Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Community
Ask the community for help and clear up your study doubts
Discover the best universities in your country according to Docsity users
Free resources
Download our free guides on studying techniques, anxiety management strategies, and thesis advice from Docsity tutors
This thesis discusses the measurement of aerosol optical properties in Seoul using the UW-Madison HSRL. The author acknowledges and thanks their mentors and those who read and commented on the thesis. tables and figures showing the methods used to estimate column AOD and the time-series of monthly bulk lidar ratio. The document could be useful as study notes or a summary for a course on atmospheric and oceanic sciences.
Typology: Thesis
1 / 78
This page cannot be seen from the preview
Don't miss anything!
by
Coda Phillips
A thesis submitted in partial fulfillment of the requirements for the degree of
Master of Science
(Atmospheric and Oceanic Sciences)
at the
The thesis is approved by the following members of the Final Committee: Steve Ackerman, Professor, AOS Grant Petty, Professor, AOS Tristan L’Ecuyer, Professor, AOS
i
I would like to acknowledge and thank Robert Holz, Ralph Kuehn, and Willem Marais, my mentors throughout my graduate work. My thanks also go to Ilya Razenkov and Ed Eloranta for helping me with technical lidar issues. Also, this thesis was composed with data graciously provided by Xian Peng and Sang-Woo Kim. Finally, I’d like to thank those who read and commented on the thesis: Steve Ackerman, Grant Petty, and Tristan L’Ecuyer.
iv
2.1 Description of symbols............................ 12
vi
3.2 The AERONET-based lidar ratio and backscatter-weighted relative hu- midity averaged every 20min are plotted for each season. A linear fit was computed for each season and for the entire year. The monthly-mean values are also plotted in red......................... 27 3.3 Various estimates of monthly-median aerosol optical depth (AOD) in Seoul. Top: All data sampled as AERONET and the effect of lidar ratio is demonstrated by comparing the constant lidar ratio assumption (blue) and interpolated mean lidar ratio (green). Bottom: AERONET AOD and HSRL estimates of AOD. "Cloudless" is defined as having no clouds or virga below the top of integration at 10km. "Cloudless 5km" only integrates to 5km and includes scenes with clouds between 5km and 10km. "Mol. AOD" is the molecular-derived AOD and is sampled the same as "cloudless". All HSRL estimates except "Mol. AOD" assume a bulk lidar ratio interpolated from AERONET-derived monthly means. 29 3.4 AERONET and HSRL AOD estimates aggregated by season and hour- of-day. Descriptions of HSRL various estimates are given in Fig. 3.3. Aggregation excludes hours without adequate representation in each month of the season to prevent some leakage from seasonal AOD trends when sunrise and sunset times co-vary. The seasonal-mean (all inclu- sive) is plotted as a dotted horizontal line.................. 31 3.5 Diurnal deviation of AERONET and HSRL AOD estimates aggregated by season and hour-of-day. Similar to Fig 3.4. but daily mean is subtracted. 32 3.6 Daily mean deviation from weekly mean grouped by day-of-week. Inte- grated aerosol backscatter coefficient is used for all but one (Molecular AOD) of the HSRL estimates. Interquartile range (IQR) is also plotted to show variance................................ 33 3.7 Time-series of coarse-mode and dust column optical properties. The black line is monthly-median AERONET coarse-mode AOD. The green line is monthly-median matched dust backscatter integrated up to 10 km. Interquartile range (IQR) is plotted to show variance........ 35
vii
3.8 Seasonal comparison of coarse-mode AOD and integrated dust backscat- ter coefficient. Blue dots represent individual matchups between AERONET and HSRL. Red dots show the monthly means for each season. The slope (m) and coefficient of determination (r^2 ) from linear regression are shown in the legend............................ 36 3.9 Time-series of decomposed bulk lidar ratio. The black line repeats the monthly bulk lidar ratio from Fig. 3.1 (using median instead of mean has little effect). The green line shows the dust bulk lidar ratio (de- fined as coarse-mode AERONET AOD divided by integrated HSRL dust backscatter). The blue line shows the non-dust bulk lidar ratio (defined as fine-mode AERONET AOD divided by integrated HSRL non-dust backscatter)................................... 37 3.10 Vertical profile of aerosol backscatter coefficient below 4km. The median value and inter-quartile range (IQR) for each season is displayed.... 38 3.11 Vertical profile of aerosol backscatter coefficient between 4km and 10km. Similar to Fig. 3.10 but at higher altitude and smaller y-scale....... 39 3.12 Seasonal-mean vertical profiles of dust and non-dust backscatter coef- ficient and dust fraction at 7.5m vertical resolution. The dashed black line shows seasonal-mean non-dust backscatter coefficient (νnd). The solid black line has been corrected for hygroscopic growth. The blue line shows dust backscatter coefficient (νd). The dashed green line shows the ratio of median dust to median aerosol backscatter coefficient ( (^) νndν+dνd ). The solid green line is the same ratio, but using the corrected non-dust backscatter coefficient (dashed black line)............ 41 3.13 Seasonal-mean vertical profiles of dust and non-dust mass concentra- tion from NAAPS aerosol model. The black line shows the non-dust mass concentration. The blue line shows dust mass concentration. The dashed gray line shows the median dust fraction of median total mass concentration.................................. 42
