Unexpected high contribution of in-cloud wet scavenging to nitrogen deposition induced by pumping effect of typhoon landfall in China

NAQPMS is a 3D Eulerian terrain-following air quality model, which is developed by Wang et al (2001). It has been widely used to simulate the air pollutants, such as ozone (Li et al 2011, 2016), dust (Tian et al 2020), anthropogenic aerosols (Chen et al 2017, Wang et al 2017, Yang et al 2018, Du et al 2019), and even toxic pollutants like Mercury (Chen et al 2015), as well as the sulfur and nitrogen wet depositions over China (Ge et al 2014). Recently, NAQPMS has been invited to participate in the Model Inter-Comparison Study for Asia III (MICS-Asia III) to compare the simulated air pollutants (Li et al 2019, Kong et al 2020) as well as the depositions (Ge et al 2020, Itahashi et al 2020) over East Asia among 14 regional and global air quality models. In this study, the cloud processes schemes for in-cloud scavenging (Dennis et al 1993, Ge et al 2014) and the below-cloud scavenging scheme (Xu et al 2017, 2019, 2020) have been updated in NAQPMS to better simulate the wet deposition of air pollutants. In order to investigate the S/R of wet deposition of IN, an online tracer tagging module is also implemented in NAQPMS to track the formation and evolution of air pollutants (Li et al 2008, Wu et al 2017). Unlike other studies that can only track the source of the total wet deposition (Lin et al 2008, Kajino et al 2013), NAQPMS coupled with the on-line tracer tagging module can quantitatively estimate the S/R of the below- and in-cloud wet deposition without introducing errors by nonlinear chemistry. Further technical details about the NAQPMS can be found in Text S1-S2 and in Li et al, (2011;2013;2016) and Ge et al (2014).

The model domain covers the East Asia (15.4°S–58.3°N, 48.5°–160.2°E), which is set on a Lambert conformal map projection with 180 × 170 grids at 45 km horizontal resolution. Vertically, the numerical model has 20 terrain-following layers from the surface to the height of 20 km with the lowest 10 layers below 3 km. The global atmospheric chemical transport model MOZART v2.4 provided initial and lateral boundary conditions for the NAQPMS. The meteorological fields for NAQPMS were provided by the Weather Research and Forecasting (WRF) model, version 3.4.1 (Skamarock et al 2008) driven by National Centers for Environmental Prediction Final Analysis reanalysis data (1° × 1°). More detailed descriptions of the WRF model including the Physics schemes can refer to in Text S3 and in Ge et al (2021). Emissions for NAQPMS are obtained from the emissions inventory in MICS-Asia III (Li et al 2017), and considering the yearly projections of the variations through the satellite-constrained NO_x and ammonia (NH₃) emissions based on vertical column density (VCD) values in 2020. More detailed descriptions of the updated emission can be found in Text S4. The model's output of IN wet deposition can be classified as oxidized N (N_ox), including particulate nitrate, gaseous nitrate acid and NO_x, and reduced N (N_rd) including NH₃ and particulate ammonium. The simulation period was from 24 July to 6 August 2020, with 8 days of model spin-up time from 24 to 31 July 2020.

Figure S1 shows 22 identified areas (ID) in the NAQPMS model simulation domain. Fifteen land areas in the territory of China (i.e., YRD, FJ, SD, NCP, SX, HN, AH, HB, JX, SW, TW, West, Middle, NW, and NE) are marked in different colors, together with North Korea (NK), South Korea (SK) and Japan. The three major oceanic regions along the pathway of typhoon Hagupit (green line depicted in figure S1) are marked as ECS (East China Sea), WP (West Pacific), and JS (Japan Sea), respectively. The definitions of the 22 IDs are listed in table S1.

