIT Delhi, Dr. Sorangsu Chowdhury, Puneet Sharma, Popat Salunke, Abhishek Anand, Kunal Bali, Prateek Sharma, and Anshul, for always being the friends in need and their excellent company made my stay at IIT pleasant and memorable. I also thank all my friends like Abhishek, Raju, Akash, Ravi, Rajeev, and Prasant, for being there for me throughout. I like to thank my seniors Dr. Pushp Raj Tiwari, Dr. Saurabh Sinha, Dr. Dhirendra Mishra, Dr. Sushant Das, Dr. Ram Singh, Dr. Piyush Srivastava, Dr. Amit Singh, and Dr.

Gavendra Pandey, for their constant support and suggestions. I also thank Vijay for helping me with the logistics.

Words cannot completely express my love and gratitude to my family members who have supported and encouraged me through this journey. I thank them for making all the efforts to nurture me. My parents were my first teachers. They scolded and moulded me as a boy and provided me with freedom as a man.

Finally, I thank the almighty God for the passion, strength, perseverance, and resources to complete this study.

Date: 20-02-2021 Abhishek Kumar Upadhyay


v Abstract

PM2.5 (particulate matter less than 2.5 µm in aerodynamic diameter) can enter deep inside the respiratory system and cause various health problems like chronic obstructive pulmonary diseases, ischemic heart diseases, stroke, and lung cancer. These fine particulate matters alter the radiation budget and impact the climate and weather system. High PM2.5 exposure over the south Asian region, including India, is a major health risk. Understanding the distribution of PM2.5

tagged to various sources and the factors that influence such distribution are key in devising feasible mitigation strategies. This thesis is aimed to model anthropogenic PM2.5 distribution over India that is required to enhance the understanding of PM2.5 situations over India and formulate efficient and feasible mitigation policies. In view of limited ground-based observation for PM2.5 over India and temporal limitations of satellite products, chemical transport models are a useful tool to fill these data gaps. It can also perform investigative experiments. WRF-Chem (Weather research forecasting coupled with chemistry) model is used in this thesis to simulate PM2.5 over India, and then the contribution of major anthropogenic sources is estimated. Further, the representativeness of emissions over India from the two widely used emission inventory is explored. Finally, PM2.5 concentration is projected under two emission pathways, and their impact on health and climate are analyzed, and the implications are presented.

The model captures the observed seasonal distribution of PM2.5 over India, with the highest level of pollution in the winter and the lowest level in the monsoon. Spatially, PM2.5 is the highest over the Indo-Gangetic plains (IGP), with a seasonal shift in pollution from the west to the east. After the monsoon is withdrawn, pollution builds up in the western IGP while pushed towards the eastern IGP in the winter season. Night-time concentration is higher than day-time with a diurnal amplitude in anthropogenic PM2.5 exceeding 15 µg m-3 in large parts of India.



These results show that meteorology, topography, and emission are the three major factors that guide the seasonal and spatial distribution of PM2.5. Since anthropogenic emissions are under our control, the contribution of each anthropogenic emission source towards PM2.5 is estimated, followed by the health impact attributed to each sectoral emission. The residential sector is the biggest contributor to anthropogenic PM2.5 in India. Population-weighted all-India averaged (±1σ) annual ambient PM2.5 exposures due to residential, transport, industry, and energy sectors in 2010 are estimated to be 26.2±12.5, 3.8±4.3, 5.5±2.7 and 2.2±2.3 µg m-3, respectively. Complete mitigation of emissions from the transport, industrial, and energy sectors combined would save 92,380 (95% uncertainty interval (UI), 40,918 – 140,741) premature deaths annually, primarily at the urban hotspots. For the residential sector, this would result in avoiding 378,295 (95% UI, 175,002 – 575,293) premature deaths due to a reduction in ambient PM2.5 exposure in addition to the benefit of avoiding all premature deaths from household exposure. Bihar and Goa are expected to have the largest (289) and smallest (48) health benefits (per 100,000 population) from complete mitigation of these four anthropogenic sources.

Understanding the projected changes in PM2.5 in the future is also critical to assess the efficacy of the various mitigation pathways. Evaluating the climate and air quality impact of short-lived pollutants (ECLIPSE) emission inventory provides projected emissions under two contrasting emission pathways. In contrast, Emission database for global atmospheric research - Hemispheric transport of air pollution (EDGAR-HTAP) only provides the current emissions. So, a comparative analysis of the representativeness of EDGAR-HTAP and ECLIPSE emission inventory for the Indian subcontinent is performed. This has provided an opportunity to justify the use of ECLIPSE emission inventory for future projections and compare with the simulations using EDGAR-HTAP inventory in the present condition. The statistical parameters show similar



accuracy for both inventories (correlation coefficients, R of 0.87 and 0.9 and RMSEs of 8.4 and 10.2 µg m-3 for EDGAR and ECLIPSE, respectively) at the national scale, but it shows significant regional variations. The simulated anthropogenic populated PM2.5 exposure is higher with ECLIPSE emission compared to EDGAR for most states. In some populated states, Delhi, Uttar Pradesh, Haryana, Bihar, Uttarakhand, Rajasthan, Punjab, and few less populated states like Sikkim, Tripura, population-weighted exposure exceeds more than 30% with ECLIPSE emission inventory compared to EDGAR.

