Dry Deposition of PM2.5 Nitrate in a Forest according to Vertical Profile Measurements

The atmospheric nitrogen compounds can serve as a nutrient; however, its excess deposition has harmful effects on terrestrial ecosystems due to acidification and eutrophication. There are still large uncertainties concerning the dry deposition process of PM2.5 nitrate in forests, even though this process affects the accuracy of chemical transport model simulations. To better understand this process, we conducted vertical profile measurements of inorganic ions in PM2.5 and SO2 above and within a forest canopy in the Field Museum Tamakyuryo site in suburban Tokyo with a particular focus on the processes observed under both daytime and nighttime and both leafy and leafless conditions. We performed two observations during leafy periods (July 21–August 1, 2015, and September 27–October 11, 2016) and one observation during a leafless period (February 23–29, 2016). To obtain daytime and nighttime vertical profiles, we set filter holders at 4 or 5 heights on an observation tower in the forest and changed the filters for each daytime and nighttime. For the PM2.5, the vertical gradients of NO3− concentration were larger than those of SO42− during both the daytime and nighttime for all observational periods, particularly during the leafy periods. In addition, the decreasing rate of NO3− in the PM2.5 within the canopy was larger than that of SO2 for some observational periods. In the daytime, the air temperature was higher near the canopy surface during the leafy period and near the ground surface during the leafless period. As also suggested by past studies, the large gradients of NO3− in the PM2.5 during the leafy period were likely caused by the volatilization of NH4NO3 near the deposition surfaces due to the higher temperature in the daytime and the lower concentration of HNO3 caused by its fast removal during both the daytime and nighttime.


INTRODUCTION
A drastic increase in the emission of nitrogen compounds on a global scale has occurred due to human activities over the last century (Galloway et al., 2008).In particular, a drastic increase in the emission of nitrogen oxides (NO x ) associated with energy consumption has been observed over East Asia in recent decades (Kurokawa et al., 2013;Ohara et al., 2007).NO x acts as a precursor of nitrate atmospheric particulate matter with diameters of less than 2.5 μm (PM 2.5 ); such particu-late matter is known to have adverse effects on human health.Moreover, these nitrogen compounds have harmful effects on terrestrial ecosystems due to acidification and eutrophication via the deposition of excess nitrogen.
PM 2.5 simulations have been conducted using chemical transport models over Japan (Shimadera et al., 2018;Morino et al., 2015;Shimadera et al., 2014); however, such models have clearly overestimated the concentration of NO 3 -in PM 2.5 .Shimadera et al. (2014) suggested that the simulated concentration was highly dependent on the uncertainty in the dry deposition process of nitrogen compounds.Moreover, for NO 3 -in PM 2.5 , large uncertainties exist in the theoretical models used to estimate the dry deposition rates, particularly on forest surfaces (Flechard et al., 2011).However, the dry deposition process of PM 2.5 has rarely been studied in East Asia, where anthropogenic emissions of NO x are higher than those found in Europe and North America.Therefore, a better understanding of the dry deposition process of NO 3 -in PM 2.5 in this region will contribute to improving the model accuracy of estimated PM 2.5 concentrations and nitrogen deposition rates.
Several field experiments to determine the deposition velocity (V d ) of PM 2.5 nitrate in forests via gradient or relaxed eddy accumulation methods have been performed in Japan (Sakamoto et al., 2018;Honjo et al., 2016;Takahashi and Wakamatsu, 2004).These experiments suggest that the equilibrium shift of NH 4 NO 3 into the gas phase likely enhances the dry deposition of nitrogen compounds, as indicated by previous studies in other regions (Nemitz et al., 2004;Wyers and Duyzer, 1997;Sievering et al., 1994;Huebert et al., 1988).Measurements of the vertical profiles of relevant matter are useful to understand dry deposition mechanisms in forests.Yamazaki et al. (2015) found different vertical profiles for NO 3 -and SO 4 2-in PM 2.5 in a forest in Tokyo over the course of a year, likely due to the abovementioned volatilization process.However, these studies conducted in Japan primarily focused on leafy forests and did not examine the diurnal deposition process.To improve the understanding of dry deposition processes for PM 2.5 in East Asia further, we conducted intensive field observations in a forest in suburban Tokyo.We obtained the vertical profiles of the PM 2.5 components together with those of SO 2 , a stereotypical gas, the deposition processes of which have been generalized (Nemitz, 2015), to compare the differences in the deposition processes of particle and gaseous matter.We particularly focused on the daytime and nighttime deposition processes of PM 2.5 in a forest during leafy and leafless periods.

