Hydroxyl (OH), hydroperoxy (HO2) radicals, collectively known as HOx, and OH reactivity, were measured during the PMTACS–NY (PM2.5 Technology Assessment and Characteristics Study-New York) summer 2002 intensive at Whiteface Mountain, Wilmington, New York. The measurement results of OH and HO2 for 4 weeks are presented. Diurnal cycles show that the average noontime maximum mixing ratios were about 0.11 pptv (2.6 × 106 cm−3) for OH and 20 pptv for HO2. Measured HO2 to OH ratios were typically between 40 and 400, which are greater than those obtained in polluted and semipolluted rural environments. Low but significant mixing ratios of OH and HO2 persisted into early evening and were frequently observed during nighttime, consistent with previous studies in different environments. Steady state OH and HO2 were calculated with a zero-dimensional chemical model using a complete Regional Atmospheric Chemical Mechanism (RACM) and a parameterized RACM which was constrained to the measured OH reactivity. Good agreement was obtained between the complete RACM and the parameterized RACM models. On average, the complete RACM model reproduced the observed OH with a median measured-to-modeled OH ratio of 0.82 and daytime HO2 with a median measured-to-modeled HO2 ratio of 1.21. The reasonably good agreement in this study is inconsistent with the significant underestimation of OH in the Program for Research on Oxidants: Photochemistry, Emissions, and Transport in 1998 (PROPHET98) study at a similar forested site. HOx budget analysis indicates that OH was primarily from the photolysis of HONO and O3 during the day and from O3 + alkenes reactions at night. The main HOx loss was the self reaction of HO2. The good agreement between the measured and calculated OH reactivity in this environment contrasts with findings in the PROPHET2000 study, in which significant OH reactivity was missing and the missing OH reactivity was temperature-dependent.
 Hydroxyl (OH) and hydroperoxy (HO2) radicals, collectively known as HOx, play important roles in the oxidation of many atmospheric compounds and in the formation of secondary pollutants such as ozone (O3) and fine particles. In the clean troposphere, the photolysis of ozone followed by the subsequent reaction of O(1D) with water vapor is the main HOx source during daytime. In more polluted environments, the photolysis of other oxygen-containing species, such as nitrous acid (HONO), formaldehyde (HCHO), and hydrogen peroxide (H2O2), can also be important HOx sources. Besides these photolytic HOx sources, other nonphotolytic HOx sources have been discovered in recent studies. For instance, O3 reactions with alkenes can produce a significant amount of OH and HO2 [e.g., Donahue et al., 1998; Salisbury et al., 2001].
 Forests emit significant quantities of biogenic hydrocarbons that remain largely unstudied. Oxidation of these biogenic species leads to the formation of ozone and secondary organic aerosol [Fehsenfeld et al., 1992; Kurpius and Goldstein, 2003]. There are a few ground-based field studies focusing on atmospheric oxidation in forested areas which include HOx observations. In some studies, observed HOx results have been compared with the expectations from constrained models.
 During the Tropospheric OH Photochemistry Experiment (TOHPE), McKeen et al.  found that the model overestimated OH concentrations by a factor of 1.5 on average, which is consistent with the model-measurement comparisons conducted at some continental sites [e.g., Poppe et al., 1995; George et al., 1999]. Possible reasons include inadequate model representation of hydrocarbon chemistry or of uptake of HOx by aerosols.
 In contrast to these studies in which OH and HO2 were overpredicted by the models, Carslaw et al.  found that observed OH was underpredicted by the model at a forested site in northern Greece during the AEROsols formation from BIogenic organic Carbon (AEROBIC) campaign. The modeled OH concentrations were 50% of those measured, whereas the measured and modeled HO2/OH ratio was greater than what has been found previously for similar NOx (NO + NO2) environments. OH was also significantly underpredicted at another forested site in northern Michigan during the Program for Research on Oxidants: Photochemistry, Emissions, and Transport in summer 1998 ((PROPHET98) [Tan et al., 2001]. Another interesting feature of this study was that significant nighttime HOx radicals were observed at this site [Faloona et al., 2001]. The nighttime OH level is consistent with that derived from isoprene decays in the evening and can be reproduced by the model by including certain extremely reactive species that can react with O3 to produce HOx.