Future projections of PM2.5 are made over India with a computationally efficient statistical model and physically resilient dynamical model using projected emissions from ECLIPSE emission inventory. The multiple regression (MLR) model is used as a statistical tool, which shows reasonable accuracy (R>0.9) during training and validation. The changes of the meteorological parameters under both Representative concentration pathway (RCP) 4.5 and 8.5 scenarios partially negates the impact of rising emissions in the future; more so in RCP8.5 than in RCP4.5 scenario. Air quality is projected to improve significantly with short-lived climate pollutant (SLCP) mitigation emission pathway in comparison with current legislation (CLE) baseline emission pathway. The spatial analysis identifies a rapid increase in anthropogenic PM2.5 in the eastern Indian states of Jharkhand, Chhattisgarh and Odisha, Peninsular India, and Delhi National Capital Region in the near future relative to the present-day.

WRF-Chem model in its climate mode is used to project PM2.5 concentration under baseline and mitigation emission pathways. All-India average PM2.5 exposure is expected to increase from 41.4±26.5 µg m-3 in 2010 to 61.1±40.8 and 58.2±37.5 µg m-3 in 2030 under RCP8.5 and RCP4.5 scenarios, respectively if India follows the CLE emission pathway. In contrast, ambient PM2.5 in 2030 would be 40.2±27.5 (for RCP8.5) and 39.2±25.4 (for RCP4.5)



µg m-3 following the SLCP mitigation emission pathway. We find that the lower PM2.5 in the mitigation pathway (34.2% and 32.6%, respectively for RCP8.5 and RCP4.5 relative to the baseline emission pathway) would come at the cost of 0.3-0.5°C additional warming due to the direct impact of aerosols. The premature mortality burden attributable to ambient PM2.5 exposure is expected to rise from 2010 to 2030, but 381,790 (5-95% confidence interval, CI: 275,620- 514,600) deaths can be averted following the mitigation emission pathway relative to the baseline emission pathway.

This study carries out a comprehensive analysis of PM2.5 distribution over India in the current and future climate and provides some useful policy suggestions. WRF-Chem is a reliable and valuable tool for understanding air pollution over India. From the policy perspective, mitigation measures should be more intense in the winter season and over the IGP. Compared to the other major (transport, industry, and energy) anthropogenic sources, controlling residential sources should be prioritized because of the effectiveness of implementing mitigation measures and the expected larger health benefit at a regional scale. However, additional mitigation measures are advised at the urban hotspots to curb emissions from these sectors to maximize possible health benefits. Exposure modeling needs to be carried out with higher temporal frequency for a better health burden assessment. Our results highlight the near future pollution hotspots, and this information would be useful in air quality management through city planning, strict implementation of regulations and guidelines across the country, air quality monitoring, and analysis at a regional scale. These findings conclude that with the expected large health benefit, the mitigation emission pathway is a reasonable tradeoff for India despite the unfavorable meteorological response.






(वायुगितक�य व्यास में 2.5 µm से कम कण) �सन प्रणाली के अंदर गहराई में प्रवेश कर सकता है और िविभन्न स्वास्थ्य समस्याओं जैसे क्रॉिनक ऑब्सट्रिक्टव पल्मोनरी िडजीज, इस्केिमक �दय रोग, स्ट्रोक और फेफड़ों के कैंसर का कारण बन सकता है। ये सू�म कण पदाथर्

िविकरण बजट को बदल देते हैं और जलवायु और मौसम प्रणाली को प्रभािवत करते हैं। भारत सिहत दि�ण एिशयाई �ेत्र में उच्च PM


जोिखम एक प्रमुख स्वास्थ्य जोिखम है। िविभन्न स्रोतों को टैग

िकए गए PM


के िवतरण को समझना और ऐसे िवतरण को प्रभािवत करने वाले कारक व्यवहायर्

शमन रणनीितयों को तैयार करने में महत्वपूणर् हैं। इस थीिसस का उद्देश्य भारत में मानवजिनत PM


िवतरण को मॉडल करना है जो भारत में PM


िस्थितयों क� समझ को बढ़ाने और कुशल और व्यवहायर् शमन नीितयों को तैयार करने के िलए आवश्यक है। भारत में PM


के सीिमत भू -आधा�रत अवलोकन और उपग्रह उत्पादों क� अस्थायी सीमाओं को देखते ह�ए, इन डेटा अंतराल को भरने के

िलए रासायिनक प�रवहन मॉडल एक उपयोगी उपकरण हैं। यह खोजी प्रयोग भी कर सकता है। इस थीिसस में WRF-Chem (रसायन िव�ान के साथ मौसम अनुसंधान पूवार्नुमान) मॉडल का उपयोग भारत में PM


का अनुकरण करने के िलए िकया जाता है, और िफर प्रमुख मानवजिनत स्रोतों के

योगदान का अनुमान लगाया जाता है। इसके अलावा, दो व्यापक �प से उपयोग क� जाने वाली

उत्सजर्न सूची से भारत पर उत्सजर्न क� प्रितिनिधत्व �मता का पता लगाया गया है। अंत में, PM


