FTIR time series of tropospheric HCN in eastern China : 1 seasonality , interannual variability and source attribution 2

30 We analyzed seasonality and interannual variability of tropospheric HCN columns in densely 31 populated eastern China for the first time. The results were derived from solar absorption spectra 32 recorded with ground-based high spectral resolution Fourier transform infrared (FTIR) spectrometer 33 at Hefei (117°10′E, 31°54′N) between 2015 and 2018. The tropospheric HCN columns over Hefei, 34 China showed significant seasonal variations with three monthly mean peaks throughout the year. 35 The magnitude of the tropospheric HCN column peak in May > September > December. The 36 tropospheric HCN column reached a maximum monthly mean of (9.8 ± 0.78) × 10 molecules/cm 37 in May and a minimum monthly mean of (7.16 ± 0.75) × 10 molecules/cm in November. In most 38 cases, the tropospheric HCN columns at Hefei (32°N) are higher than the FTIR observations at Ny 39 Alesund (79°N), Kiruna (68°N), Bremen (53°N), Jungfraujoch (47°N), Toronto (44°N), Rikubetsu 40 (43°N), Izana (28°N), Mauna Loa (20°N), La Reunion Maido (21°S), Lauder (45°S), and Arrival 41 Heights (78°S) that are affiliated with the Network for Detection of Atmospheric Composition 42 Change (NDACC). Enhancements of tropospheric HCN column were observed between September 43 2015 and July 2016 compared to the same period of measurements in other years. The magnitude 44 of the enhancement ranges from 5 to 46% with an average of 22%. Enhancement of tropospheric 45 HCN (ΔHCN) is correlated with the concurrent enhancement of tropospheric CO (ΔCO), indicating 46 that enhancements of tropospheric CO and HCN were due to the same sources. The GEOS-Chem 47 tagged CO simulation, the global fire maps and the Potential Source Contribution Function values 48

(PSCFs) calculated using back trajectories revealed that the seasonal maxima in May is largely due 1 to the influence of biomass burning in South Eastern Asia (SEAS) (41 ± 13.1%), Europe and Boreal 2 Asia (EUBA) (21 ± 9.3%) and Africa (AF) (22 ± 4.7%). The seasonal maxima in September is 3 largely due to the influence of biomass burnings in EUBA (38 ± 11.3%), AF (26 ± 6.7%), SEAS (14 4 ± 3.3%), and Northern America (NA) (13.8 ± 8.4%). For the seasonal maxima in December, 5 dominant contributions are from AF (36 ± 7.1%), EUBA (21 ± 5.2%), and NA (18.7 ± 5.2%).The 6 tropospheric HCN enhancement between September 2015 and July 2016 at Hefei (32°N) were 7 attributed to an elevated influence of biomass burnings in SEAS, EUBA, and Oceania (OCE) in this 8 period. In particular, an elevated fire number in OCE in the second half of 2015 dominated the 9 tropospheric HCN enhancement in September -December 2015. An elevated fire number in SEAS 10 in the first half of 2016 dominated the tropospheric HCN enhancement in January -July 2016. 11 12 1 Introduction 13 Atmospheric hydrogen cyanide (HCN) is an extremely hazardous gas that threaten human 14 health and terrestrial ecosystems ( exhaust, industrial processes and biomass burning. Many researchers have evaluated regional 35 emissions in various pollution regions (e.g., the Jing-Jin-Ji region, the Yangtze River Delta region, 36 and the Pearl River Delta region, Fig. S1), but the relative contribution of the biomass burning, 37 automobile exhaust, and industrial processes is seldom mentioned in the literature (Tang et al., 2012; 38 Chan, 2017; Wang et al., 2017;Sun et al., 2018a;Xing et al., 2017). This is because both industrial 39 emissions and biomass burning are major sources of the trace gases (e.g. carbon monoxide (CO), 40 formaldehyde (HCHO) and carbon dioxide (CO2)) that were used to evaluate regional emissions in 41 the literature, and it is hard to quantify their relative contribution under the complex pollution 42 condition in China (Chan et  where influences from biomass burning occurring at long distances or locally can be assessed. 55 In this study, we analyze the first multiyear measurements of tropospheric HCN in densely 56 populated eastern China. In section 2 the retrieval strategy to derive HCN from high resolution FTIR 57 spectrometry and the methods for a GEOS-Chem tagged CO simulation and potential source 1 contribution function (PSCF) calculation are summarized . In section 3 we present the seasonal and  2  interannual variability of tropospheric HCN columns measured at Hefei (32°N), China and  3  comparisons with the measurements affiliated with Network for Detection of Atmospheric  4 Composition Change (NDACC, http://www.ndacc.org/, last accessed on 3 June 2019). The potential 5 sources that drive the observed HCN variability are determined by using the GEOS-Chem tagged 6 CO simulation, the global fire maps and the PSCFs analysis in section 4. The work concludes with 7 a summary in section 5. This study aims to improve our understanding of regional biomass burning 8 characteristic and transport, and contribute to the evaluation of the global nitrogen cycle. 9 2 The diagonal elements of a priori profile covariance matrices Sa are set to standard deviation 19 of the WACCM v6 special run for NDACC, and its non-diagonal elements are set to zero. The 20 diagonal elements of the measurement noise covariance matrices Sε are set to the inverse square of 21 the SNR calculated from each individual spectrum and its non-diagonal elements are set to zero. 22 The measured instrument line shape (ILS) is included in the retrieval (