ix
3.18 Time-series of median column dry fine-mode mass (integrated PM2.5). The green line is NAAPS aerosol model fine-mode integrated mass con- centration which outputs at 6-hour time resolution. The blue line is the HSRL-derived column fine-mode mass up to 10km from the cloudless subset. The purple line is the HSRL-derived column fine-mode mass (up to 5km) from the "cloudless 5km" subset. Interquartile range (IQR) is plotted to show variance. Derivation of mass is detailed in methods. The median HSRL-derived AOD from "cloudless" subset is plotted as a dashed line for comparison.......................... 50 3.19 Seasonal comparison of near-surface aerosol backscatter coefficient dust fraction and surface relative humidity. Blue dots show hourly relative humidity from ERA5 and the matched HSRL aerosol backscatter coeffi- cient dust fraction nearest in time. Both are sampled near the surface. Coefficient of determination (r^2 ) from linear regression is shown in the legend...................................... 52 3.20 Seasonal comparison of near-surface non-dust backscatter coefficient and surface relative humidity. Figure details similar to Fig. 3.19..... 53 3.21 Seasonal comparison of near-surface dust backscatter coefficient and surface relative humidity. Figure details similar to Fig. 3.19........ 54
Coda Phillips
Under the supervision of Professor Steve Ackerman At the University of Wisconsin-Madison
The University of Wisconsin High Spectral Resolution Lidar (HSRL) was oper- ated continuously at Seoul National University as part of the Korea-United States Air Quality Study (KORUS-AQ). Using the UW-Madison HSRL, AOD can be de- rived at any time of day and with great frequency. Direct daytime comparisons of the HSRL-derived and AERONET-derived AOD show good agreement. However, the HSRL makes a much larger volume of measurements available. These mea- surements show qualitative differences from AERONET in the mean diurnal cycle for some seasons. Furthermore, AOD tendency at night can also be characterized. The HSRL is also able to separate dust and non-dust using signal depolarization, average dust and non-dust backscatter is highest near the surface and decreases with increasing altitude. Meanwhile, the fraction of backscatter attributable to dust is small near the surface and increases to about 50% in the free troposphere. Finally, surface PM2.5 and near-surface backscatter were found to be highly correlated and the ratio was dependent on relative humidity.
tively well established, as well as the primary factors driving the climatological mean. It’s important to review the present state of our understanding of the aerosol climatology in Korea, so the UW-Madison HSRL results can be presented in context and better interpreted.
First, understanding the meteorological climate is critical to understanding the aerosol climate as they are closely correlated. The transport of either dust or pollution depends on the wind. In the absence of transport, meterological factors also influence the local response of the aerosol environment to local emissions. The meteorological climate of Seoul is classified as Koppen humid continental with dry winter (Peel et al., 2007). It is located in a frontal regime, and the polar jet will often pass overhead during the year. Mid-latitude cyclones as well as typhoons often impact the Korean peninsula. The East Asian monsoon exerts the most significant control over the intra-annual variability in the region’s meteorology. In the winter, the northeastern monsoon winds bring cold Siberian air south into the East Asian region. When spring arrives, lows that form in the west cause outbreaks of dust, known as HwangSa. This is also when the highest aerosol optical depth (AOD) is recorded (Kim et al., 2007b). Summer ushers in the southwest monsoon. The slow southwesterly flow typical in the summer make transport from China to Korea favorable, which can dramati- cally increase the aerosol loading. Stagnant conditions are also more likely to occur in the summer, leading to a locally produced increase in loading. A majority of the precipitation falls during the months of June and July, during the passage of a quasi-stationary convergence zone, called Changma in Korea. Effects of the summer monsoon in Seoul typically begin between late June and mid-July and last 20-30 days (Chang, 2004). Wet deposition from precipitation efficiently removes aerosol in the boundary layer and decreases the overall aerosol loading. AODs decrease afterwards (Kim et al., 2007b).