सांद्रता को दो उत्सजर्न माग� के तहत प्र�ेिपत िकया जाता है, और स्वास्थ्य और जलवायु पर उनके

प्रभाव का िव�ेषण िकया जाता है, और इसके िनिहताथर् प्रस्तुत िकए जाते हैं।

मॉडल भारत में PM


के देखे गए मौसमी िवतरण को दशार्ता है, िजसमें सिदर्यों में प्रदूषण का

उच्चतम स्तर और मानसून में िनम्नतम स्तर होता है। स्थािनक �प से, PM


भारत-गंगा के मैदानों

(IGP) में सबसे अिधक है, िजसमें प्रदूषण में मौसमी बदलाव पि�म से पूवर् क� ओर होता है। मानसून क� वापसी के बाद, पि�मी आईजीपी में प्रदूषण का िनमार्ण होता है जबिक सिदर्यों के मौसम में पूव�

आईजीपी क� ओर धकेल िदया जाता है। भारत के बड़े िहस्सों में मानवजिनत PM


में दैिनक आयाम 15 µg m


से अिधक होने के साथ रात के समय क� सांद्रता िदन के समय क� तुलना में अिधक होती


इन प�रणामों से पता चलता है िक मौसम िव�ान, स्थलाकृित और उत्सजर्न तीन प्रमुख कारक हैं जो



के मौसमी और स्थािनक िवतरण का मागर्दशर्न करते हैं। चूंिक मानवजिनत उत्सजर्न हमारे

िनयंत्रण में है, इसिलए PM


के प्रित प्रत्येक मानवजिनत उत्सजर्न स्रोत के योगदान का अनुमान



लगाया गया है, इसके बाद प्रत्येक �ेत्रीय उत्सजर्न के िलए िजम्मेदार स्वास्थ्य प्रभाव का अनुमान लगाया गया है। भारत में मानवजिनत PM


में आवासीय �ेत्र का सबसे बड़ा योगदान है। 2010 में

आवासीय, प�रवहन, उद्योग और ऊजार् �ेत्रों के कारण जनसंख्या-भा�रत अिखल भारतीय औसत (±

1 σ ) वािषर्क प�रवेश PM


जोिखम क्रमशः 26.2 ± 12.5, 3.8 ± 4.3, 5.5 ± 2.7 और 2.2 ± 2.3 µg m


, होने का अनुमान है। । प�रवहन, औद्योिगक और ऊजार् �ेत्रों से उत्सजर्न को पूरी तरह से कम करने से 92,380 (95% अिनि�तता अंतराल (UI), 40,918 - 140,741) सालाना समय से पहले

होने वाली मौतों क� बचत होगी, मुख्य �प से शहरी हॉटस्पॉट्स में। आवासीय �ेत्र के िलए, इसके

प�रणामस्व�प 378,295 (95% UI, 175,002 - 575,293) समय से पहले होने वाली मौतों से बचने

में मदद िमलेगी, जो िक घरेलू जोिखम से सभी समय से पहले होने वाली मौतों से बचने के लाभ के

अलावा प�रवेश PM


जोिखम में कमी के कारण होगी। इन चार मानवजिनत स्रोतों के पूणर् शमन से

िबहार और गोवा को सबसे बड़ा (289) और सबसे छोटा (48) स्वास्थ्य लाभ (प्रित 100,000 जनसंख्या) होने क� उम्मीद है।

भिवष्य में PM


में अनुमािनत प�रवतर्नों को समझना िविभन्न शमन माग� क� प्रभावका�रता

का आकलन करने के िलए भी महत्वपूणर् है। अल्पकािलक प्रदूषकों (ECLIPSE) उत्सजर्न सूची के

जलवायु और वायु गुणव�ा प्रभाव का मूल्यांकन दो िवपरीत उत्सजर्न माग� के तहत अनुमािनत उत्सजर्न प्रदान करता है। इसके िवपरीत, वैि�क वायुमंडलीय अनुसंधान के िलए उत्सजर्न डेटाबेस - वायु प्रदूषण का गोलाधर् प�रवहन (EDGAR-HTAP) केवल वतर्मान उत्सजर्न प्रदान करता है।

इसिलए, भारतीय उपमहाद्वीप के िलए EDGAR-HTAP और ECLIPSE उत्सजर्न सूची के

प्रितिनिधत्व का तुलनात्मक िव�ेषण िकया जाता है। इसने भिवष्य के अनुमानों के िलए ECLIPSE उत्सजर्न सूची के उपयोग को सही ठहराने और वतर्मान िस्थित में EDGAR-HTAP इन्वेंट्री का

उपयोग करने वाले िसमुलेशन के साथ तुलना करने का अवसर प्रदान िकया है। सांिख्यक�य पैरामीटर राष्ट्रीय स्तर पर दोनों इन्वेंट्री (सहसंबंध गुणांक, 0.87 और 0.9 के आर और 8.4 और 10.2 µg m


क्रमशः EDGAR और ECLIPSE के िलए) के िलए समान सटीकता िदखाते हैं, लेिकन यह महत्वपूणर्

�ेत्रीय िविवधताओं को दशार्ता है। अिधकांश राज्यों के िलए EDGAR क� तुलना में ECLIPSE उत्सजर्न के साथ नकली मानवजिनत आबादी वाला PM


