Constraining black carbon emission inventories with observations

Sun, Tianye

Permalink

https://hdl.handle.net/2142/106417 Copy

Description

Title

Constraining black carbon emission inventories with observations

Author(s)

Sun, Tianye

Issue Date

2019-08-15

Director of Research (if dissertation) or Advisor (if thesis)

Bond, Tami C

Doctoral Committee Chair(s)

Bond, Tami C

Committee Member(s)

• Rood, Mark J
• Riemer, Nicole
• Koloutsou-Vakakis, Sotiria

Department of Study

Civil & Environmental Eng

Discipline

Environ Engr in Civil Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Black Carbon
• Emission Inventory
• Observations
• Climate Change
• Diesel Engine
• Coal Combustion
• Model Resolution

Abstract

Black carbon (BC) is a tiny particle in the atmosphere that plays an important role in affecting the Earth’s energy balance. It can absorb solar radiation, interact with clouds, and reduce the albedo of snow. More accurate quantification of the long-term climatic influence of BC requires estimates of its emissions over time. This dissertation improves the long-term estimation of historical BC emissions by applying observational constraints and provides more accurate and comprehensive emission information for modeling studies. This work develops a method to constrain the historical BC emission inventory for the U.S. from 1960 to 2000 based top-down constraints (ambient observations) and bottom-up constraints (emission measurements). Measurements of BC concentration in California and New Jersey are compared with predicted concentration trends for 1960 to 2000. Based on seasonal and weekly patterns in observations, discrepancies between observations and predicted concentrations are attributed to particular sources, indicating needed adjustment in the emission factors. Then, emission measurements further substantiate these adjustments for the identified sources. In addition, this work also investigates other possible sources for discrepancy between model and measurement, including bias in modeled meteorology, model resolution, and inconsistencies in reported fuel consumption, which were not addressed in previous emission constraining studies. Overall, the method built in this work helps improve the historical BC emission inventory and the discrepancy analysis provides better quantification of the biases in model-measurement comparison. Observational constraints indicate emission factor changes in vehicle, industry and residential sectors. Emission factors for pre-regulation vehicles increase by 80% to 250% compared to the previous inventory, and emission factors for residential heating stoves and boilers increase by 70% to 200% for 1980s and before. In addition, new technologies are included to account for in naturally aspired off-road engines and for certified wood stoves. The emission inventory updated by this work shows a significant change from previous inventories. In total, the updated U.S. BC emissions decrease from 690 Gg/yr in 1960 to 200 Gg/yr in 2000. The new inventory is higher than the a priori estimation (Bond et al., 2007; Lamarque et al., 2010) by about 80% between 1960 and 1980, and shows a decreasing trend that did not appear over this period. This work provides the first observationally-constrained U.S. black carbon emission inventory with explicit representation of activities and technologies between 1960 and 2000, and uncovers a decreasing climate forcing from BC in the U.S. for 1960 to 2000. Emission constraining studies have been relying on comparison of model against measurement, but the influence of model resolution has not been fully addressed. This work investigates the discrepancies caused by model resolution on model-measurement comparison of surface BC for urban and rural monitoring network sites in the U.S. Increasing model resolution from 2˚ to 0.5˚ increases predicted BC by 30% and 106% for rural and urban networks in California, respectively. This discrepancy with 2˚ model could explain 24% to 41% of the total discrepancy in model-measurement comparison. For rural sites elsewhere in the U.S., increasing resolution from 2˚ to 0.5˚ results in a nearly even division between over- and under-prediction at the higher resolution, with an averaged discrepancy of 6%. Adjustment for resolution discrepancy is recommended for California sites, but not for other rural sites since the averaged effect is small. Adjustment factors describing the model resolution discrepancy for California coefficient of haze network and U.S. IMPROVE network are presented.

Graduation Semester

2019-12

Type of Resource

text

Permalink

http://hdl.handle.net/2142/106417

Copyright and License Information

Copyright 2019 Tianye Sun

Owning Collections