OBSERVATIONS AND METHODOLOGY
We conducted the measurements using an observation tower (Fig. 1) in a forest at the Field Museum Tamakyuryo (FM Tama) site of the Tokyo University of Agriculture and Technology, which is located in a western suburb of Tokyo, Japan (35°38 N, 139°23 E).Deciduous trees (Quercus) were the dominant tree species around the tower in addition to some Japanese cedar (Crytomeria).The canopy height around the tower was approximately 20 m.The deciduous trees were leafy from April and leafless from December.Other details concerning the measurement site have been described in Matsuda et al. (2015).We performed two observations during leafy periods and one during a leafless period from July 2015 to October 2016.We sampled PM 2.5 and SO 2 simultaneously using a filter pack (Tokyo Dylec Corporation, NILU filter folder NL-O) with an impactor and a pump unit (Tokyo Dylec Corporation, MCI sampler).The flow rate was set to 20 L min -1 in accordance with the PM 2.5 cut off the impactor.PM 2.5 was collected on glass fiber filters coated with Teflon.SO 2 was collected on a cellulose filter impregnated with potassium carbonate following the PM 2.5 filter.To obtain vertical concentration profiles during the daytime and nighttime, we set the filter holders at four or five heights, as indicated in Table 1, on the tower and changed the filters twice a day.Outlines of the samplings during each observational period are given in Table 1.To sufficiently detect the vertical gradients, we set the sampling time to more than 9 h after considering previous measurements at the same site (Yamazaki et al., 2015).Sampling was conducted continuously, except when it was raining.We obtained 36 valid samples in total at each height level.After the samples were collected, the inorganic ions in each filter were extracted into deionized water via ultrasonic extraction and then analyzed using ion chromatography (Thermo Scientific, Dionex ICS-1100).
Meteorological conditions including the wind speed (WS), temperature (Temp), and relative humidity (RH) were observed at heights of 30 m, 25 m, 20 m, 10 m, 6 m, and 1 m along the tower using meteorological sensors (YOUNG 81000 and PREDE PHMP45A at 30 m; VAISALA WXT520 at 25 m, 20 m, 10 m, 6 m, and 1 m).The 10-min averages were used to obtain the vertical profiles of each parameter.The leaf area index (LAI) was measured using a plant canopy analyzer (LI-COR LAI-2200).

1. Overview of the Three Observational Periods
Table 2 shows the atmospheric conditions at the observation site for the three periods considered here.The estimated LAI shows a leafy canopy in 15-summer and 16-autumn and a leafless canopy in 16-winter, where The temporal variations in the mass concentrations of SO 4 2-and NO 3 -in the PM 2.5 at heights of 30 m, 23 m, 8 m, and 1 m during the observational periods are shown in Fig. 2.During the three observational periods, there were no significant differences in the mass concentrations of SO 4 2-between these heights except between 8 m and 1 m (p<0.01),but there were significant differences in the mass concentrations of NO 3 -(p<0.01).The NO 3 -concentration clearly decreased from the top of the canopy to the forest floor during the observations compared to the SO 4 2-concentration.

2 Vertical Profiles and Decreasing Rates
The vertical concentration profiles of SO 4 2-and NO 3 in the PM 2.5 and those of SO 2 during the daytime and nighttime for the observational periods are shown in Fig. 3.The daytime and nighttime averages at each height were used to obtain the vertical profiles.The concentrations at all heights were normalized by those at 30 m.In general, the dry deposition mechanisms of aerosols are thought to depend on the physical processes at each particle size.In that case, the vertical profiles of SO 4 2-and NO 3 -in the PM 2.5 should show similar tendencies.However, the decreasing trend of NO 3 -below the canopy was clearly larger than that of SO 4 2-during both the daytime and nighttime for all observational periods, especially during the leafy seasons (15-summer and 16-autumn).In the leafless season (16-winter), the differences in the SO 4 2-and NO 3 -vertical profiles decreased due to the smaller decrease in NO 3 -.The decreasing rate (hereafter referred to as DR) is a proper index to understand the tendency of whether the target component is removed or not between each of the heights.The DR is defined as the following: where C Z1 and C Z0 is the concentration of the target component at Z1 [m] and Z0 [m] ( Z1 > Z0 ), respectively.Therefore, differences in DR indicate the differences in the removal efficiency between components.Because SO 2 is a gas that is easy to deposit on forest surfaces due to its reactive and water-soluble properties, its value of DR is assumed larger than that of fine particulate matter (e.g., Erisman and Draaijers, 1995).This assumption holds between SO 2 and SO 4 2-for all observations in the canopy.However, the DR value of NO 3 -below the canopy was larger than that of SO 2 during some observational periods (daytime in 15-summer and 16-autumn) (Fig. 3).Therefore, there are likely other factors that enhance the deposition of NO 3 -in PM 2.5 in addition to physical processes.
The DR 30-1 values (the DR values between 30 m and 1 m) of SO 4 2-and NO 3 -for the observational periods are shown in Fig. 4.Each circle shows a daytime or nighttime daily value.The DR 30-1 values of NO 3 -were clearly larger than those of SO 4 2-, particularly in 15-summer and 16-autumn.The DR 30-1 values of SO 4 2-varied independently of LAI for all observations.Conversely, the DR 30-1 values of NO 3 -were clearly larger during the leafy periods and smaller during the leafless period.These results indicate that the variation in the NO 3 -decrease may be closely related to the leaf condition.In effect, the removal efficiency of NO 3 -by the leaf canopy was larger than that of SO 4 2-regardless of the time of day.