Averaging kernels and error budget 25
The partial column averaging kernels of CO and HCN at selected layers are shown in Fig. 2 2 We calculated the error budget following the formalism of Rodgers, 2000, and separated all 3 error items into systematic error or random error depending on whether they are constant over 4 consecutive measurements, or vary randomly. Table 2 summarizes the random, the systematic, and 5 the combined error budget of tropospheric CO and HCN columns. The error items included in the 6 error budget are listed in Table 1. For CO, the major systematic error is line intensity uncertainty, 7 and the major random error are zero level uncertainty and temperature uncertainty. For HCN, the 8 major systematic error are line intensity uncertainty and line pressure broadening uncertainty, the 9 major random error are smoothing error and measurement error. Total retrieval errors for 10 tropospheric CO and HCN columns between surface and 15 km are estimated to be 8.3 and 14.2%, 11 respectively. 12  implemented following the standard GEOS-Chem tagged CO simulation (http://geos-chem.org/, last 26 accessed on 8 April 2020). In this study, we only investigate the influence from the biomass burning 27 sources. The regional definition of all biomass burning tracers are shown in Fig. 1 and tabulated in 28 Table 3. 29  32 We used the potential source contribution function (PSCF) analysis method to identify air 33 masses associated with high levels of air pollutants. The PSCF assumes that back trajectories 34 arriving at times of higher concentrations likely point to the more significant pollution directions 35 (Ashbaugh et al., 1985). PSCF has been applied in many studies to locate air masses associated with In this study, PSCF values were calculated using back trajectories that were calculated by HYSPLIT. 38 The top of the model was set to 10 km. The PSCF values for the grid cells in the study domain were 39

Potential source contribution function
based on a count of the trajectory segment that terminated within each cell (Ashbaugh et al., 1985). 40 The number of endpoints that fall in the ij th cell is designated nij. The number of endpoints for the 41 same cell having arrival times at the sampling site corresponding to concentrations higher than an 42 arbitrarily set criterion is defined to be mij. In this study, we calculated the PSCF values based on 43 trajectories corresponding to concentrations that exceeded the monthly mean level of tropospheric 44 HCN column during measurement. The PSCF value for the ij th cell is then defined as: The unitless PSCF value can be interpreted as the conditional probability that the 3 concentrations of a given analyte greater than the criterion level are related to the passage of air 4 parcels through the ij th cell during transport to the receptor site. That is, cells with high PSCF values 5 are associated with the arrival of air parcels at the receptor site that have concentrations of the 6 analyte higher than the criterion value. These cells are indicative of areas of 'high potential' 7 contributions for the constituent. 8 Identical PSCFij values can be obtained from cells with very different counts of back-trajectory 9 points (e.g., grid cell A with mij = 400 and nij = 800 and grid cell B with mij = 4 and nij = 8). In this 10 extreme situation grid cell A has 100 times more air parcels passing through than grid cell B. 11 Because of the sparse particle count in grid cell B, the PSCF values are more uncertain. To account 12 for the uncertainty due to low values of nij, the PSCF values were scaled by a weighting function 13 Wij (Polissar et al., 1999

Seasonal variation 1
The monthly means of the tropospheric CO and HCN columns at the twelve FTIR stations are 2 shown in Fig. 3. As commonly observed at Hefei (32°N), three monthly mean peaks are evident for 3 tropospheric HCN and CO columns. The magnitude of the tropospheric HCN peak at Hefei (32°N) 4 in May > September > December, while for tropospheric CO column, the magnitude of the peak at 5 Hefei (32°N) in February > September > December. Note that the largest seasonal peak of HCN 6 occurs in May which is 3 months later than that of CO which occurs in February, but the other two 7 seasonal peaks for both species occur in the same months, i.e., in September and December 8 respectively. Otherwise, their seasonal cycles show similarities. 9