 In the PROPHET2000 study conducted in the summer of 2000 at the same site as PROPHET98, HOx amounts and diurnal behavior were about the same as in 1998. However, in 2000, the total first-order OH loss rate, called OH reactivity, was also measured. A significant fraction of the OH reactivity was missing, and this missing OH reactivity was temperature-dependent [Di Carlo et al., 2004]. An estimate of the amount of biogenic volatile organic compounds (BVOCs) derived from the observed missing OH reactivity is consistent with the amount of BVOCs necessary to generate the observed nighttime OH. These results are also consistent with other observations in other forests: aerosol production [Kulmala et al., 2001; O'Dowd et al., 2002] and significant missing ozone loss [Kurpius and Goldstein, 2003]. A worthy question is: Will missing OH reactivity be found in other forests?
 In this work, we present the measurement data and model comparison of OH and HO2 and calculated and measured OH reactivity during the PM2.5 Technology Assessment and Characterization Study–New York (PMTACS–NY) summer 2002 intensive at Whiteface Mountain in Wilmington, New York. The initial purpose of this work was to investigate photochemistry in a lightly polluted rural area upwind of the polluted Northeast corridor to provide a comparison for measurements made in New York City in the summer of 2001 [Ren et al., 2003a, 2003b]. In addition, this study provided an opportunity to test for missing OH reactivity in another forested environment.
2. Experimental and Model Description
2.1. Site Description
 The PMTACS–NY Summer 2002 Intensive campaign took place at an atmospheric monitoring station at Whiteface Mountain, which is operated by the Atmospheric Sciences Research Center (ASRC) at State University of New York (SUNY), University at Albany. Sampling trailers were located at a 600 m lodge site (44°23′27″N, 73°51′32″W) on a shoulder of Whiteface Mountain within the deciduous forest canopy. Most instruments were positioned in or beside the temperature-controlled trailer laboratories. The field campaign was held from the beginning of July through the beginning of August 2002.
 Whiteface Mountain is far away from major urban areas but is routinely in the path of regional air masses that have been influenced by urban emissions [Gong and Demerjian, 1997]. The nearest urban centers are Montreal, Canada, 130 km to the north; Albany, New York, 180 km to the south; New York City, 400 km to the south; Syracuse, New York, 220 km to the southwest; Boston, Massachusetts, 320 km to the southeast; and the Buffalo, New York/Toronto, Canada metropolitan area, 450 km to the west. The air at the site is characterized by relatively low levels of NOx and anthropogenic volatile organic compounds (VOCs), with periodic pollution plumes from these metropolitan areas and forest fires in Canada.
2.2. Instrumental Description
2.2.1. OH and HO2 Measurement
 The OH and HO2 radicals were measured with the Penn State GTHOS (Ground-based Tropospheric Hydrogen Oxides Sensor) which detects OH and HO2 with laser-induced fluorescence (LIF, often called fluorescent assay by gas expansion (FAGE), which was originally developed by Hard et al. ). A detailed description of the GTHOS instrument can be found elsewhere [Faloona et al., 2004] and here an abbreviated description of the field instrument is given. The detection module was mounted on top of a 5 m scaffolding tower above the ground, whereas the laser and data acquisition system were housed inside a trailer laboratory located at the base of the tower.
 The air sample is drawn into a low-pressure chamber through a pinhole inlet (1.0 mm) with a vacuum pump. As the air passes through a laser beam, OH is excited by a spectrally narrowed laser with a pulse repetition rate of 3 kHz at one of several vibronic transition lines near 308 nm (A2Σ–X2Π, v′ = 0 ← v″ = 0). Collisional quenching of the excited state is slow enough at the chamber pressure (∼4.5 hPa) that the weak OH fluorescence extends beyond the prompt scattering (Rayleigh and wall scattering) and is detected with a time-gated microchannel plate (MCP) detector. HO2 is measured by the reaction with NO followed by the LIF detection of OH. The OH and HO2 detection axes are in series: OH is detected in the first axis and HO2 in a second axis as reagent NO (>99%, Matheson, Twinsburg, Ohio, purified through Ascarite) is added to the flow between the two axes. The OH fluorescence signal is detected 60 ns after the laser pulse has cleared in the detection cells and is recorded every 0.2 s. The laser wavelength is turned on and off resonance with an OH transition every 10 s, resulting in a measurement time resolution of 20 s. The OH fluorescence signal is the difference between on-resonance and off-resonance signals.