Anthropogenic Particulate Matter In East Asia, there are two dominant types of aerosol particulate matter (PM): mineral dust and anthropogenic pollution. Anthropogenic PM is most common in large cities with the fine-mode size distribution (aerodynamic diameter <2.5 μm) (Seinfeld and Pandis, 1998). In Seoul, the chemical composition of fine aerosol near the surface is known to be dominated by organic material (42% of PM 1 , 34% of PM2.5), likely from anthropogenic sources and secondary formation pathways. An almost equal part is composed of secondary inorganic aerosol like nitrate, sulfate, and ammonium (~50%) (Kim et al., 2018b). Together, these make up about 90% of the total submicron particulate mass and most of the remaining 10% is elemental carbon (also known as black carbon) which is directly emitted soot (Heo et al., 2009; Kim et al., 2018b, 2007a; Kang et al., 2006). Components of anthropogenic PM in East Asia are moderately hygroscopic (Anderson et al., 2003). Anthropogenic emissions are known to vary with the seasons in this region. Variation is caused mostly by the residential sector as opposed to industry, power, or transportation, which remain relatively constant throughout the year (Ma et al., 2018). In China, a majority of carbonaceous aerosol (elemental and organic) is emitted by the residential sector, a large portion of which is from heating using solid fuel (Ma et al., 2018).
Dust Dust events exhibit a strong seasonality in Korea, with 87% of dust events occurring in the spring, connected to the typical north-westerly prevailing flow (Kim et al., 2007b; Kim, 2008). In Korea, the predominant sources of dust are the high-altitude deserts in China and Mongolia (Murayama et al., 2001). However, dust generated from roadways and construction is also thought to contribute. From airborne in-situ measurements taken during ACE-Asia, dust was found to be primarily coarse-mode (aerodynamic diameter >2.5 μm), nearly non-absorbing and also nearly non-hygroscopic (Anderson et al., 2003).
concentration occurs in the winter. This also seems to be true for most chemical components like sulfate, nitrate, and organic carbon. The only exception is non-sea- salt calcium, which peaks in the springtime when dust transport is highest (Kim et al., 2018a). The important factors driving PM2.5 seasonality are boundary layer height, seasonal transport patterns, and wet deposition due to rainfall (Lee et al., 2013; Kim et al., 2018a).
Diurnal Cycle Both column-based AOD and surface-based PM2.5 are thought to exhibit a diurnal cycle. The cycle is a result of cycles in human activity, meteorological variables, and photochemistry. Previous studies in Korea using AERONET have established the basic pattern of peaks in AOD at 9am and 2pm local time and a minimum at noon (Lennartson et al., 2018). However, the HSRL day/night retrievals presented in this thesis indicate a different and strongly seasonal pattern in AOD diurnal cycle.
Recent trends in emissions and air quality in East Asia make characterization of aerosols a moving target, motivating regular monitoring and reevaluation of the state of the science. In the past two decades, Korea has experienced an impressive decline in the surface aerosol concentration (Ahmed et al., 2015; Kim and Lee, 2018). From 2000 to 2015 the annual mean PM 10 concentration in Seoul declined from 70 μgm−^3 to 40 μgm−^3 (Heo et al., 2017). PM 10 decreased in six out of seven major cities observed from 1996 to 2010 (Sharma et al., 2014). MODIS AOD retrievals corroborate this decline in aerosol loading, showing a 22% decline in AOD from 2000 to 2010 (Panicker et al., 2013). Meanwhile, China’s PM2.5 concentrations have increased from 1999 to 2011 (Peng et al., 2016). Annual mean AOD had a similar increase from 2000 to 2007 (Qin et al., 2018). Such anthropogenic changes can be expected to alter many aspects of the aerosol climatologies previously established in the region including chemical composition, seasonality, vertical distribution, and amount.