12
Vertical error bars represent 1σ within that month. All stations are organised as a function of decreasing latitude. 13 The tropospheric HCN and CO columns at Hefei (32°N) are higher than the NDACC FTIR 14 observations (see Fig. S2). The tropospheric HCN column reached a maximum of (9.8 ± 0.78) × 15 10 15 molecules/cm 2 in May and a minimum of (7.16 ± 0.75) × 10 15 molecules/cm 2 in November. 16 The tropospheric CO column reached a maximum of (3.38 ± 0.43) × 10 18 molecules/cm 2 in February  17 and a minimum of (2.29 ± 0.48) × 10 18 molecules/cm 2 in July ( (0.67 ± 0.03) to (1.79 ± 0.14) × 10 18 molecules/cm 2 , respectively (Table 4). 23 In the northern hemisphere, the phase of the seasonal maxima for tropospheric HCN columns 24 generally occur in spring or summer, and for CO occur in winter or spring. While in the southern 25 hemisphere, the phase of the seasonal maxima for both tropospheric HCN and CO columns occur 26 in autumn or winter. 27

Interannual variability and enhancement 28
In order to study the interannual variability of HCN and CO, fractional differences in the 1 tropospheric HCN and CO columns relative to their seasonal mean values represented by the cosine 2 fitting at the twelve FTIR stations are shown in Fig.4 and Fig.5, respectively. Enhancements of both 3 tropospheric HCN and CO columns between September 2015 and July 2016 at Hefei (32°N) were 4 observed compared to the same period of measurements in other years. For HCN, the magnitude of 5 the enhancement ranges from 5 to 46% with an average of 26%. The significant enhancements 6 occurred in December 2015 and May 2016 with peaks of 46% and 38%, respectively. By contrast, 7 the magnitude of the enhancement in tropospheric CO column at Hefei (32°N) between September 8 2015 and July 2016 ranges from 4 to 59% with an average of 27%.The tropospheric CO columns 9 were elevated over its seasonal means by more than 20% from March to April 2016. In addition, an 10 enhancement magnitude of more than 40% were occasionally observed in August and September 11 for both HCN and CO at Hefei (32°N). 12 The enhancements of both tropospheric HCN and CO columns within the same period were