जोिखम अिधक है। कुछ आबादी वाले

राज्यों, िदल्ली, उ�र प्रदेश, ह�रयाणा, िबहार, उ�राखंड, राजस्थान, पंजाब और िसिक्कम, ित्रपुरा

जैसे कुछ कम आबादी वाले राज्यों में, जनसंख्या-भा�रत जोिखम EDGAR क� तुलना में ECLIPSE

उत्सजर्न सूची के साथ 30% से अिधक है।





के भिवष्य के अनुमान भारत में एक कम्प्यूटेशनल �प से कुशल सांिख्यक�य मॉडल और शारी�रक �प से लचीला गितशील मॉडल के साथ ECLIPSE उत्सजर्न सूची से अनुमािनत उत्सजर्न का उपयोग करके बनाए गए हैं। मल्टीपल �रग्रेशन (MLR) मॉडल का उपयोग एक सांिख्यक�य उपकरण के �प में िकया जाता है, जो प्रिश�ण और सत्यापन के दौरान उिचत सटीकता

(R> 0.9) िदखाता है। प्रितिनिध एकाग्रता मागर् (RCP) 4.5 और 8.5 दोनों प�र�श्यों के तहत मौसम संबंधी मापदंडों के प�रवतर्न भिवष्य में बढ़ते उत्सजर्न के प्रभाव को आंिशक �प से नकारते हैं;

RCP8.5 प�र�श्य क� तुलना में RCP8.5 में अिधक। वतर्मान कानून (CLE) बेसलाइन उत्सजर्न मागर्

क� तुलना में अल्पकािलक जलवायु प्रदूषक (SLCP) शमन उत्सजर्न मागर् के साथ वायु गुणव�ा में

उल्लेखनीय सुधार का अनुमान है। स्थािनक िव�ेषण वतर्मान समय के सापे� िनकट भिवष्य में पूव�

भारतीय राज्यों झारखंड, छ�ीसगढ़ और ओिडशा, प्रायद्वीपीय भारत और िदल्ली राष्ट्रीय राजधानी

�ेत्र में मानवजिनत PM


में तेजी से वृिद्ध क� पहचान करता है।

WRF-Chem मॉडल अपने जलवायु मोड में बेसलाइन और शमन उत्सजर्न माग� के तहत PM


एकाग्रता को प्रोजेक्ट करने के िलए उपयोग िकया जाता है। अिखल भारतीय औसत PM


जोिखम 2010 में 41.4 ± 26.5 µg m


से बढ़कर 2030 में RCP8.5 और RCP4.5 प�र�श्यों के

तहत क्रमशः 61.1 ± 40.8 और 58.2 ± 37.5 µg m


हो जाने क� उम्मीद है, यिद भारत सीएलई उत्सजर्न मागर् का अनुसरण करता है। इसके िवपरीत, 2030 में प�रवेशी PM


40.2-27.5 (RCP8.5 के िलए) और 39.2 ± 25.4 (RCP4.5 के िलए) µg m


SLCP शमन उत्सजर्न मागर् के

बाद होगा। हम पाते हैं िक न्यूनीकरण मागर् में िनम्न PM


(बेसलाइन उत्सजर्न मागर् के सापे� क्रमशः

34.2% और 32.6%, RCP8.5 और RCP4.5 के िलए) 0.3-0.5̊C अित�र� वािम�ग क� क�मत पर आएगा। एरोसोल का प्रत्य� प्रभाव। प�रवेशी PM


जोिखम के कारण समय से पहले मृत्यु का बोझ 2010 से 2030 तक बढ़ने क� उम्मीद है, लेिकन 381,790 (5-95% िव�ास अंतराल, CI:

275,620- 514,600) मौतों को आधारभूत उत्सजर्न के सापे� शमन उत्सजर्न मागर् के बाद टाला जा

सकता है।

यह अध्ययन वतर्मान और भिवष्य के माहौल में भारत में PM


िवतरण का व्यापक िव�ेषण करता है और कुछ उपयोगी नीितगत सुझाव प्रदान करता है। WRF-Chem भारत में वायु प्रदूषण को

समझने के िलए एक िव�सनीय और मूल्यवान उपकरण है। नीित के �ि�कोण से, शमन के उपाय

सिदर्यों के मौसम में और आईजीपी क� तुलना में अिधक तीव्र होने चािहए। अन्य प्रमुख (प�रवहन,

उद्योग और ऊजार्) मानवजिनत स्रोतों क� तुलना में, अन्य प्रमुख (प�रवहन, उद्योग) क� तुलना में,



शमन उपायों को लागू करने क� प्रभावशीलता और �ेत्रीय स्तर पर अपेि�त बड़े स्वास्थ्य लाभ के

कारण आवासीय स्रोतों को िनयंित्रत करने को प्राथिमकता दी जानी चािहए। , और ऊजार्) मानवजिनत स्रोत। हालांिक, संभािवत स्वास्थ्य लाभों को अिधकतम करने के िलए इन �ेत्रों से उत्सजर्न को रोकने

के िलए शहरी हॉटस्पॉट्स में अित�र� शमन उपायों क� सलाह दी जाती है। बेहतर स्वास्थ्य बोझ आकलन के िलए एक्सपोजर मॉडिलंग को उच्च अस्थायी आवृि� के साथ िकया जाना चािहए। हमारे

