Driving Factors for Isoprene-Surface Ozone and Carbon Dioxide Chemistry Over an Oil Palm plantation in Malaysia

Background: Biogenic Volatile Organic Compounds (BVOCs) such as isoprene (C 5 H 8 ) are ozone (O 3 ) precursors that can be emitted at signicant concentrations from the oil palm tree (Elaeis guineensis). Reactions involving BVOCs can lead to increased levels of surface O 3 which can signi ﬁ cantly impair air quality and cause crop damage. This study focuses on the link between isoprene, surface O 3 and carbon dioxide (CO 2 ) over an oil palm plantation and the effects of meteorological factors such as temperature and irradiance on the gas concentrations. The mixing ratios of isoprene, surface O 3 and CO 2 were measured using a portable gas chromatograph with a photoionization detector called ‘iDirac’, an EcoTech O 3 analyser, and LI-COR, respectively. Atmospheric models were used for measured selected gases estimation and validation purposes. Results: Results showed that isoprene and surface O 3 had maximum daytime mixing ratios of ~25 ppb and ~57 ppb, respectively. CO 2 mixing ratios were high during the night compared to the day, with a maximum night-time ratio of ~883 ppm. It is also showed that suppression of isoprene emissions from plants by high CO 2 concentrations during the night was due to the reduction of dimethylallyl diphosphate (DMADP) in the leaf cells. The meteorological parameters temperature and light intensity were signicantly correlated with isoprene and surface O 3 , with r 2 =0.91 and p<0.01 and r 2 =0.87 and p<0.01, respectively. The Model of Emissions of Gases and Aerosol from Nature (MEGAN) estimated emission rates of isoprene from the oil palm plantation are in the range ~5000 to ~7000 μgm -2 h -1 which are higher than previous studies on pristine forest. This information together with the the in-situ measurement information on isoprene emission ux is then fed into the WRF-CMAQ atmospheric chemistry model to study the effect of oil palm plantation expansion over the years (2000 – 2016) on the local atmospheric chemistry. Conclusions:

Peninsula Malaysia due to the lack of measurement data [26; 27]. Therefore, the availability of the in situ observation data in the current study has made possible the in uence of isoprene emission from oil palm on the atmospheric composition over the region using models. Weather Research and Forecasting model coupled with Community Multiscale Air Quality Modelling (WRF-CMAQ) is applied to study the compositional change of atmospheric isoprene and O 3 over years of rapid oil palm plantation expansion (2000 to 2016) as well as the transport of these atmospheric pollutants to the downwind region during the northeast monsoon season.
In this study, to the best of our knowledge, we conducted the rst continuous in situ measurements of isoprene concentrations in a palm oil (Elaeis guineensis) plantation, certainly in Peninsular Malaysia. The main objective was to investigate the isoprene and surface O 3 mixing ratios at 4 m above the canopy levels as well as their relationship with meteorological factors such as temperature, irradiance (light intensity), wind speed and direction, and relative humidity on isoprene. In addition, CO 2 mixing ratios will also be discussed to investigate their implications on isoprene emissions. The implications are studied using the chemistry transport model WRF-CMAQ to assess the modelling system performance in the prediction of O 3 levels and isoprene concentrations and emissions over the palm oil plantation. In Sect. 2, the experimental methodology is described. In Sect. 3, the measurements are presented and analysed alongside the modelling results from WRF-CMAQ. Finally, conclusions are discussed in Sect. 4.