Ground-based observations of wet deposition collected at sites shown in figures 1(a)–(c) during 1–6 August 2020 by the China National Environmental Monitoring Centre (CNEMC) (http://www.cnemc.cn) and Acid Deposition Monitoring Network in East Asia (EANET) ) (http://www.eanet.asia/, last access date: 25 June 2022) were used to validate the performance of the model with respect to the wet deposition of N_ox and N_rd. Six of 48 stations are in SW, 9 in HN, 8 in SD, 11 in YRD, 4 in West, 4 in FJ, 2 in Middle, 1 in WP, 2 in Japan and 1 in SX (table S2). The CNEMC observations of wet deposition included daily average inorganic ion concentrations (SO₄ ²⁻, NO₃ ⁻, NH₄ ⁺, F⁻, Cl⁻, Ca²⁺, Mg²⁺, Na⁺, and K⁺) and daily accumulated precipitation from 1 to 6 August 2020. Because CNEMC sites mainly covered inland China, EANET sites located in coastal area over East Asia is additionally analyzed in this study (More information about the EANET observation can refer to Ge et al (2020)). After quality control (Text S5), the measured anion and cation concentrations of precipitation samples from 48 stations were nearly in balance and there were no large deficiencies of major inorganic ions. Surface observations of hourly PM_2.5 and NO₂ concentrations at the 48 stations from 1–6 August 2020 were also used in this study. Ground meteorological observations of daily precipitation and wind were obtained from China Meteorological Administration. The best-track data of typhoon Hagupit provided by the Japan Meteorological Agency were used to identify the location of the typhoon (https://www.jma.go.jp/jma/jma-eng/jma-center/rsmc-hp-pub-e.g./besttrack.html, last access: 16 December 2021).

Zoom In Zoom Out Reset image size

The NAQPMS was able to simulate the precipitation and wet deposition of IN both in typhoon-affected regions and in the other regions of China. The simulated accumulated precipitation in the typhoon-affected region during the pre-landfall (1–2 August), landfall (3–4 August), and post-landfall (5–6 August) stages shows good correlation, with a correlation coefficient (R) of 0.79 (N = 11), 0.78 (N = 110) and 0.75 (N = 92), and normalized mean bias (NMB) of −19%, −12.6% and 12.7%, respectively (figures 1(d)–(f)). In the S-high control region, the R values are 0.66 (N = 48), 0.64 (N = 41) and 0.68 (N = 45) for the pre-landfall, landfall and post-landfall stages, respectively, and the NMB values are −19.5%, 17.6% and −9.1% (figures S2(a), (b)). Compared to the reported statistical scores in MICS-Asia III (Ge et al 2020, Itahashi et al 2020), our modeling performance in this study have been confirmed as better. In accordance with the good model performance for precipitation, the simulated wet deposition of IN agrees well with the observations, with the NMB ranging from −25% to +2% (figures 1(d)–(i)). At stations in the typhoon-affected region especially YRD, the simulated wet deposition values of N_ox and N_rd are very close to their observed counterparts, with NMBs of 4% and 8%, respectively. At stations in the S-high control region, the wet deposition of N_ox was overestimated by the model, with an NMB of +32%. Such positive bias in the simulated IN wet deposition is mainly due to the overestimation of N precursors, up to +20% for NO_x as reported by Ge et al (2020).

As the track of Hagupit in figure 1(b) (blue lines) shows, the typhoon impacted eastern China after it made landfall in the YRD on 3 August. According to the MODIS true-color image taken on 3 August (figure S3(a)), the location of mostly the cloud-free typhoon eye overlaps with the typhoon position at 00:00 UTC on 3 August. Then, the typhoon continued to move in a northeasterly direction after it entered the Bohai Sea. Finally, it made landfall in SK. The model performance of typhoon track simulation is important to reproduce the structure of typhoon surface precipitation and cloud hydrometeors (Merrill 1984, Irish et al 2008, Chavas and Emanuel 2010, Hong et al 2020, Kumar et al 2020). Besides the well-simulated precipitation by WRF, the typhoon track was also consistent with that obtained from Japan Meteorological Agency (Text S3 and figure S4). The whole typhoon track can be divided into three stages, as figures 1(a)–(c) show, i.e., the pre-landfall stage (1–2 August), the landfall stage (3–4 August), and the post-landfall stage (5–6 August).