प�रणाम िनकट भिवष्य के प्रदूषण हॉटस्पॉट को उजागर करते हैं, और यह जानकारी शहर क� योजना, देश भर में िनयमों और िदशािनद�शों के सख्त कायार्न्वयन, वायु गुणव�ा िनगरानी और �ेत्रीय स्तर पर

िव�ेषण के माध्यम से वायु गुणव�ा प्रबंधन में उपयोगी होगी। इन िनष्कष� का िनष्कषर् है िक अपेि�त

बड़े स्वास्थ्य लाभ के साथ, प्रितकूल मौसम संबंधी प्रितिक्रया के बावजूद शमन उत्सजर्न मागर् भारत

के िलए एक उिचत व्यापार है।


xiii Contents


Acknowledgements ... iii

Abstract ...v

Hindi Abstract...…...ix

Contents ... xiii

List of Figures ... xviii

List of Tables ... xxiii

Chapter 1: Introduction………..…1

1.1 Air Pollution………...2-6 1.1.1 Air Pollutants and their sources.………....………..2-3 1.1.2 Natural and anthropogenic aerosols………3-6 1.2 Impacts of PM2.5………...7-12 1.2.1 Impact on health………..……….….7-10 1.2.2 Impact on climate………..………..10-11 1.2.3 Other Impacts………..11-12 1.3 Quantifying PM2.5 pollution in India………..12-22 1.3.1 Ground measurement………...12-15 1.3.2 Satellite data………15-17 1.3.3 Air quality models………...17-22 Statistical models………...18 Chemical transport models………...18-36 Dispersion models………...19


xiv Photochemical models………..….19-23 1.4 Identifying the research gaps………..23-24 1.5 Objectives………...24 1.6 Structure of thesis………...24-25

Chapter 2: Distribution of anthropogenic PM2.5 over India……….26 2.1 Background……….27-28 2.2 Methodology………...28-35 2.2.1 WRF-Chem setup………....29-30 2.2.2 Model Inputs………...30-35 Meteorology………...32 Geography data………..…32 Emission………...33-35 2.3 Model evaluation………....35-40 2.3.1 Evaluation of temperature and wind speed against ERA5 data………..……35-38 2.3.2 Evaluation of simulated total and anthropogenic PM2.5 against Satellite-derived


2.4 Spatial and seasonal distribution of meteorology………...40 2.4.1 Spatial and seasonal distribution of temperature and wind...41 2.4.2 Spatial and seasonal distribution of ventilation factor …….………..41-45 2.5 Spatial and seasonal distributions of PM2.5………46-50 2.6 Diurnal amplitude………...51-52 2.7 Summary………52-55

Chapter 3: Health benefit analysis from sectoral interventions………...56 3.1 Background……….57-59



3.2 Methodology………..….…59-68 3.2.1 EDGAR-HTAP emission sectors………..………..…60-63 3.2.2. Subtraction Method………..………...…63 3.2.3 Estimation of Premature Mortality attributed to PM2.5………...…64-68 3.3 PM2.5 contribution by major anthropogenic sectors in India………..68-74 3.4 Health benefits from mitigation of each emitting sector………74-79 3.5 State-level analysis………..………...…80-84 3.6 Summary………..………...84-86

Chapter 4: Sensitivity of PM2.5 distribution to the choice of emission inventory………87 4.1 Background………..………...88-90 4.2 Methodology………..……….……90-93 4.2.1 Emission Inventory………..………...…91 4.2.2 ECLIPSE emission inventory………..………..……….…91-93 4.3 Comparison of emissions in EDGAR-HTAP and ECLIPSE……….…93-99 4.3.1 Total emission comparison for PM2.5, BC, OC, SO2, and NOx……….93-96 4.3.2 Sector-wise comparison for PM2.5 emission………..…….97-99 4.4 Simulated PM2.5 comparison………..………100-113 4.4.1 Simulated BC and OC comparison………..…………...103-108 4.4.2 Simulated other inorganic aerosols………..…………...108-110 4.4.3 Simulated annual PM2.5 exposure with EDGAR-HTAP and ECLIPSE emissions at State-level………..………...110-113 4.5 Summary………..………..113

Chapter 5: Near future projection of ambient PM2.5 over India……….114 5.1 Background………..………...115-117



5.2 ECLIPSE emission pathways………..………...117-121 5.3 Statistical method………..……….…121-136 5.3.1 Multiple Linear Regression (MLR) analysis………..……….123-124 5.3.2 Meteorological Data………..………..124-125 5.3.3 Satellite-derived PM2.5 for training………..……....……...126 5.3.4 Model Training and Validation………..…………..126-130 5.3.5 Projection of anthropogenic PM2.5 in the near future under baseline and mitigation emission pathways………..………..131-136 5.4 Dynamical Method………..………...136-130 5.4.1 CWRF-Chem setup………..………....137-139 5.4.2 Input data………..………...139-140 Meteorological Input………..………..139 Chemical Boundary Condition……….…139-140 Emission input………..………...…140 5.4.3 Projected PM2.5 using CWRF-Chem under bseline and mitigation emission pathway ……...………...140-143 5.4.4. Projected PM2.5 with MLR and CWRF-Chem………..……143-144 5.4.5 Impact on meteorology………144-146 5.4.6 Seasonal change in CWRF-Chem projected PM2.5……….147-149 5.4.7 Health Impacts………..………..150 Health Burden analysis………..………..….150-151 Projected Health burden………..…….……151-158 Health benefit with mitigation emission pathway (Baseline – Mitigation)