Study site description
In situ continuous measurements of isoprene, surface O 3 and carbon dioxide (CO 2 ) were made at an oil palm plantation located on the east coast of Peninsular Malaysia (3°26'17.8692" N, 103°23'23.8056" E) in the state of Pahang as shown in Figure 1. The oil palm plantation area is about 18,000 ha in extent.
The sampling location is at the town of Pekan, approximately 30 km south of Kuantan, the capital city of Pahang. Heavy industries (oil and gas facilities, steel plants, and manufacturing industries) are common in the area. Recently, the introduction of Malaysia's rst Special Economic Zone at Kuantan has boosted the local economy including through tourism. This increased industrial activity has made Kuantan a primary source for anthropogenic VOCs and other ozone precursors such as NO x . These emissions have the potential to in uence the levels of surface O 3 and aerosols over the oil palm area. During the measurement period, isoprene, surface O 3 and CO 2 mixing ratios were measured at the eddy covariance (EC) tower (30 m height), owned by the Malaysian Palm Oil Board (MPOB). Each sample inlet was 10 m above the ground. Isoprene was measured during a 38-day period (6th December 2016 to 13th January 2017), while surface O 3 and CO 2 were measured until the 31st December 2016 due to the reinstallation of the O 3 and CO 2 analysers. The atmosphere immediately above the site was sampled using a 10 m 1/4" ID Te on sample line (5 m above the oil palm canopy). Air was drawn from the inlet to the isoprene and O 3 analysers. The CO 2 measurements were made using a separate inlet system.
Surface O 3 was measured using an EcoTech (Australia) model Serinus 10 O 3 analyser. The EcoTech was well-maintained and was calibrated immediately prior to the measurement period. The calibration was based on a 5-point standard technique with low to high concentrations from 0.01 to 100 ppb (parts per billion by volume). The detection limit was estimated to be 0.05 ppb. The calibration procedure is described in [28].
Isoprene was measured using a new, portable Gas Chromatography with Photoionization detector (GC-PID) named iDirac, which is a robust, autonomous instrument with good sensitivity to isoprene in ambient conditions as described details in [29]. A dual-bed adsorbent trap (1 mg Carboxen-1016 60/80 mesh followed by 1 mg Carboxen-569 20/45 mesh) is used to pre-concentrate each sample. The trap is then heated and the gaseous constituents separated on a 10 m, 0.18 mm ID capillary column (Restek MXT 502.2) prior to measurement in the PID. Nitrogen is used as the carrier gas. The instrument is controlled by an Arduino and Raspberry Pi with Wi-Fi built in. As the instrument is designed to run unattended, the Raspberry Pi interface software runs chromatograms according to a pre-determined sequence of samples bracketed by calibration, blank and sample chromatograms. The sequence includes frequent calibrations using the same volume as the samples (20 mL) for correction of instrument sensitivity drift and precision determination. The unit can be used without a laptop in the eld with two sample inlets and one calibration port for the calibration gas injection onto the PID. Solar power (12 V) was used to power up the iDirac throughout the campaign. iDirac has a power consumption of ~10 W. It typically has a precision of <10%, a detection limit of ~20 ppt, and a sampling time of ve minutes which depends on the sampling sequence chosen for a particular experiment [29]. CO 2 mixing ratios were measured using an open-path infrared gas analyser (LI7500, LI-COR, USA). These measurements were made as part of a study to estimate CO 2 uxes over the oil palm plantation using EC [30]. A three-dimensional sonic anemometer (CSAT3, Campbell Scienti c, USA) was also installed at the EC tower. The analyser and anemometer were factory-calibrated before use. The open-path analyser was calibrated in March 2016 to ensure accurate measurements. Nitrogen gas and drierite/soda lime were used to zero the CO 2 and H 2 O sensors. Standard gas with 413 ppm CO 2 was used to analyse and monitor the CO 2 readings, and a dew point generator was used to span the H 2 O range.

Meteorological
The meteorological data over the sampling site were collected using a "Biomet" system data logger (Xlite 9210, Sutron, USA). Air temperature and relative humidity sensors (HMP155, Vaisala, Netherlands) and quantum sensors (LI-190SL, LI-COR, USA) to measure photosynthetically active radiation (PAR) were installed at a height of 30.65 m. The wind speed and direction were measured by a three-dimensional sonic anemometer (CSAT3, Campbell Scienti c, USA). A tipping-bucket rain gauge (TR525, Texas Instruments Inc., USA) was employed to collect the 30-min accumulative precipitation.

