References

Nitrous acid (HONO) plays a significant role in the atmosphere, especially in the polluted troposphere. Its photolysis after sunrise is an important source of hydroxyl free radicals (OH). Measurements of nitrous acid and other pollutants were carried out in the Kathmandu urban atmo- sphere during January-February 2003, contributing to the sparse knowledge of nitrous acid in South Asia. The results showed average nocturnal levels of HONO (1.7±0.8 ppbv), NO2 (17.9±10.2 ppbv), and PM10(0.18±0.11 mg m 3 ) in ur- ban air in Kathmandu. Surprisingly high ratios of chemi- cally formed secondary (HONO) to (NO2) (up to 30%) were found, which indicates unexpectedly efficient chemical con- version of NO2 to HONO in Kathmandu. The ratios of (HONO)/(NO2) at night were found to be much higher than previously reported values from measurements in urban air in Europe, North America and Asia. The influences of aerosol surface, ground reactive surface, and relative humidity on NO2-HONO chemical conversion were discussed. The high humidity, strong and low inversion layer at night, and high aerosol pollution burden in Kathmandu may explain the par- ticularly efficient conversion of NO 2 to HONO.


Introduction
Atmospheric nitrous acid (HONO) has been studied for a long time, and at a wide range of locations.It can be an important source of hydroxyl free radical (OH) in the atmosphere, especially in the morning (Lammel and Cape, 1996;Harrison et al., 1996).Despite its importance, and several decades of research, nitrous acid's formation mechanism remains unclear.Further investigation of the sources of nitrous acid in the atmosphere is important for improving our understanding of atmospheric chemistry processes.
One source of atmospheric HONO is the gas phase reaction of NO with OH (Seinfeld and Pandis, 1998;Lammel, 1999;Kleffmann, 2007).However, gas phase reactions (including NO with OH and NO 2 with peroxy or alkoxy radicals - Stockwell and Calvert, 1983) are too slow to explain the atmospheric HONO concentration.Direct emission from combustion engines is another source of HONO in the atmosphere (Kurtenbach et al., 2001;Kirchstetter et al., 1996;Wormhoudt et al., 2007;Geiger et al., 2002).Ratios of [HONO] to total [NO x ] ([NO]+[NO 2 ]) ranging from 0.3%-0.8%have been observed in direct emissions from vehicle engines.However, observations of [HONO]/[NO 2 ] ratios of a few percent in a wide range of urban atmospheres suggest that direct emission from combustion engines is probably not a major source of HONO.
Published by Copernicus Publications on behalf of the European Geosciences Union.
Recently, the reaction of NO 2 on soot particles (Reaction 4) has been suggested to be a source of HONO in the atmosphere (Aubin and Abbatt, 2007;Alcala-Jornod et al., 2000;Nienow and Roberts, 2006;Ammann et al., 1998;Gerecke et al., 1998;Longfellow et al., 1999;Arens et al., 2001;Al-Abadleh and Grassian, 2000;Aumont et al., 1999;Grassian, 2001Grassian, , 2002;;Karagulian and Rossi, 2007).Soot (or black carbon) aerosols often have a loading of 1.5 to 20 µg m −3 in urban air, mostly from anthropogenic sources, e.g., fossil fuel combustions (IPCC, 2001).The fractal and porous surface of soot provides a large contribution to the total reactive aerosol surface even when soot may have a relatively small contribution to the total aerosol mass.Studies using black carbon and soot as a substrate have demonstrated the reactive uptake of NO 2 and the formation of NO and HONO (Aubin and Abbatt, 2007;Kleffmann, 2007).Reaction (4) will contribute significantly to the production of atmospheric HONO only under certain conditions of high NO 2 concentrations and high and fresh soot loadings with large surface areas (Arens et al., 2001;Aubin and Abbatt, 2007;Kleffmann et al., 1999), and is thus not important in many real atmospheric conditions with low NO 2 and soot concentrations.
HONO formed in homogenous and heterogeneous reactions is released to the gas phase and accumulates during the night.Photolysis of HONO after sunrise provides a pulse of strong OH formation, which strongly influences atmospheric chemistry processes.Field measurements of HONO and its precursor NO 2 at locations with various aerosol loads, aerosol compositions, relative humidity, boundary layer heights and vertical transport efficiencies are thus essential for improving our understanding of the HONO formation mechanism.
Gaseous HONO in ambient air was measured for the first time by Perner and Platt (Perner and Platt, 1979) using differential optical absorption spectroscopy (DOAS).Besides the DOAS technique, a variety of other methods have been developed for the detection of HONO in the atmosphere with high sensitivity and good time resolution (Liao et al., 2006a,b;Takenaka et al., 2004;Trebs et al., 2004;Kleffmann et al., 2002;Huang et al., 2002;Vecera and Dasgupta, 1991;Heland et al., 2001).Measurements of HONO in the atmosphere have been carried out in many areas of Europe and the US.Elevated atmospheric HONO concentrations have also been detected in some Asian cities, suggesting HONO to be an important OH source in these places (Table 1).However, knowledge about HONO sources, sinks, and concentrations is almost nonexistent in South Asia.Due to the different source strengths, aerosol burdens, and the composition of aerosol and ground surfaces, HONO chemistry in South Asia is expected to have different features from what has been found in Europe or North America.For example, the average atmospheric aerosol burden in Kathmandu in winter is much higher than in European and US cities (Sharma et al., 2002;Sharma, 1997;Yu et al., 2008a;Giri et al., 2006).Moreover, strong nocturnal inversion layer (Kondo et al., 2002;Regmi et al., 2003;Panday, 2006;Panday and Prinn, 2009) and the In this study, we present measurements of HONO and NO 2 using long path DOAS in Kathmandu, Nepal, over several weeks in 2003.As far as we are aware, this was the first study on HONO and NO 2 in the nocturnal atmosphere in South Asia.Our observations provide insights into the particular HONO chemical formation processes taking place in the Kathmandu atmosphere.