 The instrument was calibrated both in the laboratory and during the field campaign. Monitoring laser power and Rayleigh scattering maintained this calibration during the campaign. For the calibration, water vapor photolysis by 185 nm light produced OH and HO2. Absolute OH and HO2 mixing ratios were calculated by knowing the 185 nm flux, which was determined with a NIST-calibrated photomultiplier tube from the University of Colorado, the H2O absorption cross section, the H2O mixing ratio, and the exposure time of the H2O to the 185 nm light. The absolute uncertainty was estimated to be ±32% for both OH and HO2, with a 2σ confidence level. The 2σ precisions during this campaign were about 0.01 pptv (∼2.4 × 105 cm−3) for OH and 0.1 pptv (∼2.4 × 106 cm−3) for HO2, with 1 min integration time. Further details on the calibration process may be found elsewhere [Faloona et al., 2004].
2.2.2. OH Reactivity Measurement
 During this campaign, a Total OH Loss rate Measurements (TOHLM) instrument was used to measure the first-order OH loss rate, called OH reactivity. Details about this instrument can be found elsewhere [Kovacs and Brune, 2001; Kovacs et al., 2003]. In short, ambient air was pulled by a blower through a tube and past an OH detection axis, where OH was detected in the same manner as described with GTHOS. OH was mixed into the ambient air through a movable injector. As the injector was pulled back in discrete steps, the decay in the OH signal was measured. The decay rate was determined by the equation:
where distance/velocity = time of reaction and the wall loss, kwall, was determined by looking at the OH decay in zero air (kwall ∼ 1.5 ± 0.4 s−1). The absolute uncertainty of the TOHLM measurement is about ±10% with 1σ confidence level. Because of low levels of NO in this environment, the OH reactivity correction due to the HO2-NO-OH cycling is not significant [Kovacs et al., 2003].
2.3. Ancillary Measurements
 In addition to OH, HO2 and OH reactivity, some other trace gases, such as O3, NO, NO2, CO, SO2, HONO, HNO3, HCHO, and speciated VOCs (including 47 nonmethane hydrocarbons and 11 oxygenates), and meteorological parameters were also measured during this campaign. Because of a leak problem in the VOC sampling line, valid VOC measurements are only available for the last week of the study after the leak problem was fixed. The techniques used to measure these species and parameters are listed in Table 1.
Table 1. Measurements and Techniques During the PMTACS-NY Summer 2002 Intensive Study at Whiteface Mountain
 Periodic NO spikes were observed at night when westerly wind dominated at the site. These spikes correlate with the position of the GTHOS vacuum exhaust, which contained NO, located about 50 meters to the west of the site. Some other NO spikes were determined to be from the use of a tractor to move heavy items. The NO spikes were removed from the data set to reflect the real environment at this site without the artificial NO contamination influence.
2.4. Model Description
 A zero-dimensional chemical model based on the Regional Atmospheric Chemical Mechanism [Stockwell et al., 1997] was used to calculate steady state OH and HO2 concentrations to compare with the measurements. Kinetic rate coefficients were updated using the results by Sander et al. . The model was constrained to the observed concentrations of O3, NOx, CO, SO2, VOC and meteorological parameters. All measured parameters were averaged using a time resolution of 10 min. H2 was fixed at 500 ppbv and CH4 at 1.8 ppmv for the whole campaign.
 The complete RACM constrained by the actual VOC measurements and other measured parameters was used for the last week of the campaign. In order to compare measured and modeled HOx for the first 3 weeks, a parameterized input of the VOC was required [Ren et al., 2005]. In the parameterized method, the measurements of OH reactivity, NO, NO2, SO2, CO, and O3 were used to calculate the OH reactivity due to VOCs. The speciated VOCs measured during the last week were then used to estimate the contributions of OH reactivity from individual VOCs. The average diurnal distributions of the measured VOCs, NO, NO2, CO, SO2, and O3 in the last week were used to partition the measured OH reactivity. The measured OH reactivity due to VOCs was estimated by subtracting the OH reactivity due to NO, NO2, CO, SO2 and O3 from the measured OH reactivity, and by scaling the total VOC concentrations so that the OH reactivity from the sum of the VOCs matched the measured OH reactivity due to VOCs [Ren et al., 2005].