Secondary Data
Finally, hourly averages of ozone, carbon monoxide (CO), nitrogen oxide (NO), nitrogen oxide (NO x ) and nitrogen dioxide (NO 2 ) concentrations were measured at stations managed by Alam Sekitar Sdn Bhd (ASMA), a company which measures the air quality status on behalf of the Malaysian DOE in Kuantan.
The two stations were located in Kuantan, S1 (N03 o 49.138, E103 o 17.817) and S2 (N03 o 57.726, E103 o 22.955). S1 is in a residential area, while S2 is in an industrial area which is a busy area with a high volume of motor vehicles moving in and out of the area. use categories. The gas-phase chemistry and aerosol mechanisms adopted in the model were CB05e51 and AE6 with aqueous chemistry. Dry deposition and the photolysis of gases were parameterized based on [31]. The anthropogenic emissions were re-gridded for the model domain from the MIX dataset available at 0.25° x 0.25° for the year 2010 [32]. This was not a major season for biomass burning, hence the model does not include the data for the relevant emissions. Biogenic emissions of trace species were calculated online using the Model of Emission of Gases and Aerosols from Nature (MEGAN) model version 2.1 [25], with several important updates: the incorporation of changes in land use (Ramankutty and Foley, 1999), the explicit consideration of oil palm emissions following the algorithm of [33], and the incorporation of corrections in ECMWF solar radiation elds based on surface solar radiation data [34; 35].
The input data of MEGAN was updated to the 8-day LAI index [36] and present-day plant functional types (PFT) [37]. In this paper, we have updated the emission factors for isoprene at the regions covered with oil palm with the measured results. Therefore, we have simulated two model cases with updated emission factors according to the distribution of oil palm over the Pekan region in the years (a) 2000 and (b) 2016 as shown in Figure 3. The increase of oil palm occurred across the Pekan region over these 10 years.
According to our results (discussed in Section 3.7), the emission factor for isoprene is updated to 6000 ug m -2 h -1 as the MEGAN input for oil palm regions. After the emission-response calculation that accounts for the light and temperature responses, the MEGAN output increased the isoprene over central Pekan comparing the years 2016 and 2000 as shown in Figure 3(c). The model simulates December 2016 and January 2017, in line with the measurement campaign. It runs two separate cases with oil palm coverage of the years 2000 and 2016. The conversion of high isoprene-emitting agricultural land to oil palm in the north is the major cause of isoprene reduction here, while the reduction in the south is related to the introduction of oil palm due to forest clearing.

Meteorological conditions and backward trajectories
The measurement period, December 2016 and January 2017, was during the North East Monsoon (NEM) when the prevailing wind ow is northeasterly and strong. Hourly average meteorological data consisting of temperature, relative humidity, wind (speed and direction) and irradiance were used to evaluate the weather conditions in this period. A summary of the meteorological conditions is given in Figure 4. The minima for temperature and relative humidity (RH) were 23.3 °C and 43% with average values of 27.5 °C and 82%, respectively. The maximum temperature of 32.2 °C occurred on the 13th December 2014. In January 2017, rain fell nearly continuously from the 11th to 13th January 2017 and this led to lower temperatures. The minimum temperatures recorded were during these days and were in the range 25.2 to 25.5 °C. The maximum temperature recorded in January 2017 was 29.2 °C on the 10th January.
The daily wind speed varied between 0.3 ms -1 and 8.8 ms -1 with a median value of 2.7 ms -1 during December 2016, while in January 2017 the respective values were 0.4 ms -1 and 9.3 ms -1 with a median value of 3.4 ms -1 . The wind roses for these months are shown in Figure 4. The wind directions were mainly blowing from the regions around Kuantan and Pekan. The highest daily maximum value of Photo Active Radiation (PAR) was 801 mMol m -2 s -1 on the 13 th December 2016 when the highest temperature was also recorded. The minimum daily PAR, 300 W, was recorded during the rain event on the 11 th to 12 th January 2017. The travel pathways of the 3-day (100 m) backward trajectories (BTs) at Pekan palm oil plantation were plotted ( Figure 5). As an input of the trajectory model, the data set was downloaded from the National Oceanic and Atmospheric Administration (NOAA) website (link: ftp://arlftp.arlhq.noaa.gov/pub/archives/reanalysis). The date of the BTs was the 12th December 2016. This day was selected as it was the day on which the highest mixing ratios were observed for isoprene. At the measurement site, the BTs showed transport from the northeast (NE) of Peninsular Malaysia. The mixing height was about 100 m on the 12th December 2016. The results of the trajectories can be interpreted with the elevation of surface O 3 as well as the change to the height of the boundary layer.
During the in situ measurements at the site, the mixing ratios of surface O 3 were higher at the Pekan palm oil site. The nearby city Kuantan experienced high concentrations of O 3 precursors such as NO x and CO (this will be discussed in the next Section) on this day. The noted shift of air mass in this region may in part in uence the anthropogenic impact on the formation of surface O 3 that might move with air masses around the palm site. We can conclude that the isoprene may react with the NO x and VOCs emitted from Kuantan.  This observation of high mixing ratios of isoprene during the day and low mixing ratios at night is consistent with many previous studies [38; 39; 40: 41]. The morning increase results principally from the strong in uence of PAR and temperature on the emissions from the leaves into a stable planetary boundary layer (PBL). The rate of the afternoon decrease is additionally affected by the loss to reaction with the hydroxyl radical (timescale ~1 hour), the main chemical loss process for isoprene, and transport into the lower troposphere. The rates of all these processes vary signi cantly from day to day.
A similar diurnal cycle was found in the measured ozone mixing ratios, with a maximum in the day and a minimum at night ( Figure 6). Typical daytime peaks are ~40-50 ppb, with the highest hourly value of 57 ppb recorded on the 12th December 2016. These values are under the Malaysian Air Quality Standard for 1 hour and 8 hours which are 180 and 100 ppb, respectively. Surface O 3 values are determined by three main processes: (i) mixing in of ozone-rich tropospheric air into the PBL which is responsible for the morning increase in surface ozone; (ii) photochemical reactions involving ozone whose production will be enhanced by the presence of isoprene in NO x -rich air [9]; and (iii) deposition of ozone onto the oil palm plantation.
A strong diurnal cycle of CO 2 was measured throughout December 2016 ( Figure 6). Maximum mixing ratios were observed during the night and minimum values during the daytime with typical values of 810 ppm and ~450 ppm, respectively. Plant photosynthesis and respiration are the main reasons for these variations. Plants produce CO 2 all the time through respiration, but during the day (when the light intensity is high) plants use CO 2 for photosynthesis, and x CO 2 into other molecules, resulting in more O 2 than CO 2 . At night, the CO 2 increases because the plants are giving off CO 2 in respiration and not photosynthesising. It can take until late morning before CO 2 removal by photosynthesis is observable during the day.
Physical factors effects on isoprene, surface O 3 and CO 2 .