3 Effect of the Equilibrium Shift of NH 4 NO 3 on
the Deposition Process (NH 4 ) 2 SO 4 particles have a very low vapor pressure and exist as an aerosol under atmospheric conditions.Conversely, NH 4 NO 3 particles are semi-volatile and have an equilibrium relationship with NH 3 and HNO 3 in the atmosphere.The dry deposition of NH 4 NO 3 is affected by its volatilization process and depends on Temp, RH, and the concentrations of either HNO 3 or NH 3 .Therefore, differences in the DR 30-1 values between SO 4 2-and NO 3 -were likely caused by differences in their chemical properties.
There are some previous studies mentioned below that indicate a higher V d value for NO 3 -particles compared to SO 4 2-particles.For a crested wheatgrass field in the Boul-der Atmospheric Observatory, a research facility in Colorado in the United States, Huebert et al. (1988) found that the vertical gradients of the NO 3 -concentrations in particles were larger than those of SO 4 2-and some times exceeded those of HNO 3 , which has a very high V d value.
These results are consistent with the predictions of a model coupling the volatilization of NH 4 NO 3 to the rapid dry deposition of HNO 3 presented in Brost et al. (1988).Wyers and Duyzer (1993) determined the V d values of SO 4 2-and NO 3 -above a coniferous forest in    Considering these studies, there are two possible sources of the differences in the dry deposition mechanisms between (NH 4 ) 2 SO 4 and NH 4 NO 3 : (1) An equilibrium shift of NH 4 NO 3 due to the higher temperature near the deposition surfaces and/or (2) An equilibrium shift of NH 4 NO 3 due to the low concentrations of HNO 3 caused by fast removal near the deposition surfaces.
The variations in the ensemble mean air temperature over the observational periods are shown in Fig. 5.In the daytime, Temp at 30 m was lower than Temp at 20 m, which was close to the canopy surface during 15-sum-   mer and 16-autumn, and was lower than Temp at 1 m, which was close to the ground surface during 16-winter.This occurred because direct sunlight struck the canopy surface during the daytime in the leafy periods (15-summer and 16-autumn) while it struck the ground surface in the leafless period (16-winter).In addition, Temp values of forest surfaces exposed to sunlight tend to be higher than the Temp values near the surfaces (Nakahara et al., 2019).Therefore, the volatilization of NH 4 NO 3 was likely enhanced close to the canopy surface during the leafy periods and close to the ground surface during the leafless period.This is in agreement with the higher daytime DR 30-1 values of NO 3 -than SO 4 2-during the three periods (Fig. 4).Conversely, the air temperature gradients mentioned above were not clearly seen in the nighttime data during the three periods, even though the DR 30-1 values of NO 3 -were also higher than SO 4 2-in the nighttime.The NH 4 NO 3 aerosols have an equilibrium relationship with the concentrations of the HNO 3 and NH 3 gases in the atmosphere.The V d value of HNO 3 is known to be greatly higher than those of NH 3 and NH 4 NO 3 .The V d values calculated by the resistance models are 3-10 times higher than those of other gaseous and particulate matter (Ban et al., 2016).During the leafy periods, HNO 3 was quickly removed by the canopy surface due to the high V d values and the large leaf area.When HNO 3 is quickly removed by deposition surfaces, its concentration near these surfaces drastically decreases.Then, the gas-particle equilibrium is shifted to the gas phase.After that, NH 4 NO 3 near the surfaces volatilizes to resupply the decreased level of HNO 3 and quickly deposits to the surfaces as gaseous matter.This can cause a high DR 30-1 value for NO 3 -not only in the daytime but also in the nighttime.This is likely because NH 4 NO 3 can be removed from the atmosphere just as quickly as SO 2 (Fig. 3) and therefore the DR 30-1 value of NH 4 NO 3 can become significantly larger than that of (NH 4 ) 2 SO 4 (p<0.01)(Fig. 4).Katata et al. (2020) applied a new multi-layer land surface model, which was coupled with the NH 4 NO 3 gasparticle conversion processes, to our results of the 16-autumn observation.The model reproduced the differences in the vertical profiles between NO 3 -and SO 4 2in the PM 2.5 well, particularly in the daytime.This revealed that the volatilization of NH 4 NO 3 below the canopy under dry conditions enhanced the deposition flux of HNO 3 converted from NH 4 NO 3 .The DR 30-1 value of NO 3 -calculated by the model was similar to that observed in the daytime but was smaller than that observed in the nighttime.These results are likely due to the uncertainty in the process of the equilibrium shift of NH 4 NO 3 due to the low concentrations of HNO 3 and other nighttime processes.

