2.1 Satellite observations

The Atmospheric Chemistry Experiment Fourier transform spectrometer (ACE-FTS) is a high-spectral-resolution (0.02 cm⁻¹) limb sounder instrument on board the Canadian Science Satellite mission (SCISAT). SCISAT moves along a circular low Earth orbit at an altitude of 650 km with an inclination of 74^∘ and an orbit period of 97.7 min (Bernath et al., 2005). This orbit gives a sampling pattern with a high density of observations at high latitudes. The original aim of the SCISAT mission was to obtain complete insight into the physical and chemical processes driving the distribution of ozone by measuring changes in atmospheric composition, in particular in polar regions. ACE-FTS operates in the infrared absorption spectrum over the range 750–4400 cm⁻¹ (2.2 to 13.3 µm) using the solar occultation technique, measuring a wide range of molecules in the upper troposphere and stratosphere (Bernath et al., 2005). Its profiles have been widely used as a reference for interpreting ground-based and nadir satellite measurements of HCN and to help in validating tropospheric–stratospheric transport in atmospheric models (Park et al., 2013; Viatte et al., 2014; Glatthor et al., 2015; Sheese et al., 2017). The retrieved HCN profiles extend from the middle troposphere (∼ 6–8 km) to ∼ 42 km with a vertical resolution on the order of ∼ 3 km (Boone et al., 2005, 2020). ACE-FTS observations extend higher than the profiles measured by balloon-borne missions (Kleinböhl et al., 2006), allowing us to investigate stratospheric HCN variability and to test the different processes driving HCN loss. Here we present ACE-FTS version 4.1 HCN data (Bernath et al., 2021), the most recent update version, and use it to evaluate different modelled HCN tracers.

Figure 1 shows the seasonal ACE-FTS of the sample year 2008–2009 HCN zonal mean cross sections in 10^∘ latitude bins. To highlight the HCN stratospheric distribution, the dashed black line shows the seasonal average location of the tropopause, based on the World Meteorological Organization (WMO) thermal tropopause definition (Maddox and Mullendore, 2018) using ECMWF ERA-Interim reanalyses (Dee et al., 2011). The first panel of Fig. 1 shows the December–February (DJF) season, which exhibits a high upper-tropospheric HCN concentration in the Southern Hemisphere (SH) from midlatitudes to high latitudes. In the March–May (MAM) season an enhancement of upper tropospheric HCN is observed in the Northern Hemisphere (NH) from the tropics to midlatitudes, while in the June–August (JJA) season the enhancement is observed in the NH midlatitudes to high latitudes. The observed behaviour is attributable to biomass burning emissions during the wildfire seasons in the tropical and midlatitude regions, respectively. During the September–November (SON) season an enhancement of HCN in the upper troposphere is observed over southern tropics, due to biomass burning emissions from South America, Africa, and South-East Asia.

Figure 1Latitude–height cross sections of ACE-FTS HCN zonal mean profiles (pptv) in 10^∘ latitude bins for four seasons, December–February (DJF), March–May (MAM), June–August (JJA), and September–November (SON), from December 2008 to November 2009. The dashed black lines show the season-averaged location of the thermal tropopause (Maddox and Mullendore, 2018) based on ECMWF ERA-Interim reanalyses for the same periods.

Download

Figure 2 shows the time–altitude cross sections of the 2-month mean HCN mixing ratios measured by ACE-FTS averaged over three 30^∘ latitude bands. Here, the strong HCN tropospheric seasonal variability is clearly visible and is followed by a similar seasonal cycle in the stratosphere at all latitudes, in agreement with Park et al. (2021). During the months following the period of high tropospheric HCN concentration, a large amount of HCN is transported to the stratosphere, where it persists for the following years. In particular, after the El Niño events, which influence the fire season of South-East Asia, this amount is notably high. Looking specifically at the tropical region, two of the largest El Niño events ever recorded in Indonesia, in 2006 and 2015, are highlighted by an extremely high concentration of HCN emitted during peat fires. During the months following the two events, a large quantity of HCN was transported to the stratosphere and persisted for a longer time than the typical seasonal variation. Specifically, after the 2015 Indonesian fire season, the HCN transported to the stratosphere persisted for the following 2 years before being completely reduced. Sheese et al. (2017) observed the transport of the HCN from the upper troposphere to the lower stratosphere during January and February 2016 and its persistence during the entire year using ACE-FTS measurements. This behaviour is also visible in the latitude band 15–45^∘ S.