 During the field campaign, photolysis frequencies, J values, were not measured directly. However, a Yankee ultraviolet multifilter rotating shadowband radiometer (UV-MFRSR) was used to measure the UV components of solar irradiance at seven different wavelength bands near 299, 305, 311, 317, 324, 332, 367 nm, each with a 2 nm effective bandwidth. J values were calculated using the NCAR Tropospheric Ultraviolet and Visible (TUV) transfer model (http://cprm.acd.ucar.edu/Models/TUV/) for clear sky values with the observed O3 column density specified for each day measured by the Total Ozone Mapping Spectrometer-Earth Probe satellite sensor (data available at http://toms.gsfc.nasa.gov/teacher/ozone_overhead.html). In order to correct for solar attenuation, the calculated J values on a clear day were used to calculate a cloudiness factor. The field experiment log and the profile of the solar UV radiation measured on 30 July indicated that it was a clean day with a clear sky. Thus the ratio of calculated photolysis frequency to UV intensity for each photolytic species on this day was calculated and used as a reference to correct the cloud effects. The J values on other days were then scaled to the measured UV radiation intensity. Because of a data communication problem, there were no UV radiation data available between 19 July and 23 July and on 2 August, so no J values could be calculated for those days. The model only calculated OH and HO2 during nighttime for these days when J values were assumed to be zero.
 The model was run using the FACSIMILE software (UES Software Inc) that employs a Gear method to solve the ordinary differential equations. The uncertainty in the parameterized RACM model is estimated to be about 35% for OH and 58% for HO2 with a 2σ confidence. These uncertainties are based on the combined uncertainties in the kinetic rate coefficients [Sander et al., 2003] and in the measured concentrations of species, as estimated using a Monte Carlo approach [Ren et al., 2003b].
3. Measurement Results
 During the campaign, typical summer weather (clear and sunny or sunny with some clouds) dominated in the Whiteface Mountain area. There were some periodic rain showers on the days of 18 July, 22 July, 23 July, 26 July, and 28 July. Daytime temperature was as high as 30°C, with an average of 18°C throughout the campaign. The average relative humidity throughout the campaign was 72%. The diurnal profiles of trace gases including O3, NOx, HONO, CO, SO2, HCHO, total VOC and meteorological parameters indicate the prevailing physical and chemical conditions in this forested environment (Figure 1).
 OH was measured on 28 days, from 10 July to 7 August. The measured OH shows expected diurnal variations with maxima around noon when the solar radiation reached the highest levels and with minima at night (Figure 2a). On average the maximum OH mixing ratio was about 0.11 pptv (∼2.6 × 106 cm−3) while daytime peak values varied from 0.05 pptv (∼1.2 × 106 cm−3) to 0.3 pptv (∼7.2 × 106 cm−3). The observed OH mixing ratios in this study are comparable to those in the PROPHET98 [Tan et al., 2001] and AEROBIC [Creasey et al., 2001] studies conducted at two different forested sites.
 Also shown in Figure 2 is the diurnal profile of O3 photolysis frequency, J(O1D), calculated by averaging all data into hourly bins. OH peaked around solar noon at about 1300 LT. In the early evening, J(O1D) decreased to zero, but the OH mixing ratio remained at a significant level of about 0.04 pptv (∼9 × 105 cm−3) on average, indicating that there were some nonphotolytic HOx sources such as the reactions of O3 with alkenes. Details about nighttime OH are discussed in section 3.2.
 HO2 was measured on 25 days, from 14 July to 7 August. Composite diurnal variations of HO2 and J(O1D) are shown in Figure 2b. Typical daytime HO2 peak mixing ratios varied from 5 pptv (∼1.2 × 108 cm−3) to 50 pptv (1.2 × 109 cm−3). On average, the diurnal profile shows a peak mixing ratio of 20 pptv (4.8 × 108 cm−3), occurring at midday (about 1300 LT). The observed HO2 mixing ratios in this study are much greater than those obtained in urban environments [George et al., 1999; Ren et al., 2003a; Holland et al., 2003], but are similar to those in relatively clean environments [Tan et al., 2001; Kanaya et al., 2001; Creasey et al., 2002]. HO2 mixing ratios as high as 2–3 pptv were observed at night. These significant amounts of HO2 acted as a source of OH. Nighttime HO2 measurements are discussed further in section 3.2.
 OH reactivity has a slight diurnal variation with higher values during daytime and lower values at night. The average OH reactivity was about 5.6 s−1, which is much lower than the values obtained in urban environments [Kovacs et al., 2003; Ren et al., 2003a; Sadanaga et al., 2004] and in the PROPHET2000 study [Di Carlo et al., 2004]. However, the OH reactivity at Whiteface Mountain is similar to that observed at a rural site at Rock Springs, PA, ∼10 km from the main campus of the Pennsylvania State University [Ren et al., 2005].