CONCLUSIONS
To better understand the dry deposition process of NO 3 -in PM 2.5 in a forest, we conducted vertical profile measurements of SO 4 2-and NO 3 -in PM 2.5 , as well as SO 2 , in a forest in suburban Tokyo, Japan, focusing in particular on the daytime/nighttime and leafy/leafless conditions.The observations were performed during the daytime and nighttime during two leafy periods and one leafless period.The vertical gradients of NO 3 -were clearly larger than those of SO 4 2-in the PM 2.5 during both the daytime and nighttime, especially for the leafy periods.Moreover, the daytime decreasing rate of NO 3 in the PM 2.5 below the canopy was larger than that of SO 2 during the leafy periods.
The large NO 3 -gradients in the PM 2.5 were caused by the equilibrium shift from NH 4 NO 3 to NH 3 and HNO 3 near the deposition surfaces.In the daytime, the air temperature was higher near the canopy surface during the leafy periods and near the ground surface during the leafless period.These conditions enhanced the volatilization of NH 4 NO 3 near the deposition surfaces in the daytime.Moreover, the lower concentration of HNO 3 near the surfaces caused by its fast removal enhanced the volatilization during both the daytime and nighttime.Therefore, NO 3 -in the PM 2.5 was quickly removed by the forest and its vertical gradients were larger than those of SO 4 2-in the PM 2.5 , even to the point of being equal to those of SO 2 .
Because the abovementioned chemical processes during dry deposition are not treated in current chemical transport models, future studies focused on the quantification of these processes are required to improve the model accuracy to estimate PM 2.5 and nitrogen deposition.

Fig. 1 .
Fig. 1.Schematic diagram of the observation tower at the Field Museum Tamakyuryo site.

Fig. 2 .
Fig. 2. Temporal variations in the mass concentrations of SO 4 2− and NO 3 − in the PM 2.5 at 30 m, 23 m, 8 m, and 1 m at the tower during the three observational periods.D and N indicate daytime and nighttime, respectively.
Speulderbos in the Netherlands using the gradient method.They found that the V d value of NO 3 -was larger than that of the maximum theoretically possible value when the temperature was high (above 20°C) and much larger than that of SO 4 2-, possibly due to the equilibrium shift from NH 4 NO 3 to NH 3 and HNO 3 .Van Oss et al. (1998) used these results in their simulations, which considered the influence of the gas-to-particle conversion on surface

Fig. 3 .
Fig. 3. Normalized vertical profiles of the mass concentrations of SO 4 2− in the PM 2.5 (open circles), NO 3 − in the PM 2.5 (closed circles), and SO 2 (open squares) during the daytime and nighttime for the three observational periods.The relative concentration is the concentration ratio with respect to the concentration at 30 m.The gray layers indicate the leafy canopies.
exchange processes above a forest.Their simulation results suggest that the volatilization of particulate NH 4 NO 3 during the daytime can lead to the emission of HNO 3 and NH 3 above the forest, along with the observed anomalously large V d value for NO 3 -compared to the theoretical value.Based on the observations in Nemiz et al. (2004) of a dry inland heath dominated by Calluna vulgaris in Elspeetsche Veld in the Netherlands, the surface concentration products of HNO 3 and NH 3 were well below the thermodynamic equilibrium value and the Damköhler numbers indicated that the chemical conversion was sufficiently fast to modify the exchange fluxes.

Fig. 4 .
Fig. 4. Distributions of the decreasing rates of SO 4 2− and NO 3 − in the PM 2.5 from 30 m to 1 m for the three observational periods.Each circle shows a daytime or nighttime daily value.The bars indicate the mean values.

Fig. 5 .
Fig. 5. Variations in the ensemble mean values of the air temperature at 30 m, 25 m, 20 m, 10 m, 6 m, and 1 m for the three observational periods.The figures ((b), (d), (f)) are an expansion of the period from 10:00 to 14:00.

Table 1 .
Experimental periods and sampling strategies.

Table 2 .
Atmospheric conditions at a height of 30 m at the tower and the leaf area index.WS, Temp, RH, and LAI indicate the wind speed, temperature, relative humidity, and leaf area index, respectively.All values except LAI are given as the average±standard deviation from the 10-min means of the data for the meteorological parameters and the half-day data for the concentrations.