Location and instruments
We carried out a field campaign in Kathmandu in winter 2003, investigating air quality and atmospheric chemical processes.NO x (NO, NO 2 ), SO 2 , O 3 , HONO, HCHO, PM 10 , and meteorological parameters were measured during this period.
A home-made long-path DOAS (Chalmers University) was used to measure HONO and NO 2 in Kathmandu with a time resolution of around 5-10 min.DOAS measurements were carried out from 10 January to 11 February 2003.The measurement site (27 • 43.2190 N, 85 • 21.417 E) was about 2 km east of the Kathmandu Ring Road, along the main road figure in this file.5. Journals from API (i.e.J. Appl.Phys.) and AGU (i.e.J. Geophys.Res.& Geophys.Res.Lett.) have very special citation system.Please use revised references in the revised printout.from Kathmandu to Sankhu (Fig. 1).A DOAS telescope, a NO x analyzer (Model 42S Thermo Environmental Instruments Inc.), an O 3 monitor (2B Technologies), and a PM 10 aerosol monitor (TSI Dusttrak) were installed on the 5th floor of a hotel (15 m above ground).DOAS retro-reflectors were mounted at the top of a 7-story building, 25 m above ground and 957 m away from the DOAS telescope, giving a DOAS light path of 1914 m.Low buildings, roads, and some bare fields were under the light path.An automated weather station (Rainwise), which measured temperature, wind speed, wind direction, relative humidity, dew point temperature, and solar radiation, was installed on the roof of the hotel (22 m above ground) near the DOAS telescope and other instruments.Details of the DOAS instrument and the experimental sites have been described in earlier papers (Yu et al., 2008a(Yu et al., ., 2004)).
In our experiments, several steps were taken before the spectral retrieval to determine concentrations of HONO, NO 2 , and other gases.The spectra were corrected for electronic offset and light from sources other than the Xe arc lamp.The spectra were divided by a recorded halogen lamp spectrum, which only contains broadband structures, to eliminate spectral structures originating from the pixel-to-pixel variation of the quantum yield.A digital high pass filter was used to separate narrow absorption structures from the broadband structures caused by the lamp, Rayleigh and Mie scattering, and other instrument effects.Then, a digital low pass filter was applied to reduce high frequency noise.Finally, the spectra were analyzed for NO 2 (cross section by Vandaele et al., 2002), HONO (cross section by Stutz et al., 2000), and HCHO (cross section by Meller and Moortgat, 2000) using a nonlinear least-squares method (Stutz and Platt, 1996) in the spectral range of 310-380 nm with a spectra resolution of 0.4 nm.The sources of errors of the measured HONO and NO 2 mixing ratios include the error (5% for HONO and 3% for NO 2 ) of the cross sections, the spectral fit error (less than 5%), and the experimental error of the DOAS system (5%).The total errors for the measured HONO and NO 2 mixing ratios are estimated to be around 9% and 8%, respectively.