Figure S5 shows latitude–height cross-sections of the ratio of the PM_2.5 concentration to that at ground level with the vertical wind vector superimposed. More than 30%–49% of the PM_2.5 concentration at ground level could be lifted to altitudes higher than 2–3 km in less than 10 h. All cross-sections over the landfalling regions of the typhoon simulated by NAQPMS captured these fast and violent uplifts of PM_2.5. The gaseous precursor of nitrate and ammonium, NO_x, and NH₃ in the same cross-sections, were also found to be uplifted to high altitudes within a very short time (figures 2 and 3). In particular, NO_x was strongly lifted to as high as about 4 km altitudes (figure 2(c)) due to the strong upward winds in the typhoon, while the lifting of NH₃ was significantly weaker than that of NO_x and PM_2.5. This is reasonable that NH₃ is mainly emitted by agricultural sources (e.g., livestock and synthetic fertilizers) located on the surface (Aneja et al 2003), whereas NO_x pollution occurs primarily through emissions from the industry at higher altitudes, for example, stack chimney emission (Aneja et al 2001). The higher altitude the emission source is, the easier for the pollutants to be uplifted to a higher altitude and hence the more contribution to long-range transport.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Vertical transport of air pollutants can be caused by different mechanisms. Boundary layer turbulent mixing can contribute to the vertical transport of pollutants, but it is limited by the boundary layer height (0.6–0.8 km). The cold-front structure of most cyclones can also lift pollutants to 1–2.5 km above the ground (Kang et al 2019). However, only very strong cyclones such as typhoons may lift air pollutants to almost double the high altitudes that most cyclones can. This indicates that pollutants in the surface layer, such as PM_2.5 and NO_x, were quickly lifted to altitudes higher than 2–3 km when typhoon Hagupit made landfall. This elevation of pollutants influenced by the typhoon is hereafter referred to as the typhoon's 'pumping effect' (figure 4).

Zoom In Zoom Out Reset image size

In order to illustrate the influence of the 'pumping effect', wet deposition of IN from in-cloud and below-cloud fractions was recorded by NAQPMS. The typhoon-affected regions in YRD as well as the precipitation region in Shanxi province affected by subtropical anticyclone (hereafter called S-high) are selected to compare the different mechanisms of IN wet deposition. The definition of the S-high region is followed by He et al (2015), who use the 5880m geopotential height at 500 hPa to relate with the S-high regions as shown in figure S6, and hence was displayed as the blue rectangle in figure 1.

Box plots of the below- and in-cloud wet deposition and their contributions to the total wet deposition of N_ox and N_rd during the pre-landfall, landfall, and post-landfall stages in the YRD and S-high region are displayed in figure 5. In YRD, with the average precipitation increasing from 5.6 to 21.6 mm, the in-cloud N_ox wet deposition contribution changed from 75% during pre-landfall to 89% during landfall. In the S-high region, however, in-cloud N_ox wet deposition contribution increased only slightly from 63% to 67%, with the average precipitation increasing from 14 to 27.6 mm. A similar phenomenon is found for N_rd wet deposition. Xu et al (2017) pointed out that the in-cloud contributions to the total wet deposition will increase with the increase in rainfall as the effect of the washout process weakens. However, the increase of in-cloud contributions is limited by the concentrations of pollutants within the clouds. The larger increase in the typhoon-affected region than in the S-high region indicates the significance of the typhoon's 'pumping effect', which elevates the surface air pollutants to high altitudes and contributes more to the in-cloud scavenging during precipitation events. It is noteworthy that the upper limit of the in-cloud contribution of N_ox (N_rd) reached 89% (80%) during the typhoon, which is significantly higher than the average value of 39% (53%) in 2014 observed in China (Xu et al 2017) and the value of 34% observed in Japan (Aikawa and Hiraki 2009, Aikawa et al 2014). The highest value (82%) was found in the cold vortex events during the four years observation in Beijing, China reported by Ge et al (2021). This observation confirms that in-cloud contributions for strong convective weather systems like typhoons and cold vortexes can be much higher than thought (figure 6, figure S7).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To better characterize the mechanism of nitrogen deposition induced by typhoons, detailed simulations were done using the updated NAQMPS. Different from the previous studies, the source of wet deposition during the typhoon event was firstly identified in relation to wet scavenging processes (figure 7). Three typical typhoon-affected regions, i.e., YRD, Bohai, and SK were chosen as the receptor regions under the influence of typhoon Hagupit to quantitatively investigate the source of in- and below-cloud wet deposition.

