Chemical evidence of inter-hemispheric air mass intrusion into the Northern Hemisphere mid-latitudes

The East Asian Summer Monsoon driven by temperature and moisture gradients between the Asian continent and the Pacific Ocean, leads to approximately 50% of the annual rainfall in the region across 20–40°N. Due to its increasing scientific and social importance, there have been several previous studies on identification of moisture sources for summer monsoon rainfall over East Asia mainly using Lagrangian or Eulerian atmospheric water vapor models. The major source regions for EASM previously proposed include the North Indian Ocean, South China Sea and North western Pacific. Based on high-precision and high-frequency 6-year measurement records of hydrofluorocarbons (HFCs), here we report a direct evidence of rapid intrusion of warm and moist tropical air mass from the Southern Hemisphere (SH) reaching within a couple of days up to 33°N into East Asia. We further suggest that the combination of direct chemical tracer record and a back-trajectory model with physical meteorological variables helps pave the way to identify moisture sources for monsoon rainfall. A case study for Gosan station (33.25°N, 126.19°E) indicates that the meridional transport of precipitable water from the SH accompanying the southerly/southwesterly flow contributes most significantly to its summer rainfall.


Gosan station:
The Gosan station (GSN, 33.25°N, 126.19°E, Jeju Island, Korea) is located on the boundary between the Pacific Ocean and the Asian continent, and experiences distinct seasonal wind and weather patterns characterized by warm wet East Asian Summer Monsoon (EASM) and cold dry winter (Fig. S1), and thus provides an ideal place to monitor both Asian continental outflows and maritime air mass intrusion. Figure S1. (a) The Gosan AGAGE (Advanced Global Atmospheric Gases Experiment) station is located atop a 72-m cliff on the remote south-western tip of Jeju Island, 100 km south of the Korean peninsula, allowing for monitoring of long-range transport from the surrounding region. (b) The residence time analysis using 5-day back-trajectories arriving at the Gosan station for year 2008 made by HYSPLIT 4.8 model shows that major air masses dominating the region vary seasonally: predominant northwesterly and northeasterly continental outflows from fall through spring versus air flows of clean air directly from the Pacific in summer and from northern Siberia in winter. Maps and plots are generated with MATLAB R2013a.

Selection of chemical tracers:
Criteria for chemical tracers to capture the imprint of meridional transport are: the chemical species should be (1) abundant in the atmosphere to assure a high signal-to-noise ratio; (2) have its dominant emission sources located in the NH; (3) have a chemical lifetime between one to ten years, and thus show clear latitudinal and/or NH-SH gradients in its atmospheric mixing ratio. Among synthetic compounds, therefore mainly emitted/produced in the NH, hydrofluorocarbons (HFCs) have lifetimes similar to or longer than the time scale of inter-hemispheric exchange. In particular, HFC-134a with a lifetime of 14 years 1 and HFC-152a with a lifetime of 1.6 years 1 have been most widely used among HFCs since their introduction in the 1990s. Their major uses are as refrigerants, aerosol propellants and foam-blowing agents 1 , replacing use of chlorofluorocarbons (CFCs) and hydrochlorofluorocarbons (HCFCs) in these applications.

Influences of OH seasonality and vertical dilution on the HFCs drawdowns:
We found the same drawdown features in the time series of CF4 and SF6 that have much longer atmospheric lifetime than HFCs, and thus are not affected by seasonal variations of the OH radical concentrations. Thus, their drawdowns cannot be the results of OH seasonality (Fig. S3).
Vertical dilution by free tropospheric air masses is also an unlikely explanation for the drawdowns given the fact that we cannot detect a clear vertical gradient up to the altitude of 8 km in HFC-152a mixing ratio observed from the fourth and fifth HIAPPER Pole-to-Pole Observations (HIPPO-4 and HIPPO-5) (http://hippo.ornl.gov/dataaccess) during the boreal summer (Fig. S4).

Dropdowns of HFCs in late April:
The climatological summary 2 of the regional onset dates of Asian monsoon shows that the earliest onset of the monsoon arrives first at the central Indochina Peninsula Fin late April and early May, and then the monsoon system advances northward along the BOB (Bay of Bengal) and the south China sea, finally arriving in the East Asian region in June. The dropdowns of the HFC mixing ratio observed at Gosan in late April coincided with the ISM onset dates defined by zonal wind shear index 3 (Fig.S5). Given the fact that the ISM is initiated by development of an active BOB cyclone 2 , the HFC dropdowns in spring seem to be a fingerprint of coherent propagation of the accelerated low-level westerly and significant convection activity over the BOB, thereby indicating a tele-connected perturbation of the East Asian air masses by the mechanism(s) controlling the ISM and/or the subsequent convective motion.

Trajectory calculations:
To further examine the relationship between HFC variations observed during the EASM periods and air-mass transport pathways and origins, we analyzed 5-day back trajectories of air masses arriving at Gosan in June, July, and August (JJA) from 2008 to 2013. The back trajectories were calculated using the HYSPLIT 4.8 modeling 4 with meteorological output fields from the National Center for Environmental Prediction (NCEP) Global Data Assimilation System (GDAS; http://www.ready.noaa.gov/archives.php). These data are provided on a 1° latitude-longitude grid and every 3 hour. Statistics of air clusters were calculated for the 4078 back trajectories.
When we examined the HYSPLIT model runs in comparison with a particle dispersion back trajectory model (i.e., FLEXPART), we found the results at 500-m altitude consistent between the two models. So we used the single particle trajectory model since it has been proven in both computational readiness and actual use for the cluster analysis 5 . Then, the resulting three types of air clusters were presented by using the FLEXPART software in Fig. 2(a), because it illustrates the air distribution of each cluster in a more realistic way.

Sensitivity of HFCs tracer:
The HFCs measurements were associated with a type of monsoonal air flow (Figs. 2(a), (b) and Fig. S6). Tropical air masses (type A in Fig. 2(a)) are typically moving very fast via the low-level southerly flow along the western boundary of the western North Pacific (WNP) subtropical high, passing through the South China Sea. As expected, most of the HFCs measurements corresponding to tropical air masses (type A: Fig.S6(c)) are lower than those measured at Mauna Loa (MLO) and closer to SH background values while some measurements, especially in July, show higher values than those for most of type A air masses. We found that those measurements correspond to the air masses that originated from the tropical SH, but curved over the east coast of China ( Fig. S6(b)), where intensive industrial activity for HFCs production and consumption occurs, and therefore, reflect the influence of regional pollution plume on the low-level tropical air masses. This demonstrates the HFCs tracer signals are sensitive not only to air mass origins, but also to pathways.  Fig. 2(b). (b) The HFC-152a data associated with tropical air masses group (type A) passing over the east coast of China are closer to MLO's values, but (c) most of data corresponding to tropical air masses passing through the South China Sea are lower than MLO's level and closer to SH background values. Maps and trajectories are generated with MATLAB R2013a.

s/kg
Relative contributions of the three air mass types to the EASM rainfall: Precipitation amount for each air mass type was estimated by integrating the precipitation amounts observed for all air masses with trajectories categorized into the corresponding type arrived at Gosan during 2008 to 2013. The relative contribution of each air mass type was then derived by considering total EASM rainfall during JJA.