………..158-162 5.5 Summary………..………...162-167



Chapter 6: Conclusions and future directions………..………..…168 6.1 Summary and Conclusions………..………...…169-175 6.2 Limitations of this study………..……..175 6.2 Future directions………..………...175-176

References………177-208 Annexure I: Abbreviations……….………...209-213 Annexure II: Glossary………214-217 Annexure III: List of publications from thesis…..…...………...218


xviii List of Figures

Figure captions Page No.

Figure 1.1 Global distribution of satellite-derived Surface PM2.5 concentration (µg m-3)

6 Figure 1.2 Environmental pathway for health impact from pollution source (fuel combustion)

8 Figure 2.1 Model Domain with political state boundaries and crude demarcation of the Indo-Gangetic Plains (in states Punjab, Haryana, Delhi, Uttar Pradesh, Bihar and West Bangal) and Southern Peninsula (in states Andhra Pradesh, Tamil Nadu, Karnataka and Kerala).


Figure 2.2 WRF-Chem simulated seasonal temperature (K) (upper panel), ERA- 5 reanalysis temperature (K) (middle panel), and the difference between simulated and reanalysis temperature K) (lower panel) over the domain for the four seasons winter, pre-monsoon, monsoon, and post-monsoon (each panel from left to right).


Figure 2.3 WRF-Chem simulated seasonal wind speed (m/s) (upper panel), ERA-5 reanalysis wind speed (m/s) (middle panel), and the difference between simulated and reanalysis wind speed (m/s) (lower panel) over the domain for the four seasons winter, pre-monsoon, monsoon and post-monsoon (each panel from left to right).


Figure 2.4 (a) Comparison of WRF-Chem simulated and observed PM2.5 time- series at Delhi, (b) Scatter plot between biases in simulated concentration and observed PM2.5.


Figure 2.5 (a) Scatter plot between satellite-derived and simulated

anthropogenic PM2.5 over India (b) the spatial bias (in µg m-3) in simulated anthropogenic PM2.5 (represented as satellite-derived minus WRF-Chem simulated PM2.5). Green to blueish tinge demonstrates under prediction, and yellow to brown tinge shows over prediction by the model for annual PM2.5



Figure 2.6 (a) Satellite-derived total PM2.5 (µg m-3) over India (van Donkelaar et al., 2016) (b) WRF-Chem simulated total PM2.5 (µg m-3) over India.

43 Figure 2.7 (a) Satellite-derived anthropogenic PM2.5 (µg m-3) over India (van

Donkelaar et al., 2016) (b) WRF-Chem simulated anthropogenic PM2.5 (µg m-3) over India.


Figure 2.8 WRF-Chem simulated 2 m temperature and 10 m wind over India in (a) winter, (b) pre-monsoon, (c) monsoon, and (d) post-monsoon seasons.

44 Figure 2.9 Ventilation factor over India in (a) winter (b) pre-monsoon (c) monsoon and (d) post-monsoon seasons.




Figure 2.10 Spatial distribution of simulated PM2.5 (µg m-3) over India for the four seasons (a) winter, (b) pre-monsoon, (c) monsoon, and (d) post-monsoon.

48 Figure 2.11 Monthly anomaly (%) in simulated PM2.5 with respect to annual

PM2.5 average over India.

49 Figure 2.12 Monthly anomaly (%) in observed PM2.5 average from all the available observations from CPCB monitoring stations across India during 2015- 17, with respect to annual observed PM2.5 concentration, averaged for all monitoring stations.


Figure 2.13 Monthly anomaly (%) in PM2.5, OC, and BC emission with respect to annual emission over India.

50 Figure 2.14 Difference between day (6 UTC) and night (11 UTC) PM2.5 (µg m-

3) over India in (a) winter (b) pre-monsoon (c) monsoon and (d) post-monsoon seasons.


Figure 2.15 Diurnal amplitude of PM2.5 obtained from available CBCB observation site over India between 2015-2017 in (a) winter (b) pre-monsoon, (c) monsoon, and (d) post-monsoon seasons.


Figure 3.1 PM2.5 (kt/year) emission distribution over India from (a) Residential, (b) Industrial, (c) Transportation and (d) Energy sectors given in EDGAR- HTAP.


Figure3.2 Percentage shares of Energy, Transportation, Industry, and Residential sectors in total anthropogenic PM2.5 emission for each Indian state.

Emission data is from the EDGAR-HTAP emission inventory.


Figure 3.3 Population for each 5 year age category from 25 - 29 years to >80 year age category for India, taken from the Indian census 2011.

67 Figure 3.4 The distribution of population for age group >25 years in each district of India taken from the Indian census 2011.

67 Figure 3.5 Baseline mortality for COPD, IHD, and Stroke for different states of India.

68 Figure 3.6 Shares of (a) residential, (b) industrial, (c) transportation, and (d) energy sectors to annual ambient anthropogenic PM2.5 exposure in India.