Correlation with CO and enhancement ratios 3
The tropospheric HCN columns at the twelve FTIR stations from 2015 to 2018 have been 4 plotted against the coincident CO partial columns (Fig.6). In Fig.7, the correlations between the 5 tropospheric HCN and CO columns at Hefei (32°N) for all spectra recorded throughout the year 6 (gray dots) and those recorded within the selected periods (green dots) are compared. We the year probably because the portion of the fire-affected seasonal measurements at these stations 22 are larger than those at other stations (Fig.6). For the measurements at Hefei (32°N), the high 23 correlations between HCN and CO tropospheric columns deduced from the measurements without 24 March and April (R=0.67, Fig.7 (a)), in May (R=0.69, Fig.7 (b)), in September(R=0.77, Fig.7 (c)), 25 and in December (R=0.65, Fig.7     In order to determine what drives the seasonality and interannual variability of tropospheric 9 HCN in eastern China, it is necessary to match the observed time series with actual biomass burning 10 events, and show that the generated plumes are capable of travelling to the observation site. We did 11 this by using various independent data sets. 12 1. The 1-hourly instantaneous CO VMR (volume mixing ratio) profiles of the tracers listed in 13 Table 3 provided by a GEOS-Chem tagged CO simulation performed as described in Section 2.  The GEOS-Chem tagged CO simulation provides a means of evaluating the contribution of 28 CO from anthropogenic, biomass burning and oxidation sources to the measured CO columns at 29 Hefei (32°N). Source attribution is performed as follows. First, the GEOS-Chem CO VMR profiles 30 of all tracers in the grid box containing the Hefei (32°N) site were linearly interpolated and regridded 31 onto the FTIR vertical retrieval grid. This was necessary in order to account for the differences in 32 the vertical levels of the model and the FTIR (Barret et al., 2003). Then, the interpolated GEOS-33 Chem CO profiles were smoothed by the FTIR CO averaging kernel following Rodgers and Connor 34 (2003). Finally, we compared the partial columns calculated from the smoothed GEOS-Chem CO 35 profiles with the FTIR ones. Fig.8 shows the daily-averaged GEOS-Chem and FTIR CO 36 tropospheric columns (surface-15 km) for the simulation period from 2015 -2018 . The relative  37 contribution of anthropogenic, biomass burning and oxidation tracers are also shown. The GEOS-38 Chem and FTIR CO tropospheric columns are in good agreement. 1 The combination of the anthropogenic source and the oxidations of CH4 and NMVOCs is the 2 greatest contribution to the tropospheric CO column at Hefei (32°N). The magnitude of this 3 combination source varies over 80 to 95% throughout the year. In contrast, the magnitude of biomass 4 burning source varies over 5 to 20%. As shown in Fig.9, the anthropogenic, biomass burning and 5 oxidation sources are all seasonal dependent due to the magnitude of the emissions and the influence 6 of seasonally variable transport. The onset of the anthropogenic contribution begins in July with a 7 maximum in December. In contrast to the anthropogenic influence, the onset of the oxidation 8 contribution begins in January with a maximum in July, as a result of maximum NMVOC emissions 9 in Summer (Sun et al., 2018b). For biomass burning contribution, two onsets were observed. One 10 begins in January with a maximum in April and the other one begins in July with a maximum in 11 October. 12 After normalizing each biomass burning tracer listed in Table 3 to the total biomass burning 13 contribution, the normalized relative contribution of each individual biomass burning tracer to the 14 total biomass burning associated CO tropospheric column was obtained in Fig.10. The results show 15 that the seasonal maxima in May is largely due to the influence of SEAS biomass burning (41 ± 16 13.1%). Moderate contributions from EUBA (21 ± 9.3%) and AF (22 ± 4.7%), and small 17 contributions from SA (7.8 ± 2.9%), OCE (1.5 ± 0.8%), and NA (7.7 ± 1.9%) are also observed. The 18 seasonal maxima in September is largely due to the influence of EUBA (38 ± 11.3%) and AF (26 ± 19 6.7%) biomass burnings. Remaining contributions are from SA (5.1 ± 2.7%), SEAS (14 ± 3.3%), 20 OCE (8.9 ± 7.4%), and NA (13.8 ± 8.4%). For the seasonal maxima in December, contributions 21 from AF, SA, SEAS, EUBA, OCE, and NA are 36 ± 7.1%, 11 ± 1.9%, 11 ± 3.6%, 21 ± 5.2%, 4.8 ± 22 2.7%, and 18.7 ± 5.2%, respectively. 23  4.2 Attribution for transport pathway 8 For each seasonal enhancement of the tropospheric HCN, transport pathway is determined as 9 follows. First, the GEOS-Chem tagged CO simulation is used to calculate the relative contribution 10 of each biomass burning tracer (Fig. 10). For the tracer with a high contribution, the FIRMS global 11 fire map is used to search for potential fire events occurring before the phase of tropospheric HCN 12 enhancement within a one month period. Then, we generated an ensemble of HYSPLIT back 13 trajectories with different travel times and arrival altitudes to judge whether these plumes are 14 capable of travelling to the observation site. For example, for each intensive biomass burning event 15 detected at a specific period, we generated ten back trajectories at different arrival altitudes ranging 16 from 1.5 to 12 km, and modified the end time of these back-trajectories within one day of the 17 observed enhancement. If the back-trajectories intersect a region where the FIRMS fire data 1 indicates an intensive fire event and the travel duration is within a reasonable range, then this 2 specific fire event could contribute to the observed enhancements at Hefei (32°N) in eastern China. 3 The transport pathway for this enhancement is finally determined. 4 Fig. 11 demonstrates travel trajectories of the plumes occurred in AF, SEAS & OCE, EUBA, 5 and NA that reached Hefei (32°N) through long range transport. Fig. 12  Additionally, a small to moderate portion of wildfire events in central SA, eastern NA, and 29 Northern OCE in autumn or winter could transport to the observation site through large-scale 30 atmospheric circulation, which contributed 5 -20% of the tropospheric HCN in these periods. 31