Retrieval method of aerosol parameters
To investigate the role of aerosols in HONO formation, measurements of aerosol parameters such as the aerosol surface density are required, along with the ambient concentration of HONO and NO 2 .Box and Lo (1976) developed a simple inversion method to calculate the approximate aerosol size distribution from light attenuation.This has been further applied to calculate other aerosol parameters such as the surface density and the volume using DOAS measurements (Notholt et al., 1991(Notholt et al., , 1992a;;Notholt et al., 1992b;Notholt and Raes, 1990;Flentje et al., 1997;Reisinger, 2000).
Light transmission measurements through the atmosphere can yield bulk aerosol parameters such as diameters, number density, total surface, and total volume of particles (Livingston and Russell, 1989).In Kathmandu, we used the DOAS system to measure the light transmission, aerosol parameters and trace gas concentrations simultaneously in the same volume of air.
The spectra were recorded in the 250-390 nm range.The range of 330 nm to 390 nm was used to retrieve concentrations of HONO, NO 2 and aerosol parameters simultaneously.A background spectrum was recorded over a 1 m optical path using retroreflectors.Given the concentrations of all gases provided by the DOAS spectra retrieval processes (SO 2 , O 3 , NO 2 , HONO, and HCHO), the optical depth (τ ) at wavelength λ can be calculated according to Eq. (1).
Here I(λ) is the light intensity at the end of the full path length while I 0 (λ) is the background light intensity at 1 m path length.Inversion of the optical depth is performed using a look-up table (Box and Lo, 1976).The spectral fit at 330-390 nm region was based on an assumed exponential wavelength-dependent function for optical depth: where τ exp (λ) is the optical depth, α and β are the fitting parameters, and k is the offset constant.The offset k must be considered because of instrument factors such as the reflecting ratio of retroreflectors.Using α, β, and the pre-calculated tables of Box and Lo (1976), values of a and b can be obtained for the calculation of Deirmendjian's particles size distribution n(r) (Box and Lo, 1976;Notholt and Raes, 1990;Notholt et al., 1992bNotholt et al., , 1991)): where r is the particle radius, and the constants a and b are specified for different particles types.By using the same method as described by Notholt et al. (1992a) but with a narrower spectral range, we calculated the aerosol size distribution and further obtained the total number, surface density and the volume of aerosols in time intervals of 5-10 min.
Our light attenuation retrieval processes give a large uncertainty of aerosol parameters.Errors in this method have been extensively discussed by others (Andres-Hernandez et al., 1996;Notholt et al., 1992a;Notholt and Raes, 1990;Box and Lo, 1976).The total error of our results was estimated to be 55%.The sources are: (1) The unknown refractive index of the aerosols (we have tested different indices from 1.33 to 1.70.Trends with different indices are similar but with different values) could contribute an uncertainty as large as 50% (Box and Lo, 1976); (2) The relatively narrow spectral range compared to the application by Notholt et al. (1992a) gave 20% uncertainty as reported by Reisinger ( 2000); (3) The deviation of the true aerosol size distribution from the calculated results by the model (Deirmendjian's particle size distribution) could introduce an additional uncertainty; (4) The consistency of the lamp intensity and DOAS mechanical stability is sufficient during a time period of a few weeks but with an uncertainty of 10% (Andres-Hernandez et al., 1996;Notholt et al., 1992a).
With such limitations, the results of aerosol parameters are only approximations of the true aerosol situation.Nevertheless, the retrieval results of aerosol surface and volume (assuming a refractive index n =1.5 - Yu et al., 2008b) agree well with the trends seen in our PM 10 data (Fig. 2), which confirms the reliability of our calculated aerosol parameters.