3.1. Relationship Between HOx and NO Concentrations
 Atmospheric OH and HO2 mixing ratios depend strongly on HOx production, P(HOx), and NO levels. The dependence of the measured OH mixing ratios upon P(HOx) and NO is shown in Figure 3a. For a fixed value of NO, OH increased as P(HOx) increased. OH slightly increased as NO increased when NO level was less than ∼0.1 ppbv. When NO level was more than 0.1 ppbv, OH slightly decreased as NO increased. Similar behavior has been observed in previous studies conducted in relatively clean environments [McKeen et al., 1997; Carslaw et al., 2001].
 The decrease in HO2 mixing ratios with increasing NO mixing ratios was observed (Figure 3b) at higher P(HOx). Similar to OH, HO2 shows roughly the expected qualitative variation with NO and P(HOx). For a fixed value of NO, HO2 increased as P(HOx) increased. At higher NO levels, the reaction of HO2 with NO was faster and HO2 mixing ratios were therefore lower.
3.2. Nighttime HOx Observations
 In the early evening both measured OH and HO2 mixing ratios dropped as photolytic HOx sources vanished. Elevated mixing ratios of nighttime radicals were frequently observed throughout the night during this campaign (Figure 2). In the late afternoon and early evening, the radical mixing ratios were about 0.04 pptv for OH and 5 pptv for HO2. Small but significant levels of OH (∼0.02–0.03 pptv) and HO2 (∼2 pptv) remained through midnight to 0400–0500 LT. At around 0500–0600 LT, OH reached a minimum level of 0.02 pptv while HO2 reached a minimum level of 1.5 pptv. With these nighttime OH and HO2 levels, HOx chemistry can play an important role in the nighttime oxidation processes in this forested environment.
 Isoprene decays in the evening can be an authentic indication for nighttime OH observations [Faloona et al., 2001; Stroud et al., 2002]. With the assumption that observed isoprene decay during nighttime was mainly dominated by its reaction with OH, OH concentration can be derived from the decay of isoprene. As discussed by Faloona et al.  and Hurst et al. , the derived OH level from the isoprene decay is only valid if other chemical reactions and dynamical processes are not important to the isoprene decay.
 Besides the OH reaction with isoprene, two possible chemical effects for the nighttime isoprene decay are the reactions with O3 and nitrate radical (NO3) because isoprene is known to react rapidly with these species [Atkinson and Arey, 2003]. The reaction with O3 was not important on the timescales of interest because the lifetime of isoprene with respect to ozonolysis was about 24.5 hours for a median evening ozone mixing ratio of 44 ppbv observed during this study. There were no NO3 measurements made during this campaign but model estimates of NO3 indicate that the evening median value of NO3 was only 0.61 pptv, corresponding to an isoprene lifetime of ∼22.4 hours. By comparison, the lifetime with respect to the median OH concentration of 106 molecules cm−3 was only 2 hours and 40 min. Therefore the overnight isoprene decays were most likely not controlled by O3 and NO3 oxidation but by the OH reaction.
 Another possibility for the nighttime isoprene decay that is not dependent on chemistry is vertical and/or horizontal transport as shown in the model result for the PROPHET98 study [Sillman et al., 2002]. However, a study by Hurst et al.  indicates that both vertical and horizontal transport might not be important for nighttime isoprene decay and the contribution of vertical mixing has large uncertainty.
 Other evidence that supports the OH removal of isoprene is the measured isoprene variations during the nights of 5–6 August and 6–7 August, which were significantly different from other nights in the last week of the campaign. In the early evening of 5–6 August, isoprene slowly decreased and after midnight remained at a sustained high level of about 0.3 ppbv until early morning. In the early evening of 6–7 August, a slow isoprene decay was also observed in the early evening. Just before midnight, isoprene started to increase from 0.1 to 0.4 ppbv till the early morning of 7 August. The slow decays before midnight on these two nights are consistent with the observed low OH mixing ratios that were near the OH detection limit and were much lower than what was observed during other nights in the last week of the campaign. Although meteorological conditions on these two nights were comparable to other nights in the last week, the chemical composition was different. On the two nights of 5–6 August and 6–7 August, much cleaner air masses were encountered than the rest nights in the last week, with lower levels of O3 and high concentrations of isoprene compared to other nights. Thus little isoprene reacted away.
 There is good agreement between measured and derived OH in this study, as depicted in Figure 4. Isoprene was measured postcampaign for another 2 weeks. From the measured nighttime OH decay rates, the average OH mixing ratio during those nights was 0.045 ± 0.015 pptv, which is comparable to the nighttime OH observations during this campaign. Similar results were obtained in the PROPHET98 study [Faloona et al., 2001; Hurst et al., 2001]. The good agreement indicates that the decay of isoprene observed at the site was mainly due to the OH reaction with isoprene.