69 Figure 3.7 Percentage shares to annual ambient anthropogenic PM2.5 from each of the 4 anthropogenic source sectors (a) residential, (b) industrial, (c) transportation, and (d) energy considered in this study.


Figure 3.8 Variation of the share of residential sources to ambient anthropogenic PM2.5 exposure with the percentage of the population using solid fuel for residential use in each state/UT (represented by each dot).


Figure 3.9 Normalized ratio of population-weighted PM2.5 exposure and PM2.5

emission per square kilometer for each state.

74 Figure 3.10 Expected health benefit in terms of premature deaths (per 100,000 population) that could be avoided by complete mitigation of emission from (a)




residential, (b) industrial, (c) transportation, and (d) energy sectors.

Figure 4.1 EDGAR (left column) and ECLIPSE (second from left column) emission for the year 2010, EDGAR and ECLIPSE ratio (EDG/ECL) (third from left column) and ECLIPSE and EDGAR difference (ECL-EDG) (right column) for pollutant PM2.5 (a, b, c and d), BC (e, f, g and h), OC, (i, j, k, and l), SO2 (m, n, o and p) and NOx (q, r, s and t).


Figure 4.2 PM2.5, BC and OC emission (kt/year) from EDGAR-HTAP and ECLIPSE emission inventory in 7 different zones (North India, northwest India, Indo-Gangetic planes, northeast India, Central India, West India, and South India) of India.


Figure 4.3 The difference of primary PM2.5 emissions between EDGAR and ECLIPSE from the (a) Residential, (b) Industrial, (c) Transport, and (e) Energy sector.


Figure 4.4 (a) Satellite-derived total PM2.5 and (b) simulated PM2.5 using EDGAR and (c) using ECLIPSE. Unit is in 𝜇𝜇g m-3.

101 Figure 4.5 (a) Satellite-derived anthropogenic PM2.5 and (b) simulated anthropogenic PM2.5 using EDGAR and (c) using ECLIPSE. Unit is in 𝜇𝜇g m-3.

102 Figure 4.6 Difference between CTM simulated and satellite-derived anthropogenic PM2.5 using (a) EDGAR-HTAP and (b) ECLIPSE inventory.

102 Figure 4.7 BC (upper panel) and OC (lower panel) concentrations over India from (a and d) MERRA-2 (b and e) EDGAR-HTAP, and (c and f) ECLIPSE.

105 Figure 4.8 The difference of simulated BC from (a) EDGAR and (b) ECLIPSE with MERRA-BC between (upper panel) and simulated OC from (c) EDGAR and (d) ECLIPSE with MERRA-OC (lower panel). Unit is in 𝜇𝜇g m-3.


Figure 4.9 Comparison between model-simulated BC (values lower than 10 𝜇𝜇g m-3) and OC (values higher than 15 𝜇𝜇g m-3) and ground-based measurements of BC and OC. WRF-Chem simulated BC and OC are used from this study and in- situ observations are compiled from the literature (Pachauri et al., 2013; Ram et al., 2012; Ali et al., 2016; Kompalli et al., 2014; Joshi et al., 2016). The dotted, solid line is the 1:1 line.


Figure 4.10 Simulated other inorganic aerosols (OIA) using EDGAR-HTAP (left) ECLIPSE (middle) emission inventory and their difference (right).

110 Figure 5.1 Projected changes in primary PM2.5 emission (kt/year) from anthropogenic sources in 2030 relative to 2010 in the Indian subcontinent for the (a) baseline and (b) mitigation emission pathways.


Figure 5.2 Changes in emission (2030 - 2010) for BC, OC, SO2, and NOx represented in kt/year over India in the year 2030 following the (left panel) baseline emission pathway and (right panel) mitigation emission pathway compared to the year 2010 over India.


Figure 5.3 Total PM2.5 emission (kt/year) over the domain from 6 major sectors 122



residential (res), industry (ind), transportation (trans), energy (ene), agriculture (agr) and waste (wst) for the year 2010 and under baseline and mitigation emission pathways for the year 2030.

Figure 5.4 Scatter plot between MISR derived and MLR projected anthropogenic PM2.5 binned at each 10 µg m-3 interval for (left panel) the training period (2010-2012) and (right panel) validation period (the year 2013) under (upper panel) RCP8.5 scenario and (lower panel) RCP4.5 scenario. The error bars represent ±1σ for each bin.


Figure 5.5 The difference in projected PM2.5 (in µg m-3) in the year 2030 compared to the base year 2010-2013 under baseline emission pathway (upper panel) and mitigation emission pathway (lower panel) and in RCP8.5 (left) and RCP4.5 scenario (right).


Figure 5.6 Difference between simulated PM2.5 under baseline emission pathway and mitigation emission pathway in the RCP8.5 scenario (left) and RCP4.5 (right).


Figure 5.7 Projected changes in ambient PM2.5 concentration in 2030 relative to 2010 in the Indian subcontinent for the (top panel) baseline and (bottom panel) mitigation emission pathways under the (left) RCP8.5 and (right) RCP4.5 scenarios.