Time series of HONO, other pollutants and meteorological data
Continuous measurements were carried out during January and February, 2003. Figure 3 shows, as an example from our one-month observation, the behavior of HONO, NO x , [HONO]/[NO 2 ], PM 10 , and meteorological parameters (radiation, wind speed and wind direction) during a four day period.The wind direction becomes unimportant as wind speeds drop to zero during our observation.NO 2 and HONO measurements by the long path DOAS show gaps every morning and on some evenings due to heavy fog and smog.The concentration of NO had two peaks (with values as high as 60 ppb) at 07:00-08:00 and 19:00-20:00, respectively, and remained low (a few ppb except for short duration plumes) at night.As our campaign was at a time of www.atmos-chem-phys.net/9/6401/2009/Atmos.Chem.Phys., 9, 6401-6415, 2009   considerable political instability in Nepal, with frequent vehicle searches at night, night-time driving was quite low and thus the night-time sources of NO were limited.Peaks in NO 2 appeared concurrently with the NO peaks at 07:00-08:00 and 19:00-20:00.NO 2 values were around 10-20 ppb at night, and started to increase around 03:00-04:00.HONO concentrations climbed to several ppb after sunset and then decreased to around 1-1.5 ppb.It increased again after 03:00.PM 10 shows elevated values (up to 1 mg m −3 ) during the morning and evening.The diurnal variation of the observed wind pattern shows a long calm period from sunset (∼18:00) until around 10:00 each day.The morning and evening peaks in NO x and PM 10 did not correspond to rush hour alone.In particular, the evening peaks started after the rush hour.The timing of pollutant peaks in Kathmandu is determined by the valley's ventilation pattern (Panday, 2006;Panday and Prinn, 2009).The correlation of HONO with NO 2 and PM 10 (especially after sunset) in Fig. 3 suggests three possibilities: (1) the same emission source; (2) chemical conversion of NO 2 to HONO associated with aerosols; (3) all three species have their sources near the ground (not necessarily the same sources).The HONO formation mechanism will be discussed in later sections.
Table 2 shows several pollutants measured in Kathmandu's urban air.Elevated mixing ratios of HONO (up to ∼7.5 ppbv) were observed during the whole measurement period.The elevated HONO concentrations in the presence of high [NO], [NO 2 ] and PM 10 values right after sunset appear to be the result of several processes: (1) direct emission at evening rush hours; (2) decrease of boundary layer height During afternoon rush hours and other plume events (around 19:00, 20:00, and 21:00), [HONO] followed the same pattern as [NO 2 ] (Fig. 3).However, during those times a lower [HONO]/[NO 2 ] value was observed compared to a few hours later, similar to the observation in Houston (Stutz et al., 2004).It is likely that conversion of NO 2 to HONO in relatively freshly polluted air has not proceeded as far as in the aged air in this period.
A scatter plot of nocturnal [HONO] and [NO 2 ] is shown in Fig. 4. The regression coefficient, maximum and minimum ratio of [HONO] to [NO 2 ] were 0.081, 0.30 and 0.008, respectively.The maximum [HONO]/[NO 2 ] ratio (30%) found in our observation is, to our knowledge, the highest reported atmospheric HONO to NO 2 ratio, with the exception of daytime findings in and above snow (Zhou et al., 2001).The high ratios of [HONO]/[NO 2 ] in Kathmandu at night indicate high HONO direct emissions, or highly efficient NO 2 to HONO chemical conversion, or both.In the next sections, we discuss the contribution of direct emissions and chemical processes separately.

Contribution of direct emissions to the observed HONO
It is necessary to separate HONO formed by chemical reactions from HONO that is directly emitted (mainly from vehicles; the contribution from industrial combustion processes is not clear).No previous study reported the ratio of [HONO]/[NO x ] in emissions in Kathmandu.The highest reported average emission ratio anywhere in the recent literature was (0.8±0.1) % (Kurtenbach et al., 2001) This maximum ratio of 30% is substantially higher than the previously observed ratios at other urban and suburban sites in Europe, North America and Asia (i.e., 1-12% reported by Lammel and Cape (1996), 19% reported by Wang et al. (2003) during dust storms, and 1-13% around Asia - Qin et al., 2006;Park et al., 2004;Hao et al., 2006).

