Modeling the Impact of Human-Driven Fires on Air Quality from Regional and Global Perspectives

Abstract

Natural and anthropogenic fires emitted 2.2 Pg C yr−1 on average from 1997-2016, with widespread impacts on climate, ecosystem dynamics, air quality, and public health (van der Werf et al., 2017). In contrast to the 24% decline in global burned area over the past two decades (Andela et al., 2017), agricultural intensification and expansion have enhanced fire activity in India (Liu et al., 2019) and Indonesia (Marlier et al., 2015). In India, post-harvest crop residue burning, concentrated in the state of Punjab, severely degrades the air quality downwind in both rural and urban centers, including Delhi and the densely-populated Indo-Gangetic Plain (Kaskaoutis et al., 2014; Liu et al., 2018; Cusworth et al., 2018; Bikkina et al., 2019). In Indonesia, the combination of deforestation fires and dense network of carbon-rich peatlands spells a recipe for severe haze and elevated mortality risk during drought years (Fanin & van der Werf, 2017; Koplitz et al., 2016, 2018a). Together with daily, global monitoring of fires and vegetation from satellites, recent advances in cloud computing and geospatial analysis with the Google Earth Engine platform have accelerated the data processing and analysis pipeline by orders of magnitude (Gorelick et al., 2017). However, large uncertainty in satellite fire observations, fuel loading, small fire and cloud corrections, and emissions factors has led to substantial disagreement among global fire emissions inventories and lowered our confidence in projected health outcomes (Liu et al., 2020a). Here I use a combination of observations and models to understand factors contributing to these differences, improve current emissions estimates, and quantify spatio-temporal trends in fires in the context of human activity.

This dissertation is partitioned into two themes, focused on India (Chapter 1-3) and globally (Chapters 4-6). In north India, cascading delays in the monsoon growing season and post-monsoon agricultural burning have led to increased air quality degradation. In Chapter 1, I quantify the temporal shift in the post-monsoon fire season in the state of Punjab over 2003-2016 and discuss how a 2-week delay is connected to changes in the monsoon rice growing season, the implementation of a 2009 groundwater policy in India, and increased aerosol loading (Liu et al., 2021a). In Chapter 2, I combine survey and satellite data to build a new agricultural fire emissions inventory, SAGE-IGP (Liu et al., 2020b). In Chapter 3, I use atmospheric transport modeling with the SAGE-IGP inventory as input to quantify the effect of the post-monsoon fire season delay on air quality in downwind cities, such as New Delhi (Liu et al., 2022).

In Chapter 4, I develop metrics to understand the spatial uncertainties and biases in global fire emissions inventories, using Indonesia as a case study (Liu et al., 2020a). In Chapter 5, I discuss how tile-based processing in satellite burned area products can lead to discontinuities in burned area mapping and impact analyses (Liu & Crowley, 2021). In Chapter 6, I search globally for regions with temporal shifts in the primary fire season; I focus on north Australia and southwest Russia as two case study regions where human decisions and interventions either directly or indirectly influenced the timing of the fire season (Liu et al., 2021b).