Effect on isoprene
It has been observed that temperature and light intensity in uence the isoprene emissions from plants [42; 43; 44; 45]. The light-dependent regulation of isoprene emissions is due to the production of isoprene in the plant photosynthesis process, therefore the photosynthetic active radiation during the day leads to higher daytime isoprene [46]. In this study, the irradiance term is similar to light intensity, thus irradiance in Watt per meter square (Wm -2 ) will be used throughout. Figure 6 shows that increases in temperature and irradiance led to increases in isoprene mixing ratios. Strong correlations between temperature and irradiance with isoprene were observed with r 2 =0.91, p<0.01 and r 2 =0.82, p<0.01, respectively. The trends in isoprene, temperature and irradiance showed that the maximum peaks were observed during the middle of the day, from1100 to 1530. This was also observed by [41] for isoprene mixing ratios over Calabozo, Venezuela (Savanna site). They measured maximum mixing ratios of about 3.1 ppb between 1200 and 1500 local time.
The isoprene synthesis and light dependence relationship can be explained by the synthesis of isoprene in the stomata. First observation of the light dependence of isoprene emissions was by [42], indicated by the rapid appearance of the 13C label from 13C-labelled CO 2 in the isoprene signal [43; 47]. An observation by [47] showed isolated chloroplasts are capable of isoprene emission which provided conclusive evidence of the functional interdependence between photosynthesis and isoprene emission.