Figure 2Time–altitude sections of 2-month mean HCN mixing ratios (pptv) from ACE-FTS averaged into three 30^∘ latitude bins, 45–15^∘ N, 15^∘ N–15^∘ S, and 15–45^∘ S, for 2005–2020. Note that the colour scale saturates and the maximum observed HCN mixing ratio in the tropical band in 2015 is over 1300 pptv.

Download

2.2 Ground-based FTIR observations

NDACC is an international global network of more than 90 ground-based stations that measure atmospheric composition in order to detect long-term changes and trends in the chemical and physical state of the atmosphere (De Mazière et al., 2018). The stations employ a variety of techniques and instruments, and observations at some sites extend from the early 1990s. Because of the different instruments and station set-ups, not all species are available at all stations. The Fourier transform infrared (FTIR) instruments are used at around 25 NDACC sites. They consist of a high-quality FTIR spectrometer combined with a high-precision solar tracker, which records the direct solar absorption spectra in the mid-infrared spectral region. The instruments work only under clear-sky conditions; i.e. the instrument view must be free of clouds, and no measurements are possible during the night, including, of course, polar night. This impacts the ground-based FTIR sampling frequency, with typically an average of 120 observation days every year, when considering all sites. Note, however, that for some stations, yearly coverage can be significantly larger, lying regularly in the range 240–270 d of measurements every year. The NDACC sites which measure HCN vertical columns are distributed globally, with a higher concentration in the NH, especially in Europe and North America. Here we selected HCN ground-based measurements from four sites (Table 1), having a long and continuative measurements time series: Thule (Greenland) at NH high latitudes, Bremen (Germany) and Jungfraujoch (Switzerland) at NH midlatitudes, and Mauna Loa (Hawaii, USA) in the tropics.

Table 1Ground-based NDACC FTIR observation sites used in this study.

Download Print Version | Download XLSX

3.1 Upper-troposphere–stratosphere HCN loss

The TOMCAT simulation included four idealized HCN tracers (HCN1 to HCN4) to test the different atmospheric loss mechanisms of HCN. For each of these tracers, the global surface HCN volume mixing ratio (VMR) was constrained to a fixed value of 200 ppt, approximately the background tropospheric VMR measured during the Transport and Chemical Evolution over the Pacific (TRACE-P) aircraft campaign and modelled by Li et al. (2003) and Singh et al. (2003). The HCN atmospheric chemistry was modelled using the parameters summarized in Table 2. The main focus here is on the HCN removal process via oxidation by OH radicals and the comparison of the two different reaction rates, the recommended rate proposed by JPL (Burkholder et al., 2015, 2019) and the rate presented by Kleinböhl et al. (2006) based on the experimental measurements of Strekowski (2001). The other loss process included in the four tracers is the HCN reaction with O(¹D) (Kleinböhl et al., 2006; Strekowski, 2001). The loss of HCN by photolysis is very slow and can be ignored (Burkholder et al., 2019).

Table 2HCN atmospheric photochemical loss mechanisms used for the different TOMCAT model tracers.

^a The rate coefficient is expressed in cm³ molec.⁻¹ s⁻¹, and T_prof is the temperature profile expressed in kelvin (K). ^b k₀ is given in cm⁶ molec.⁻² s⁻¹, k_∞ is in cm³ molec.⁻¹ s⁻¹, and [M] is the molecular air density.

Download Print Version | Download XLSX

Figure 3 compares the four model tracers with example average profiles measured by ACE-FTS averaged over 60^∘ latitude bands. Tracer HCN1 includes only the JPL-recommended rate for HCN oxidation by a reaction with OH. This tracer substantially underestimates the amount of HCN in the stratosphere. The HCN1 profile shows a drastic reduction in the HCN VMR at altitudes above ∼20 km, where the measured value is greater than 150 pptv. HCN2 is used to evaluate the HCN oxidation by OH radicals using the Kleinböhl et al. (2006) rate constant. For this tracer the model VMRs are closer to the measurements but, above 30 km at high latitudes in both hemispheres, the modelled HCN amount clearly overestimates the ACE observations by about 60 pptv. Introducing the HCN destruction by O(¹D) this gap between observations and the model is greatly reduced. Considering the O(¹D) sink alone (tracer HCN3) or in combination with the HCN reaction with OH (tracer HCN4), we obtain a much more reasonable agreement with the measured HCN profile in the middle–upper stratosphere. Tracers HCN3 and HCN4 show that the stratospheric HCN loss is driven by the reaction with O(¹D). This good agreement is further support that loss of HCN by photolysis in the stratosphere is negligible.