Figure 5.8 Difference between CWRF projected PM2.5 under baseline emission and mitigation emission pathway in RCP8.5 (left) and RCP4.5 (right) scenario over India in the year 2030.


Figure 5.9 Change in PM2.5 concentration in the year 2030 compared to the year 2010 over India observed with two methodologies MLR and CWRF-Chem under baseline (CLE) and mitigation (SLCP) emission pathways in RCP8.5 and RCP4.5 scenario.


Figure 5.10 The differences (baseline-mitigation) in the projected (top panel) 2m air temperature and (bottom panel) wind speed between the baseline and mitigation emission pathways in 2030 over the Indian subcontinent under the (left panel) RCP8.5 and (right panel) RCP4.5 scenario.


Figure 5.11 Change in CWRF-Chem simulated PM2.5 (µg m-3) in 2030 compared to 2010 for (a) Winter, (b) Pre-monsoon, (c) Monsoon and (d) Post monsoon baseline emission pathway RCP8.5 scenario.

148 Figure 5.12 Change in CWRF-Chem simulated PM2.5 (µg m-3) in 2030 compared to 2010 for (a) Winter, (b) Pre-monsoon, (c) Monsoon and (d) Post monsoon baseline emission pathway RCP8.5 scenario.

149 Figure 5.13 Percentage changes in exposed population (for age more than 25 years) in the year 2030 compared to the baseline year 2010 for each state of India. UT_D means all other UTs except Delhi.


Figure 5.14 Percentage change in exposed population (for age more than 25 years) at every 5-year age intervals for 2030 compared to the baseline year 2010.




Figure 5.15 Percentage changes in age-specific baseline mortality (BM) values for adult non-communicable diseases (NCDs) at the state level in 2030 relative to 2010 at the respective age ranges shown by the numbers (e.g., NCD_BM 80+

represents NCD baseline mortality for the population aged more than 80 years).


Figure 5.16 Percentage changes in baseline mortality (BM) for adult lower respiratory infection (LRI) at the state level in 2030 relative to 2010.

156 Figure 5.17 (Upper panel) Percentage changes in premature mortality burden attributed to ambient PM2.5 over India in 2030 considering the projected population and baseline mortality with the (left) baseline emission and (right) mitigation emission pathways under the RCP8.5 scenario relative to the baseline year 2010. (Lower panel) Percentage changes in premature mortality burden attributed to ambient PM2.5 over India in 2030 considering the population and baseline mortality similar to 2010 and with the (left) baseline emission and (right) mitigation emission pathways under the RCP8.5 scenario relative to the baseline year 2010.


Figure 5.18 Percentage changes in premature mortality burden attributed to ambient PM2.5 over India in 2030 with the (left) baseline emission and (right) mitigation emission pathways under the RCP8.5 scenario relative to the baseline year 2010.


Figure 5.19 Averted premature mortality attributable to ambient PM2.5 exposure over India in 2030 due to the mitigation emission pathway relative to the baseline emission pathway under the RCP8.5 scenario.



xxiii List of Tables

Table caption Page No.

Table 1.1 Advantages and disadvantages of PM2.5 observation through listed methodologies.

21 Table1.2 List of some model based studies in the field of air quality study. 22-23 Table 2.1 Physics and dynamics schemes used in the WRF-Chem set up along

with their references.

34-35 Table 2.2 List of inputs to the WRF-Chem model and their sources. 35 Table 3.1 A comparative summary of source attributed health burden estimates for the four major continuous anthropogenic sources in India.

77-78 Table 3.2 Premature deaths (95% UI are shown within parentheses) per 100,000

population in each state/union territory could have been avoided by completely mitigating emission from the corresponding sector across the country.


Table 3.3 Total premature deaths (95% UI are shown within parentheses) (× 102) in each state/union territory could have been avoided by completely mitigating emission from the corresponding sector across the country.


Table 4.1 Brief comparative description of EDGR-HTAP and ECLIPSE emission inventory for source sectors and sector comparison.

98 Table 4.2 (First value) Pearson correlation coefficient (statistically significant at 99% CI) and (second value) RMSD (in 𝜇𝜇g m-3) between simulated OC, BC, and anthropogenic PM2.5 with MERRA-2 OC, BC, and satellite-derived anthropogenic PM2.5.


Table 4.3 Population-weighted mean (±1𝜎𝜎) annual ambient anthropogenic PM2.5

exposure (in 𝜇𝜇g m-3) for 2010 simulated by WRF-Chem for each state and UTs (denoted by *) by EDGAR and ECLIPSE inventory.


Table 5.1 Correlation matrix for the dependent and independent variables.

Correlation coefficients are significant at a 95% Confidence level. ‘E’ denotes emission.


Table 5.2 Statistics of MLR predicted PM2.5 for cities with more than one million population under RCP8.5 and RCP4.5 scenario over India.

135-136 Table 5.3 Concentrations of the GHGs used in the simulation. Note that the GHG

concentrations are identical in the baseline and mitigation emission pathways.

138-139 Table 5.4 Mortality (with 95% CI is in the second row) attributable to PM2.5

exposure for states and union territories (UTs) of India in the base year 2010 and the year 2030 using the baseline and mitigation emission pathway under RCP8.5 scenario. The values are rounded off to the nearest 10s.





Related subjects :