Effect on surface O 3
Previous studies have shown that surface O 3 mixing ratios in ambient air are controlled by temperature [17; 18; 16]. Increases in temperature will increase the surface O 3 mixing ratio in the presence of high levels of NO x . Our observed variations in the daily maximum temperature and the daily maximum O 3 mixing ratios are depicted in Figure 6. Overall, the average daily maximum temperatures and surface O 3 measurements for Pekan palm station show increments, especially during the hours 1100 to 1630, then gradual decreases after 1630. This pro le of surface O 3 was consistent throughout December 2016. The highest temperature recorded was 33 °C on the 13th December 2016 when the highest surface O 3 also was observed. A similar effect of temperature on surface O 3 was found over Kuala Terengganu, also on the east coast, ~200 km from Pekan was seen in the study by [48]. Other than temperature, irradiance and wind speed were also found to play a signi cant role in in uencing O 3 mixing ratios [48].
The daily maximum temperature increased steadily from 0900 due to the increase of light intensity during the peak sun hours, where the irradiance that a particular location would receive if the sun were shining at its might i.e. as strongly as it can, as in no cloud cover. The increase in irradiance causes an increase of temperature, which in turn escalates the O 3 mixing ratio maximum value. Similar to the temperature effect, the daily maximum O 3 mixing ratio tended to follow the intensity of irradiance, causing higher levels during daytime and lower levels at night. The correlation between daily maximum temperature and daily maximum solar radiation was r 2 =0.91 (p<0.01). This shows that these two parameters are strongly correlated with surface O 3 production.
Effect on surface CO 2 Light and temperature are two of the most in uential parameters on photosynthesis activity. CO 2 emissions are linked to the photosynthesis activity of plants. Plants consume more CO 2 during the daytime during photosynthesis and release more during respiration at night. In this study, in contrast to the meteorological effects on isoprene and surface O 3 , temperature and solar radiation showed negative correlations with the CO 2 mixing ratios as shown in Figure 6. The daily CO 2 maxima were in the range 580 to 780 ppm during the day while the minims were in the range 440 to 468 ppm during the night. This pro le of surface CO 2 was consistent throughout December 2016.
These results imply that photosynthesis activity controls the mixing ratios of CO 2 . During the day (when light and temperature are high), CO 2 is consumed to produce O 2 . High light intensity will lead to high temperatures usually during the daytime and results in stomata closing, which is also associated with high vapour pressure de cit (VPD). This means the effects of temperature and light intensity are reduced at low CO 2 mixing ratios. According to [49], blue light (BL) has signal functions regulating many processes in chloroplast and stomatal opening in the leaf that may, in turn, affect isoprene emission. The BL also will be adsorbed by the stomata and in uence CO 2 diffusion in the leaf. Thus, CO 2 and isoprene are believed to be in uenced by the temperature and light intensity (especially the BL) over the observation site.
The relationships between isoprene-surface O 3 and isoprene-CO 2 .

Isoprene-surface O 3
As mentioned, the photochemistry of isoprene can dominate the photochemical production of O 3 [50; 51; 52; 53:54]. Both the production of isoprene and surface O 3 were observed during the middle of the day throughout the measurement period. It is clearly shown that isoprene production in uenced the surface O 3 formation during this time (1100 to 1400) (see Figure 6). Secondary products of the isoprene and OH reaction interact with NO x from anthropogenic sources to form surface O 3 . However, we were expecting the NO x mixing ratios over the palm oil region to be much lower compared to those over the urban area Kuantan (this will be discussed in the next Section).
The movement of vehicles (oil palm collection) is also an expected source of NO x . According to [55], O 3 production resulted from the oxidation of isoprene (C 5 H 8 + OH →→ RO2 + NO →→ O 3 ). Higher levels of peroxyl radicals (RO2) during isoprene oxidation will enhance O 3 production [55]. Therefore, if NO x levels were high over the observed area, the production of ozone due to the oxidation of isoprene is expected to be higher than in areas with low NO x emissions. [55] also suggested from their model calculation that the oxidation of isoprene also contributed to the enhancement of levels of carbonyls (such as formaldehyde and acetaldehyde). The formation of formaldehyde and acetaldehyde from isoprene oxidation will lead to the enhancement of HO 2 radicals. The combination of increased carbonyls and RO2 in the atmosphere suggested the high observations of O 3 , especially during the middle of the day.
During the evening and night, isoprene removal is much faster (isoprene has a lifetime of ~3 hours) compared to the surface O 3 removal process. These chemical removals can be explained by the lifetime of both gases, where surface O 3 has longer than isoprene (a lifetime ~3 hours) in the atmosphere (depending on the NO x emissions). In this study, strong daytime correlations between isoprene and surface O 3 were observed with r 2 =0.81 (p<0.01), while weak correlations were observed during the evening and night with r 2 =0.43 (p>0.01). The correlations between both gases were believed to be in uenced by atmospheric lifetime and meteorological factors.