Figure 3Average HCN profiles for March–May 2009 for four TOMCAT model tracers (HCN1–HCN4) compared with the profiles measured by ACE-FTS with ±1 median absolute deviation for latitude bands 90–30^∘ N (a), 30^∘ N–30^∘ S (b), and 30–90^∘ S (c). See Table 2 for details of the photochemical loss reactions included in the different model HCN tracers.

Download

3.2 Ocean uptake

Li et al. (2000) and Li et al. (2003) both modelled HCN variability and developed two different schemes to reproduce the ocean uptake (Li et al., 2000, 2003). The first scheme was developed to test, using the 3-D model GEOS-Chem (Goddard Earth Observing System chemistry model), the hypothesis that the biomass burning provides the main HCN source and the ocean uptake provides the main sink by focusing on the HCN seasonal features (Li et al., 2000). The Li et al. (2003) scheme was introduced into the GEOS-Chem model to perform new simulations with a longer HCN lifetime and weaker global sources than those of the previous study (Li et al., 2000), in order to match the TRACE-P constraints. The authors of Li et al. (2003) assert that their model achieved a similar or better simulation than in the study of Li et al. (2000).

In our study, a second set of six HCN tracers was implemented in TOMCAT to evaluate the two different ocean uptake schemes from Li et al. (2000) and Li et al. (2003). For these tracers the atmospheric photochemical loss was included as in tracer HCN4, with the HCN reactions with OH (using the rate constant from Kleinböhl et al., 2006) and O(¹D), as this tracer gave the best model representation of stratospheric HCN variability compared to ACE-FTS.

The ocean uptake flux of HCN proposed by Li et al. (2000) is defined as

where $k_w=\mathrm{0.31}{u}^{\mathrm{2}}(S_c/\mathrm{666}{)}^{\mathrm{1}/\mathrm{2}}$ (m s⁻¹) is the air-to-sea transfer velocity with u (m s⁻¹), the wind speed at 10 m, and the dimensionless parameter $S_c=\mathit{\nu }/D$ is the Schmidt number of HCN in water, with ν (m² s⁻¹) as the kinematic viscosity and D (m² s⁻¹) as the diffusion coefficient of HCN in water. C_g (kg m⁻³) is the concentration of HCN in surface air, K_H is the temperature-dependent Henry’s law constant defined as K_H=12 M atm⁻¹ at 298 K, and $\mathrm{\Delta }{H}_{\mathrm{298}}/R=-\mathrm{5000}$ K. R=287.05 J kg⁻¹ K⁻¹ is the gas constant, and T (K) is the sea surface temperature.

The second ocean uptake scheme from Li et al. (2003) uses the results of a box model of the marine boundary layer (MBL) to derive an oceanic deposition velocity of 0.13 cm s⁻¹ (Singh et al., 2003). The resulting flux is defined as

where C_g (kg m⁻³) is the HCN concentration near the surface.

The NDACC ground-based column measurements are used to evaluate the ocean uptake schemes, which act as the HCN surface sinks in the model, due to the large impact on the HCN budget and mean tropospheric VMR. The TOMCAT simulation is sampled at each NDACC station location, and the profiles are smoothed using the instrument-specific averaging kernels. The total column time series are then compared as shown in Figs. S1–S4 in the Supplement. The agreement is evaluated considering the values of the root mean square error (RMSE) and the coefficient of determination (R²) in reference to the measurements (Tables S1–S2 in the Supplement).

Table 3HCN ocean uptake rates used for the different TOMCAT model tracers.

Download Print Version | Download XLSX

Tracers L2000 and L2003, which make use of the two published schemes from Li et al. (2000) and Li et al. (2003), respectively, show a good agreement with NDACC total column time series only in terms of the HCN seasonality. In fact, L2000 and L2003 show the highest RMSE values among all the tracers and a strongly negative R² at each location. The two tracers give greatly different results in our 3-D model compared to the HCN amount measured by FTIR instruments; L2000 underestimates the HCN total columns, with the values being almost two-thirds of the observed values, while L2003 greatly overestimates them, with the model being almost double the FTIR-measured values. The observed mismatch is attributable to the ocean uptake fluxes being unable to capture contributions of this process to the HCN variability. We therefore need to scale the two sinks, in different directions, in order to reach a better representation of HCN variability. The new tracers, L2000_0.5 and L2000_0.25, are created by applying the scaling factors 0.5 and 0.25, respectively, to reduce the Li et al. (2000) ocean uptake flux, while the tracers L2003₂ and Li2000₃ apply factors of 2 and 3, respectively, to the Li et al. (2003) flux in order to increase the HCN ocean uptake.