4. Model Comparison
 The overlap of all necessary ancillary measurements allowed model calculations to be made between 1 and 7 August for the complete RACM model and between 12 July and 7 August for the parameterized RACM model. Observed OH and HO2 mixing ratios were then compared to the steady state box model calculations.
4.1. OH Comparison
 The model calculations show similar diurnal and day-to-day trends as the measurements (Figure 5 and Figure 6a), with maxima at midday and minima during nighttime. The agreement between the measurements and model calculations is generally good. Unlike most of the study, the parameterized RACM model significantly overpredicted the daytime OH on 24, 26, 27, and 29 July. Because no instrument effects have been found that can explain the discrepancy and OH sinks are constrained in the model, the likely reason is that OH sources were biased in the parameterized model VOC inputs for these days.
 The agreement of the model calculation results between the complete RACM and the parameterized RACM models is well within uncertainties, as can be seen from the model results in the last week of the campaign (Figures 5 and 7). This agreement indicates that the parameterized RACM method works well for the steady state model calculation of OH and HO2 radicals.
 The measured daytime OH and its variation are generally captured by the model when constrained to the total OH reactivity (Figure 6a). Both the measurements and the model calculations show that the peak occurred at about local solar noon. In the parameterized RACM model, the median observed-to-modeled OH ratio is about 0.84, which is within the combined 1σ uncertainty (∼24%) of the measurements and the model calculations. In the complete RACM model constrained by actual VOC measurements, the agreement between the measured and modeled daytime OH is good, with a median measured-to-modeled OH ratio of 0.82. The linear fit in the scatterplot of the measured OH versus modeled OH in the complete RACM model gives a slope of 0.74 with a correlation coefficient, r2, of 0.60. The reasonably good agreement in this study contrasts with the poor agreement in the PROPHET98 study, where the daytime OH was significantly underpredicted by a factor of 2.7 [Tan et al., 2001].
4.2. HO2 Comparison
 The measured and modeled HO2 exhibits similar diurnal and day-to-day profiles (Figures 7 and 6b) as OH does. The model overestimated the daytime HO2 on 24 and 27 July. Both the complete and parameterized RACM mechanisms underpredicted the daytime HO2 from 3 to 5 August. Similar to the modeled OH, good HO2 agreement between both model mechanisms was also achieved. The measured-to-modeled daytime HO2 ratio is about 0.80 (Figure 6b) for the parameterized model and is 1.21 (Figure 6d) for the complete RACM model constrained by the actual VOC measurements. Both the complete RACM and parameterized RACM overpredicted nighttime HO2; the parameterized RACM was higher by a factor of 2.3 and the complete RACM was higher by a factor of 2.0 on average. The linear fit in the scatterplot of the measured HO2 versus modeled HO2 in the complete RACM model gives a slope of 1.04 with a correlation coefficient, r2, of 0.71. The fairly good agreement between measured and modeled HO2 is consistent with the PROPHET98 study, where the measured-to-modeled HO2 ratio is 0.85 during daytime.
4.3. Observed and Modeled HO2/OH Ratios
 The HO2/OH ratio reflects the interconversion between OH and HO2. It is a very useful parameter for testing of the understanding of the HOx photochemistry because the photochemical equilibrium between OH and HO2 is closely tied to the interconversion of NO to NO2 in the troposphere [Stevens et al., 1997]. Measured HO2/OH ratios ranged between 40 and 400 in this forested environment and are greater than those in polluted urban environments [George et al., 1999; Martinez et al., 2003; Ren et al., 2003a, 2005; Mihelcic et al., 2003; Emmerson et al., 2005] but are comparable to those in other clean environments [Stevens et al., 1997; Carslaw et al., 2001; Tan et al., 2001; Kanaya et al., 2001; Sommariva et al., 2004]. As shown in Figure 8, both the measured and modeled HO2/OH ratios decrease with increasing NO. This decrease occurs because NO shifts HOx into OH by reacting with HO2. The relatively good agreement between the observed and modeled HO2/OH ratios is inconsistent with the results in PROPHET98, in which the measured HO2/OH ratios were overpredicted by a factor of 3.7 [Tan et al., 2001] because of the significant underestimation of OH. Similar overestimation of HO/OH ratios was also obtained in the PROPHET2000 study.