NO 2 -HONO chemical conversion and correlation studies
The contribution by NO to the formation of HONO through Reactions (2-3) could be ruled out for Kathmandu late at night because of the low NO concentrations.According to HONO formation mechanisms (Reactions 1 to 6 in the introduction), the major source of HONO in the nocturnal air in Kathmandu was the NO 2 to HONO chemical conversion through heterogeneous reactions (i.e.Reactions 1, 4, and 5).HONO formation has been found to be a first order reaction in NO 2 (Finlayson-Pitts et al., 2003;Kleffmann et al., 1998a).It depends on the surface absorbed water, and increases as the reactive surface to air volume ratio (S/V) increases (Stutz et al., 2004;Trick, 2004).The value of S/V is determined by a combination of the aerosol surface, ground reactive surface and the influence of vertical mixing (Febo et al., 1996;Trick, 2004).Here we investigate S/V ratio and relative humidity in the nocturnal atmosphere in Kathmandu and their influence on HONO formation.
The correlation of HONO with other pollutants in Fig. 3 indicates the following.(1) the presence of all the compounds in the same air mass; (2) a chemical conversion of NO 2 to HONO associated with aerosols; (3) all three species have their sources near the ground (not necessarily the same sources).In order to investigate the efficiency of NO 2 to HONO chemical conversion without the influence of other processes (i.e.transport), we study the correlation of [HONO]/ [NO 2 ] instead of [HONO] to the other parameters.

Aerosol surface and ground reactive surface
Long path DOAS measurements can provide information about aerosol surface-to-air volume ratios (see Sect. 2).The ground reactive surface area density is defined in this paper as the inverse value of the mixing layer height (H −1 ) (Harrison et al., 1996).We use average mixing layer heights from Sonic Detection & Ranging (SODAR ) measurements (Panday, 2006).The average measured mixing layer height deceased from 12:00 (800 m) until 00:00 (120 m), and then re-mained approximately constant (120 m) from 00:00 to 06:00.It has to be noted that mixing layer height measurements by SODAR are less accurate at night than during the day.
The porosity of aerosols and the roughness of the ground surface were not considered in this study.The area densities of aerosol surfaces and ground reactive surface might be higher (up to 10 times) (Lammel and Cape, 1996;Andrés-Hernández et al., 1996) than the values calculated from light attenuation and estimation by mixing layer height, respectively.
Figure 5a shows the behaviors of [HONO]/[NO 2 ], and during a selected night.The black line in Fig. 5a is the combination of the two surface area densities.[HONO]/[NO 2 ] increased after sunset and remained relatively constant after 00:00, then decreased after 03:00.S V aerosol increased quickly after sunset for a short period (17:00-19:00), and then stayed nearly constant until 03:00.S V aerosol , as well as PM 10 , dramatically increased after 03:00.The change of S V aerosol has a positive correlation with [HONO]/[NO 2 ] right after sunset.However, for the period of 17:00 to 03:00, S V aerosol has a poor correlation (R 2 = 0.33) to [HONO]/[NO 2 ] (see Fig. 5b), which disagree with observations in some field measurements elsewhere (Notholt et al., 1992a;Reisinger, 2000).It is interesting that S V ground shows good correlation with [HONO]/[NO 2 ] (R 2 = 0.93) for the whole night except between 03:00 and 06:00 in Fig. 5b.The good correlation between [HONO]/[NO 2 ] and S V ground indicates that the NO 2 heterogeneous reaction on the ground reactive surface was a major source of HONO in Kathmandu.Our observation of ground reactive surface as a source of atmospheric nitrous acid is consistent with other field observations (Harrison and Kitto, 1994;Harrison et al., 1996;Kleffmann et al., 2003;Vogel et al., 2003).
Although the general correlation of [HONO]/[NO 2 ] with S V aerosol was poor, the positive correlation in the early hours after sunset (17:00-19:00) points to the possible contribution of aerosols to HONO formation under certain conditions.Recently many researchers have reported the fast conversion of NO 2 on fresh soot surface to form HONO (Reaction 4).Since the soot surface is shown to rapidly deactivate as the reaction proceeds, it is believed to be an insignificant source of HONO in the real atmosphere.However, the aerosol concentrations in Kathmandu air are much higher than in those cities, where research on this topic was carried out.It is thus worthy to investigate the potential impact of soot on HONO formation in Kathmandu atmosphere.
Although aerosol chemical properties in the Kathmandu urban area were not provided by our measurements, literature and many reports in local newspapers suggest that large amounts of soot are often present in the winter Kathmandu urban air.Some researchers have attributed the dark haze over Kathmandu valley to the presence of soot particles (Sapkota and Dhaubhadel, 2002).Others investigated the aerosol chemical properties in Nagarkot, at the rim of the Kathmandu valley, and found that the aerosol featured a dominant contribution from carbon, likely from fuel combustion and biomass burning processes in the Kathmandu valley (Carrico et al., 2003).More recent research shows elevated levels of black carbon and organic carbon in the Kathmandu urban air from September 2004 to August 2005 (Adhikary et al., 2007;Ramanathan et al., 2007).It has been estimated that brick-kilns are responsible for about 28% of annual PM 10 emissions in the valley, almost all of which is concentrated during the winter months (World Bank, 1997).The operation of 125 brick kilns in the Kathmandu valley during the dry season from December to April evidently contributes to the aerosol pollution, including black carbon emissions (Dhakal, 2003;Adhikary et al., 2007).
In this study, we used an integrated HONO formation rate for unit surface area of soot particles, 1.3×10 14 molecules cm −2 (average of the values provided by Aubin and Abbatt, 2007;Kleffmann and Wiesen, 2005;Kirchner et al., 2000).By assuming that all observed parti-cles were soot, we got an upper estimate of the average soot particle surface.The calculated [HONO]/[NO 2 ] values corresponding to Reaction (4) are plotted in Fig. 5a (triangles).The maximum HONO formed by NO 2 conversion on soot is found to be ∼ (0.3 ± 0.2) ppb, which corresponds to a [HONO]/[NO 2 ] ratio of 3 ± 2%.Compared to our observations during most parts of the nights, this is a very small contribution.However, for the early hours after sunset (17:00-19:00), the majority of the observed [HONO]/[NO 2 ] ratio (below 5%) could come from Reaction (4) on soot particles.Despite the fact that soot may play a role in NO 2 -HONO conversion in Kathmandu right after sunset based on our calculation, we should note that: (1) there is inadequate information about aerosol chemical composition; (2) the contribution of this process to total nocturnal atmospheric HONO formation is most likely limited.