Attribution for interannual variability 4
In Fig. 9, the biomass burning contribution was elevated by 5 -15% between September 2015 5 and July 2016, while no elevations were observed for anthropogenic and oxidation influence. As a 6 result, enhancements of both tropospheric HCN and CO columns between September 2015 and July 7 2016 at Hefei (32°N) were attributed to an elevated influence of biomass burning. In Fig.10, the 8 relative contribution (%) of the SEAS, EUBA, and OCE biomass burning tracers to the total biomass 9 burning associated CO tropospheric column were elevated by 5 -20%, 8 -27%, 8 -31%, 10 respectively, in the second half of 2015 compared to the same period in other years. The relative 11 contribution (%) of the SEAS and OCE biomass burning tracers to the total biomass burning 12 associated CO tropospheric column were elevated by 8 -39% and 2 -7%, respectively, in the first 13 half of 2016 compared to the same period in other years. 14 The statistical results of the FIRMS fire atlas data in Fig.14 (such as  3 rice, corn, and wheat straws) after harvest to fertilize the soil for the coming farming season. Post-4 harvest crop residue is a fine fuel that burns directly in the field and mostly by flaming in many 5 mechanized agricultural systems. In contrast, when crops are harvested by hand the residue is often 6 burned in large piles that may smolder for weeks. 7 This seasonal crop residue burning season typically occurs in the spring and summer seasons 8 and also occasionally occurs in the autumn and winter. Pollution gases, dust, and suspended particle 9 matters resulting from crop residue burning emissions result in poor air quality that threaten human 10 health and terrestrial ecosystems. The Chinese presidential decree included the prohibition of crop 11 residue burning into the Law of the People's Republic of China on the Prevention and Control of 12 Atmospheric Pollution in August 2015 (http://www.chinalaw.gov.cn, last access on 17 July 2019 ), 13 and since then the crop residue burning events were banned throughout China. Therefore, we obtain 14 a decrease in fire numbers in China since 2015. 15 6 Conclusion 16 The first multiyear measurements of HCN in the polluted troposphere in densely populated 17 eastern China have been presented. Tropospheric HCN columns were derived from solar spectra 18 recorded with ground-based high spectral resolution Fourier transform infrared (FTIR) spectrometer 19 at Hefei (117°10′E, 31°54′N) between 2015 and 2018. The seasonality and interannual variability 20 of tropospheric HCN columns in eastern China have been investigated. The potential sources that 21 drive the observed HCN seasonality and interannual variability were determined by using the 22 GEOS-Chem tagged CO simulation, the global fire maps and the PSCFs (Potential Source 23 Contribution Function) calculated using HYSPLIT back trajectories. 24 The tropospheric HCN columns over eastern China showed significant seasonal variations with 25 three monthly mean peaks throughout the year. The magnitude of the tropospheric HCN peak in 26 May > September > December. The tropospheric HCN column reached a maximum monthly mean 27 of (9.8 ± 0.78) × 10 15 molecules/cm 2 in May and a minimum monthly mean of (7.16 ± 0.75) × 10 15 28 molecules/cm 2 in November. In most cases, the tropospheric HCN columns at Hefei (32°N) are 29 higher than the NDACC FTIR observations. Enhancements of the tropospheric HCN columns were 30 observed between September 2015 and July 2016 compared to the same period of measurements in 31 other years. The magnitude of the enhancement ranges from 5 to 46% with an average of 22%. 32 Enhancement of tropospheric HCN (ΔHCN) is correlated with the coincident enhancement of 33 tropospheric CO (ΔCO), indicating that enhancements of tropospheric CO and HCN were due to 34 the same sources. 35 The GEOS-Chem tagged CO simulation, the global fire maps and the PSCFs analysis revealed 36 that the seasonal maxima in May is largely due to the influence of biomass burning in South Eastern 37 Asia (SEAS) (41 ± 13.1%), Europe and Boreal Asia (EUBA) (21 ± 9.3%) and Africa (AF) (22 ± 38 4.7%). The seasonal maxima in September is largely due to the influence of biomass burnings in 39 EUBA (38 ± 11.3%), AF (26 ± 6.7%), SEAS (14 ± 3.3%) and NA (13.8 ± 8.4%). For the seasonal 40 maxima in December, dominant contributions are from AF (36 ± 7.1%), EUBA (21 ± 5.2%), and 41 NA (18.7 ± 5.2%). 42 The enhancements of both tropospheric HCN and CO columns between September 2015 and 43 July 2016 at Hefei (32°N) were attributed to an elevated influence of biomass burnings in SEAS, 44 EUBA, and Oceania (OCE) in this period. In particular, an elevated fire numbers in OCE in the 45 second half of 2015 dominated the tropospheric HCN enhancement in September -December 2015. 46 An elevated fire numbers in SEAS in the first half of 2016 dominated the tropospheric HCN 47 enhancement in January -July 2016. 48 Most high resolution FTIR instruments are located in Europe and Northern America, whereas 49 the number of sites in Asia, Africa, and South America is very sparse. As one of few FTIR stations 50 on Asian continent, the long-term observations of trace gases at Hefei are crucial to understand 51 global warming, regional pollution, long term transport, and contribute to the evaluation of satellite 52 data and model simulations. 53 54 Data availability. The CO and HCN measurements at the selected NDACC sites can be found by 55 the link http://www.ndaccdemo.org, and the CO and HCN measurements at Hefei are available on 56 request. 57