4.4. Measured and Calculated OH Reactivity
 In the troposphere many gas phase species such as hydrocarbons, CO, NO2, and other species are mainly removed through their reactions with OH. The sum of the rates of these reactions divided by the OH concentration is the OH reactivity. In the last week of the campaign, the speciated VOCs were measured, so OH reactivity can be calculated from the concentrations of the measured species and their rate coefficients in the reactions with OH. Comparison between the measured and the calculated OH reactivity can test the assumption that most OH reactants have been measured and can uncover if there is any missing OH reactivity.
 A good relationship between measured and calculated OH reactivity (Figure 9a) was obtained, in contrast to the PROPHET2000 study, where significant OH reactivity was missing [Di Carlo et al., 2004]. The average difference between measured and calculated OH reactivity is 0.00 ± 0.74 s−1 and is within the combined uncertainty of the measurements and calculations. There is no significant temperature dependence of the difference between the measured and calculated OH reactivity at this forested site (Figure 9b). This observation is inconsistent with what was found during PROPHET2000 in which the missing OH reactivity was temperature-dependent and increased as temperature increased.
 The behavior of OH reactivity was not the same at Whiteface Mountain as it was at the PROPHET site. This raises questions and has potential implications for the distribution of the biogenic VOCs. The measured isoprene concentrations in this study were relatively low, typically less than 1 ppbv during daytime, which is similar to the measurements obtained in 1994 and 1995 at the same site [Gong and Demerjian, 1997]. These concentrations were much lower than the daytime isoprene levels at the PROPHET site, which were usually greater than 1 ppbv and had peak values up to 10 ppbv. At Whiteface Mountain, isoprene made up a median fraction of 14% of the OH reactivity budget. This fraction is much smaller than that in the PROPHET2000 study, in which isoprene accounted for almost half (49%) of the measured OH reactivity [Di Carlo et al., 2004].
 The reason for the different biogenic VOC levels at the two sites is not clear and needs further investigation. One possible reason could be spatial variance in biogenic VOC emissions at these two sites [Guenther, 1997] and that trees in more polluted environments do not put out as much isoprene and terpenes [Owen et al., 2003]. There is also a slight difference in the tree types at the two sites. The forest surrounding the sampling location at the Whiteface Mountain is identified as transition forest and is composed of a mix of hardwood and conifer species including white and yellow birch, sugar maple, beach and some red spruce and balsam fir. The forest surrounding the PROPHET site is also characterized as mixed or transition forest but with northern hardwood, mixed aspen, bog conifers, pines and red oaks [Carroll et al., 2001].
4.5. Balance Between OH Production and Loss Rates
 Further testing of our understanding of the sources and sinks of HOx can be accomplished by comparison of the total OH loss rates (molecules cm−3 s−1) with the expected OH production rates. Because the OH lifetime is about 0.2 s, OH production and loss should balance for time periods greater than a second. The total OH loss rates were determined by multiplying the total OH reactivity measurements from TOHLM by the measured OH concentrations from GTHOS. OH production was calculated from the HO2 reactions with NO and O3, in which the measured HO2 concentrations were used, the UV-B photolysis of O3 followed by the O(1D) + H2O reaction, and the photolysis of HONO and H2O2. Because TOHLM measures the sum of OH losses, a comparison of OH production and loss tests for the presence of additional OH sources.
 The measured OH loss rate and the calculated OH production rate show a similar diurnal variation (Figure 10). The agreement between the OH production and loss is good in the early morning and in the late afternoon. The OH production is greater than OH loss during the time period between 0900 and 1600 LT, while in the evening and at night the OH loss is significantly greater than the OH production. The difference between the daytime OH production and OH loss is marginally significant in terms of the 1σ uncertainties. However, the difference becomes statistically significant in the evening and at night, when the median OH loss is about 2–3 × 106 cm−3 s−1 while the OH production rate is about 1.6 times less. In order to balance nighttime OH production and loss, a significant unknown OH source of about 1.2 × 106 cm−3 s−1 (equal to two thirds of the known OH production) would be required. This missing OH production rate is not necessarily primary OH production, but could be OH production in the OH-HO2 cycling, e.g., an additional species which reacts with HO2 to produce OH like NO does [Faloona et al., 2001].
4.6. HOx Budget
 It is also interesting to investigate the HOx production and loss rate because the balance between HOx production and loss indicates a good understanding of HOx sources and sinks. The HOx production consists of the production from the following processes: the O3 photolysis followed by the O(1D) + H2O reaction, the HONO photolysis subtracted by the HONO formation from the OH + NO reaction to reflect the net OH production, the HCHO photolysis (the radical-produced pathway only), and the ozonolysis of alkenes. HOx loss includes the OH reaction with NO2, and the reactions among OH, HO2 and RO2. RO2 was not measured and was calculated from the parameterized RACM.