Isoprene-CO 2
Isoprene and CO 2 mixing ratios measured above the canopy are shown in Figure 6. The diurnal cycle for both measured gases shows night and daytime variability for CO 2 and isoprene. The maximum CO 2 levels during the night are believed to be linked to plant respiration, while during daytime CO 2 is used in the photosynthesis process. Average daily mixing ratios for isoprene and CO 2 were 8.9 and 527.3 ppm and 0.2 and 724.1 ppm during the daytime and night, respectively. We believe that in a peat soil ecosystem there will be higher releases of CO 2 , mainly from the peat, but during the daytime, CO 2 is taken up by the oil palm leading to decreasing CO 2 mixing ratios.
There is another unknown factor which drives the isoprene emission CO 2 response. According to previous studies, high CO 2 concentrations inhibit isoprene release into the atmosphere [45]. This has been hypothesized to re ect the inhibition of the activity of isoprene synthase (IspS) or another enzyme of the methylerythritol 4-phosphate (MEP) pathway [56]. This also supported by the evidence from a natural CO 2 spring which showed that decreasing of isoprene emissions with increasing CO 2 concentration could be linked to a reduction in IspS activity [57].
A study by [58] showed that across CO 2 concentrations from 240 to 520 ppmv isoprene emissions from Eucalyptus globulus were enhanced at the lowest CO 2 concentration. The study showed that the in uence [58;59]. Process-based leaf-level isoprene production models were demonstrated by [59] and [58], also supporting the observation of Wilkinson et al. (2009) that enhanced CO 2 levels will depress isoprene production. [60] showed that Triose Phosphate Utilization (TPU) limitation plays a key role in the suppression of isoprene emission from plants under high CO 2 concentrations. Under conditions of high CO 2 where TPU limits photosynthesis, both ATP and NADPH production are reduced, which under the energy hypothesis would reduce DMADP production [60]. Thus, during the night the reduction of DMADP in the Elaeis guineensis cell reduced the production of isoprene.

Background information of local in uenced from nearby city.
Since we did not measure in situ NO x at the oil palm site, we used secondary 2017 data for surface O 3 , NO, NO 2 , NO x and CO from the continues air quality monitoring station (CAQMS) from the Department of Environment (DOE) sites in residential (S1) and industrial (S2) areas in Kuantan (Figure 1). These provide some information on the regional background composition. Two sites are available, one in an industrial No other data over the region nearby Pekan are available and so we assume that they are representative of Kuantan as a whole and indicative of the atmospheric composition at Pekan when the wind is from Kuantan (see Figure 5). Table 1 shows a summary of monthly average mixing ratios for O 3 , CO, NO x , NO 2 and NO over the Kuantan city region during December 2017. Table 1 Overall, the observations at the two sites are remarkably similar. The maximum average concentrations of surface O 3 for the residential and industrial sites are 10 ppb and 13 ppb, respectively, while those for NO x are 10 and 15 ppb (Figure 7). The mixing ratios of O 3 and its precursors show typical diurnal variation patterns as found in previous studies in locations on the east coast of Malaysia including Kuantan [16][17][18].  Figure 8. The model captured the diurnal variations and variabilities of the isoprene mixing ratio well throughout the study period. However, the maximum level of modeled isoprene occurred around 1200 LT, two hours earlier than the observed isoprene that was measured at 1400 LT. The average modeled isoprene was 2.49 ppb, slightly lower than the average measured isoprene of 2.63 ppb. The correlation coe cient between the simulated and observed isoprene mixing ratios was found to be 0.32 with an associated mean bias of -0.14 ppb over the location during December 2016.
High concentrations of isoprene were observed on the 12 th , 20 th and 23 rd December 2016 at noon and the model is able to capture these peaks. Meteorological parameters such as temperature, RH, precipitation, PBL, wind speed and wind direction have immense control over the BVOC production, distribution and dispersion processes.  Precipitation can trigger wet scavenging and leads to decreases in isoprene concentrations. In January 2017, the model performed well on the rst few days, then deviated from the observations. Simulated low temperature and high RH during the study period clearly satisfy the persistence of low isoprene mixing ratios. Like other meteorological parameters, PBL heights also have a critical role in the model to simulate BVOCs [32]. The simulated PBL height over the location is directly proportional to the simulated isoprene and inversely proportional to the observed isoprene concentrations.
On the one hand, direct involvement of PBL height can be found on the emission, dispersion and transport of isoprene in the model. The in uence of meteorological parameters on surface O 3 means their inclusion in the model is essential due to their effects on isoprene by formation, transport and dispersion of surface O 3 , especially temperature, PBLH and RH. On the other hand, direct involvement of PBL height can be found on the emission, dispersion and transport of isoprene in the model. Previous study reported that model simulated BVOCs such as isoprene, propene, acetone found low correlation coe cients with observations and large biases [61]. In general, the model was able to capture diurnal variations of the isoprene mixing ratio over the study location, but highly underestimated (30-40 %) most of the days. Such discrepancies in the model can be due to the uncertainties in the dynamics, physics, surface process and the land use/land cover.