Poor air quality in Korea has been partially attributed to transport of aerosol and gaseous precursors from China; however, there is still much uncertainty due to the difficulty of establishing such a causal link. That said, in a modeling study, Kim et al. (2017b) found that 60% of Seoul’s surface particulate matter was attributable to foreign emissions, and other modeling studies have found similar results (Koo et al., 2008; Kim et al., 2017a). Trends in air quality can also be attributed to changing climate. Kim et al. (2017c) found that recent decreases in mean wind speed since 2012 have reversed past improvements in air quality in Seoul.
The novel observations made by the UW-Madison HSRL are a core component of this thesis. In this section, the state of aerosol observations is summarized with emphasis on the particular strengths and weaknesses inherent to each instrument.
The state-of-the-art instrumentation for aerosol remote-sensing are ground-based sun-photometers and satellite visible imagers. AERONET and MODIS are the most common remote-sensing platforms used for both forecasting and analysis. Passive observations from ground (AERONET) or from space (MODIS) are limited to only column aerosol optical properties and cannot be made at night or in cloudy scenes. The AERONET instrument has a global network, but only provides information at discrete locations. Satellite imagers such as MODIS provide global coverage; however, unlike AERONET, MODIS cannot observe the diurnal cycle of aerosol properties because it is in a sun-synchronous orbit with overpasses at the same local time every day. The recent launch of Himawari-8 and GOCI geostationary satellites with coverage in Asia provide similar spectral measurements to MODIS but with continuous 10 minute observations. However, as with the LEO imagers the AOD retrieval only provides the total column AOD for cloudless, daytime observations,
transport in Seoul in January 2018 (Coffey, 2018). For this reason, continuous in-situ aerosol measurements of PM2.5 are common in large cities.
Interactions between aerosols and clouds are difficult to measure, in-part due to the inability of most remote sensing instruments to observe cloudy scenes. The challenge with observing cloudy scenes with passive instrumentation is not only the obvious obscuration, but also the 3D radiative effects of scattering near the cloud, which foils passive retrievals typically used for aerosol properties (Várnai and Marshak, 2009). This is not a worry for an active instrument like the UW- Madison HSRL or CALIOP, which makes these instruments important supplements to passive instrumentation. Confusion between clouds and thick haze aerosol is another problem that afflicts both passive and active aerosol measurement in East Asia (Shi et al., 2014). There is also the possibility of that thin cirrus, undetected by cloud screening, can contaminate aerosol retrievals. As a result of aerosol remote sensing being dominated by column-based mea- surements, data on the vertical distribution of aerosol in the atmospheric column is limited. The space-based lidar CALIOP is the only global measurement ca- pable of resolving vertical distribution, but it is still limited by low SNR and a sun-synchronous orbit, limiting diurnal sampling. The paucity of information on vertical structure impacts modeling efforts and the task of estimating surface concentrations from the available column-only remote sensing data like MODIS or AERONET (Toth et al., 2019).
In March of 2016, the UW-Madison HSRL was deployed on the campus of Seoul National University in Seoul, South Korea for the KORUS-AQ field campaign, an international collaboration between Korea and the US to study air quality in the region. Aside from the UW-Madison HSRL, airborne measurements from the NASA DC-8, ground-based in-situ and remote sensing, and satellite measurements
participated. Despite the intensive operations only lasting approximately one month, the UW-Madison HSRL remained operational for almost 2 years, until the laser began experiencing issues in February 2018. Funded by NASA, the primary objectives of KORUS-AQ relate to identification of the factors controlling air quality on the Korean Peninsula. Secondary formation of aerosol and point source emissions were of particular interest. The months (May-June) for intensive operations were specifically chosen to maximize trans- port of pollutants and photochemical processes. Weather conditions were mostly stagnant during the experiment. One exception was the period from May 25-28, where conditions allowed direct transport from China; this coincided with the most prominent of the few days where Korean PM2.5 standards were exceeded during the experiment. Considering the large influence of weather on air quality and the variability of weather conditions, such a short experiment cannot be considered automatically representative. The role of the UW-Madison HSRL is to provide high-quality, vertically resolved measurements of aerosol above Seoul, as a long-term seasonal context for the intensive observations conducted in the area.