Relative humidity (RH)
The RH dependent HONO formation has been studied in laboratory research (Harrison and Collins, 1998) and in field observations (Stutz et al., 2004).Syomin and Finlayson-Pitts (2003) have studied the decomposition of HONO under different RH conditions.The RH in the winter of Kathmandu increased from a sunset value of ∼60% to ∼100% by 03:00 and remained close to 100% until sunrise during most nights.However, the correlation of [HONO]/[NO 2 ] with RH in Fig. 6 might not be a proof for the humidity dependence of HONO formation in Kathmandu atmosphere.The increasing RH after sunset was the result of the decreasing temperature (panel C in Fig. 6).Meanwhile, the increasing [HONO]/[NO 2 ] during the night was due to the combination of ceasing photochemistry, heterogeneous formation and a decreasing boundary layer height.Simultaneous increase of RH and [HONO]/[NO 2 ] as shown in Fig. 6 might be two independent processes.In order to test the influence of RH on the NO 2 -HONO conversion reactions, we study the average HONO formation rate constant during the night against the average humidity.
HONO formation has been found to be first order in NO 2 , dependent on the surface absorbed water, and increasing with aerosol surface to air volume ratio (Kleffmann et al., 1998b;Sumner et al., 2004;Finlayson-Pitts et al., 2003).
We can thus use Eq. ( 4) to describe the HONO formation at night: k RH is an assumed RH dependant HONO formation rate constant.
Here we determine the HONO formation by assuming a linear increase of its mixing ratio during the time period of 18:00-00:00 in Kathmandu air.The estimation of the HONO formation rate constant in each individual night is based on the equation introduced by Alicke et al. (2002): The average nighttime (18:00-00:00) HONO formation rate is (0.014 ± 0.004) (ppb HONO per hour per ppb NO 2 ) during our observation period, which is slightly higher than the observations in Milan (Alicke et al. 2002).The HONO formation rate constant is normalized by the S/V, and then plotted against the average RH value of each individual night (see Fig. 7).The HONO formation evidently decreases as RH increases from 65-100% during the night in Kathmandu.The HONO formation in RH range of 95-100% is only about half of the value in RH range of 65-75%.This negative RH dependence of HONO formation in our observations is most likely due to HONO lost in the growing aqueous layer on the ground.The water adsorption on the ground surface can be determined by BET isotherm.The number of monolayers formed by absorbed water on various surfaces increases rapidly when RH reaches 60% (Sumner et al., 2004;Lammel, 1999).Lammel (1999) had reported that the water coverage of various materials (stone, soil, and vegetation) increases dramatically (equivalent to up to 100 monolayers) when RH is over 95%.A surface with a RH of over 95% is classified as an "aqueous" surface.Such  an "aqueous" surface affects the reaction efficiency of NO 2 -HONO conversion (Stutz et al., 2004;Lammel, 1999), which is in accordance with our observation.The water uptake processes can take place on the ground surface (major HONO formation source in this paper) as well as aerosol surfaces.As many earlier papers reported, water droplets (on ground or aerosol) at a RH of over 96% play a role in the HONO sink, leading to lower [HONO]/[NO 2 ] values, which is in agreement with our observation (Lammel and Perner, 1988;Zhou et al., 2007;He et al., 2006;Acker et al., 2005Acker et al., , 2001;;Lammel, 1999).