Zoom In Zoom Out Reset image size

For the YRD, the largest source contribution to the in-cloud wet deposition in the YRD was from local emission (table S3), accounting for 52% (38%) of the in-cloud wet deposition of N_ox (N_rd). The relative contribution to the below-cloud from YRD was 68% (48%), corresponding up to 8% (9%) of total wet depositions of N_ox (N_rd). This reveals that local emissions were the dominant source of both in-cloud and below-cloud wet deposition in the YRD. For Bohai, the largest contribution was from emissions in the YRD, reaching up to 47% of the total wet deposition of N_ox, which was about four times higher than other sources (table S3). However, pollutants from the YRD contributed significantly less to the in-cloud wet deposition of N_rd than that of N_ox (1.5 versus 7.2 × 10⁻³ mg m⁻² h⁻¹, table S3). For SK, the largest in-cloud contributor was also from the YRD. However, the largest contributor to the total wet deposition of N_ox in SK was the local emission, which accounted for 32% of the total wet deposition, with 5% from the below-cloud part. Detailed analysis of the S/R relationship in three regions can be found in Text S6 and table S3.

Analysis of IN deposition caused by typhoons in terms of scavenging pathways can shed light on typhoons' effect on the coastal region, optimize the source identification and help to protect the marine ecological environment. This updated S/R relationship of in-cloud wet deposition during the typhoon event challenges the traditional view that the in-cloud wet deposition of acidic pollutants is mainly linked with medium- or long-range transport, which is mainly in non-typhoon events. This significantly higher contribution from the YRD region can be attributed to the 'pumping effect' of the typhoon, which influenced greatly not only the long-range transport of IN from the YRD to Bohai and SK but also the local wet deposition in the YRD itself.

What will the mechanism so-called 'pumping effect' of typhoon affect the wet deposition of nitrogen in the downwind typhoon-affected regions. We have quantified the effect of the 'pumping effect' according to the S/R relationship of N_ox wet deposition as figure 4 shows. As section 3.2 described, when typhoon HAGUPIT made landfall in YRD, the PM_2.5 and NO_x in the multiple cross-sections, were found to be uplifted to higher than 2–3 km in less than 10 h due to deep convective transports associated with cyclones (Fadnavis et al 2011). These pollutants were scavenged by the clouds, making the in-cloud wet deposition significant increase in this study.

According to table S3, when typhoon made landfall in YRD, 106 × 10⁻³ mg m⁻² amount of oxidized N were uplifted into the cloud by the pumping effect of typhoon. Meanwhile, with the development of rainfall in YRD, 37 × 10⁻³ mg m⁻² of uplifted N from YRD were scavenged by in-cloud process, which accounts for 52% of the total wet deposition in YRD (a). With the movement of typhoon, the remained amount of 69 × 10⁻³ mg m⁻² oxidized N pollutants (g) induced by pumping effect were transported into downwind typhoon-affected regions such as Bohai and South Korea. This made 7.2 × 10⁻³ mg m⁻² in-cloud wet deposition of N_ox (c) in Bohai and 42 × 10⁻³ mg m⁻² in-cloud wet deposition of N_ox (e) in South Korea. Different from the in-cloud scavenging, the below-cloud scavenging was much smaller. Almost 6 × 10⁻³ mg m⁻² amounts of oxidized N from local region were washed out and deposited into YRD (b). Due to the airflow supported by typhoon, 5 × 10⁻³ mg m⁻² amounts of oxidized N from YRD (h) were transported below the cloud to downwind areas such as Bohai (0.4 × 10⁻³ mg m⁻²) and South Korea (<1 × 10⁻³ mg m⁻²).

The contributions of long-range transport from the YRD were almost maintained above 40% in Bohai and 30% in SK mainly via the in-cloud process. This means that, as the typhoon moved to downwind areas, the long-range transport of pollutants from the YRD constantly influenced the in-cloud wet deposition in Bohai and SK. The largest source contribution to the in-cloud wet deposition in the YRD was local, which contributed to 52% of the in-cloud wet deposition of N_ox. This reveals that local emissions were the dominant source of the in-cloud wet deposition in the YRD. This updated S/R relationship of in-cloud wet deposition during this typhoon event challenges the traditional view that, the in-cloud wet deposition of acidic pollutants is mainly linked with medium- or long-range transport. The above analysis reveals the importance of contributions from emissions in typhoon landing to the typhoon-affected regions. This mechanism of wet deposition due to the 'pumping effect' may have profound effects on the wet deposition of N in typhoon-affected areas.