The tracers including the scaled ocean uptake schemes show a substantial improvement in the agreement with the NDACC measurements. In particular, the best agreement, considering the RMSE and R² values, is obtained by reducing the HCN flux in the Li et al. (2000) scheme by three-quarters or by doubling the Li et al. (2003) flux in the L2003₂ tracer.

Figure 4 shows a comparison between the measured HCN total column time series and the two best-performing tracers, L2000_0.25 and L2003₂, characterized by the lowest RMSE and a large R², and the worst one, L2003. HCN total columns from the L2003 model tracer are larger than the HCN measured by FTIR instruments at all the locations; it is clearly visible especially for Jungfraujoch, where the HCN estimated by L2003 is more than twice that the measured one. Both L2000_0.25 and L2003₂, the best-performing model tracers, agree very well with the measured values, confirming that both ocean uptake schemes published by Li et al. (2000, 2003) need to be scaled to be more accurate in reproducing HCN variability.

Figure 4HCN total column time series (molec. cm⁻²) measured at four NDACC stations (Thule (Thu), Bremen (Bre), Jungfraujoch (Jfj), and Mauna Loa (Mlo); black lines) and modelled by the L2000_0.25 (blue dots), L2003₂ (green dots), and L2003 (red dots) tracers (with averaging kernels applied; ak). Note the different time periods at the different stations.

Download

3.3 Atmospheric lifetime and global budgets

The global budgets of atmospheric HCN for the model tracers L2000, L2003, and L2000_0.25 are shown in Table 4. Here we report only the two tracers using the original ocean uptake schemes from Li et al. (2000, 2003), which have the worst performance, and the best one using the ocean uptake from Li et al. (2000) scaled by 0.25. The total atmospheric burdens of HCN are 0.33, 0.89, and 0.55 Tg N, respectively, reflecting the HCN concentration underestimation of the L2000 tracer and the overestimation of the L2003 tracer, while the burden of L2000_0.25 is close to the values reported in Li et al. (2000, 2003) and Singh et al. (2003). Each HCN tracer uses the same emission scheme, with biomass burning as the main contribution, producing 2.42 Tg N yr⁻¹. Ocean uptake provides the main sink for all the tracers, 2.36 Tg N yr⁻¹ in L2000, 2.32 Tg N yr⁻¹ in L2003, and 2.38 Tg N yr⁻¹ in L2000_0.25, which all largely balance the emissions despite the variation in the first-order rate of uptake (i.e. the change in atmospheric HCN burden compensates). The sink from a reaction with OH is 0.05, 0.12, and 0.12 Tg N yr⁻¹ for L2000, L2003, and L2000_0.25, respectively. Despite its importance for determining the stratosphere loss of HCN, the reaction with O(¹D) is a relatively very small sink globally, 7 × 10⁻⁴ for the L2000 tracer, 1.9 × 10⁻³ for L2003, and 1.9 × 10⁻³ Tg N yr⁻¹ for L2000_0.25. The resulting tropospheric lifetimes are 1.6, 4.4, and 2.6 months for the three tracers reported in Table 4.

The budgets of the L2000_0.25 tracer, which best reproduces the HCN variability, are in good agreement with the HCN budgets of Li et al. (2000), while the ocean uptake loss and the emissions are substantially larger than the Li et al. (2003) and the Singh et al. (2003) results. Li et al. (2003) estimate a global HCN loss to the ocean of 0.73 Tg N yr⁻¹, and Singh et al. (2003) estimate one of 1.0 Tg N yr⁻¹; both of these estimations are less than a half of the ocean uptake calculated in the present study. Similarly our emissions are more than twice the total emissions from Li et al. (2003), 0.63 Tg N yr⁻¹ from biomass burning and 0.2 Tg N yr⁻¹ from residential coal burning, and Singh et al. (2003), 1.1 Tg N yr⁻¹. The resulting global mean atmospheric lifetime is also in agreement with the range of 2.1–4.4 months calculated by Li et al. (2000).

Table 4Global burden, budget terms, and atmospheric lifetimes for three model HCN tracers.

Download Print Version | Download XLSX