Conclusions
Measurements of NO x , HONO, PM 10 , surface density, and meteorological parameters in winter in Kathmandu were presented.To our knowledge, this was the first time that HONO was measured in South Asia.
Very high values of chemically formed HONO to NO 2 ratios were found (up to 30%), which is higher than any other previous atmospheric observations that we are aware of.
We believe that direct emissions and heterogeneous reactions involving NO are not the major sources of HONO in Kathmandu's nocturnal atmosphere.NO 2 heterogeneous chemical processes appear to be responsible for the high and periodic HONO levels under the conditions of high aerosol pollution burden, strong inversion, and high humidity.Kathmandu's very low nocturnal mixing layer height provides a large reactive surface density for HONO formation through NO 2 heterogeneous reaction, which is confirmed by the good correlation of ground reactive surface area density and [HONO]/[NO 2 ] values.
The high relative humidity found in Kathmandu at night provides efficient water uptake on reactive surfaces.When the relative humidity exceeds 96% in the early morning, water condenses on aerosol and ground to form water droplets, which appears to be a sink for gaseous HONO even before sunrise.

Fig. 1 .
Fig. 1.Map of Kathmandu showing the measurement site.

Figure 4
Figure 4. [HONO] vs. [NO 2 ] in the nocturnal atmospheric boundary layer at an urba Kathmandu.The error bars are 1σ standard deviation.

Fig. 4 .
Fig. 4. [HONO] vs. [NO 2 ] in the nocturnal atmospheric boundary layer at an urban site of Kathmandu.The error bars are 1σ standard deviation.
Fig.6 Fig. 6. [HONO]/[NO 2 ], relative humidity, and temperature in the nocturnal Kathmandu atmosphere during the whole observation period.Lines with plus marks show average values.

Figure 7 .
Figure 6.[HONO] / [NO 2 ], relative humidity, and temperature in the nocturnal Kathman atmosphere during the whole observation period.Lines with plus marks show average va

Fig. 7 .
Fig. 7. Normalized HONO formation rate constants decrease with the increasing RH.Red open circles are individual HONO formation values of each night.Black dots are average values of HONO formation in RH range of 65-75%, 75-85%, 85-95% and 95-100%.Error bars are the standard deviations.

Table 1 .
HONO and NO 2 measurements in Asia.

Table 2 .
Pollutants in Kathmandu urban air.
Yu et al.:Observations of enhanced NO 2 -HONO conversion in Kathmandu minimum value of [HONO]/[NO 2 ] was ∼ 0.8%; corresponding to the maximum possible contribution by HONO direct emissions.The high value of [HONO]/[NO 2 ] (upto 20% during most nights, exceeding 25% during some nights, and reaching a maximum of 30% during one night) is thus mostly due to secondary HONO.
2 ] ratio was low (around a few %) after sunset when emissions were high and high NO x concentrations were observed.Second, NO x concentrations were low in the middle night, ruling out the possibility of direct emissions as the major source of HONO.Third, as we see in Fig.4, the www.atmos-chem-phys.net/9/6401/2009/Atmos.Chem.Phys., 9, 6401-6415, 2009 Y.