Ozone
The correlation coe cient between the simulated and the observed surface O 3 is 0.38 with rmsd of 19.1 ppb. The in uence of meteorological parameters on surface O 3 is essential as its effect on isoprene by formation, transport and dispersion of surface O 3 , especially temperature, PBL height and RH. In contrary with isoprene simulated surface O 3 is overestimated most of the days (~ 35 %), which shows the model producing surface O 3 irrespective of isoprene de ciency. According to [62] and [63] the enhanced surface ozone formation is mainly attributed to factors such as oxidation of BVOCs and CO and the emission of isoprene from natural source. The simulated CO and NO x are compared with simulated surface O 3 and found a high correlation among each species irrespective of isoprene. This strengthens the in uence of CO on surface ozone formation and has a greater effect than isoprene in the model, and led to its overestimation over the study location.   Figure 9 (a-d). The update of isoprene emission factor on oil palm region has shown clear increment over the source region. The change mainly occurred during the day from 0900 -1900 LT while isoprene is emitted in respect of the photochemical reactions. Despite a short-lived reactive gas, emitted isoprene is able to transport to inland up to 100 km downwind. This is likely due to the prevailing northeast monsoon during the December month. With the same period of time, the change of O 3 is more localized over the source region (Figure 9 (e-f)). O 3 has clearly increased over the region where oil palm plantation has expanded. On the other hand, during January 2017, the isoprene does not show apparent difference compared to December 2016 and therefore O 3 is not affected. The emission of isoprene greatly depends on the weather condition. The high temperature stress triggered large emission of isoprene that is potentially a self-protection mechanism. Figure 10 has shown the temperature difference between the January 2017 and December 2016. It is illustrated that the temperature in January 2017 is lower during the day as compared to December 2016. This is mainly attributed to the boreal winter cold air mass that is carried over to the SEA region along with the northeast monsoon. The model results clearly show a seasonal emission of isoprene, particularly over the east coast of west Malaysia. The WRF-Chem model used by [65] showed lower emission rates of isoprene with 172.9 µg g -1 h -1 over palm oil areas in Distrito Metropolitano de Quito, Ecuador.
We concluded that the uncertainties of land use database input in different models may give different values to the emission rates. To better understand isoprene emission rates, we need more chemical transport models for comparison purposes for future emission predictions. In addition, ecosystem databases can be used to compute reasonable estimates of annual global isoprene emissions but may not produce accurate regional distributions [77]. Isoprene is believed to be emitted by Elaeis Guineensis as shown by our measurements. Meteorological factors such as temperature and irradiance showed strong correlations with the mixing ratios of isoprene, with r 2 = 0.91, p < 0.01 and r 2 = 0.82, p < 0.01, respectively. Surface O 3 was high during the middle of the day with a maximum value of ~ 57 ppb. Local and regional winds carrying NO x from nearby cities such as Kuantan were also found to in uence the mixing ratios of surface O 3 produced from isoprene over the Pekan oil palm plantation. Surface O 3 decreased during the night due to dry deposition and titration by NO x . Furthermore, isoprene mixing ratios were suppressed by the elevated CO 2 emissions during the night, believed to be due to the reduction of DMADP in the leaf cells. During photosynthesis, CO 2 is consumed in the stomata leading to high isoprene emissions in the day. During respiration, CO 2 is produced, leading to lower isoprene emissions from the leaves during the night. Thus, our ndings concluded that temperature, irradiance and CO 2 were the main factors controlling the levels of isoprene over the observation site. The WRF-Chem and WRF-CMAQ models were able to capture diurnal variations in the isoprene mixing ratio over the study location, although it highly underestimated (30-40%) the mixing ratio most days. Such discrepancies in the model can be due to the uncertainties in the dynamics, physics, surface processes and the land use/land cover. In addition, the MEGAN model predicted The backward trajectories used in this study was downloaded from ftp://arlftp.arlhq.noaa.gov/pub/archives/reanalysis.

Competing interests
The authors declare no competing interests.   Figure 1 Location of the Pekan oil palm plantation (left) on the Malaysian Peninsula and (right in respect to the city of Kuantan. S1 and S2 are the two DOE air quality monitoring stations whose data are used in this study. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.  Wind roses and daily average wind speeds for December (8th to 22nd December 2016) and January (1st to 12th January 2017). The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 6
Daily mixing ratios of isoprene, ozone, carbon dioxide and meteorological parameters measured above the palm tree canopy during December 2016 and January 2017.