 The calculated diurnal profiles of HOx production and loss rates from different processes are shown in Figure 11. In the early morning and later afternoon, the most important HOx production was the photolysis of HONO. This is due to relatively high HONO levels in this environment, with a morning peak of 130 pptv and an average value of about 100 pptv in the rest of day (Figure 1d). Similar HONO levels and its importance to HOx production were also found at a forested site in Germany, where a contribution of 33% to the primary OH production and some evidence for a large unexplained HONO source were found [Kleffmann et al., 2005]. At midday the O3 photolysis became a major contributor to the HOx production, comparable to HONO photolysis. In the evening and at night, HOx was mostly from the ozonolysis of alkenes. The photolysis of HCHO was not important because of its low level in this environment with an average mixing ratio of 1.3 ppbv for the whole campaign. On average, the total HOx production rate was 16.3 ppbv per day, of which 34% was from the photolysis of HONO, 30% from the photolysis of O3, 27% from the O3 reactions with alkenes, and 9% from the photolysis of HCHO.
 For the calculated HOx loss, the clearly dominant process was the self-reaction of HO2 throughout the day, which accounted for 56% of the HOx loss. The reactions between HO2 and RO2 accounted for 32% of the HOx loss and the OH reaction with NO2 only accounted for 6% because of low levels of NO2 in this environment.
 The measurements of OH, HO2, OH reactivity, and other gas-phase components during the PMTACS–NY summer 2002 intensive at Whiteface Mountain provide a good opportunity to test our understanding of HOx chemistry in this forested area. We can draw several conclusions from this study.
 First, the constrained steady state model can reasonably predict the measured OH and HO2 in this environment. Fairly good agreement between the measured and modeled OH during daytime in this study is inconsistent with the poor agreement in the PROPHET98 and AEROBIC studies, in which the measured OH was significantly underpredicted. This suggests that the behavior of OH and HO2 in this environment was more like the HOx behavior in semipolluted rural or urban environments, rather than the HOx behavior in clean forested sites.
 Second, the good agreement between the measured and calculated OH reactivity at this forested site contrasts with what was found during the PROPHET2000 campaign where the difference between the measured and calculated OH reactivity was significant and temperature-dependent, indicating missing OH reactivity. No significant temperature dependence of the difference between the measured and calculated OH reactivity was found in this study. The fact that the measured isoprene makes up a small fraction of the OH reactivity budget at Whiteface Mountain suggests that there is a significant difference in the biogenic VOC emissions between these two sites.
 Third, the HOx budget analysis in this study indicates that during daytime HOx is primarily from the HONO photolysis and the O3 photolysis followed by the O(1D) reaction with H2O. The importance of the HONO photolysis to the HOx production is consistent with the results found at a forested site in Germany [Kleffmann et al., 2005]. Nighttime HOx is mainly from the O3 + alkenes reactions. As expected, the main HOx loss is the self reaction of HO2 in this environment due to low levels of NO2 in this environment. To balance the measured OH loss rate at night, an additional OH production of about 1.2 × 106 cm−3 s−1 is required, which is about equal to two thirds of the current known OH production rate.
 Finally, low but significant mixing ratios of OH and HO2 persisted into the early evening and were frequently observed at night, suggesting that HOx chemistry plays an important role in the nighttime oxidation processes in this forested area. This observation is consistent with the observations of nighttime OH and HO2 in some previous studies in various environments [Faloona et al., 2001; Salisbury et al., 2001; Kanaya et al., 1999, 2002; Creasey et al., 2002; Martinez et al., 2003; Ren et al., 2003a, 2005]. In this study, the observed nighttime OH levels agree with the OH levels derived from the isoprene decays in the evening, which provides evidence that the decay of isoprene observed at the site was mainly due to the OH reaction with isoprene.
 We thank other participants in the PMTACS–NY summer 2002 field campaign for the use of their data in the model. This work was supported by NSF (ATM–9974335 and ATM–0209972), the New York State Energy Research and Development Authority (NYSERDA) (contract 4918ERTERES99), the U.S. Environmental Protection Agency (EPA) (cooperative agreement R828060010), and New York State Department of Environmental Conservation (NYS DEC) (contract C004210). Although the research described in this article has been funded in part by the U.S. Environmental Protection Agency, it has not been subjected to the Agency's required peer and policy review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred.