CONDENSED ATMOSPHERIC PHOTOOXIDATIONMECHANISMS FOR ISOPRENEbyWilliam P. L. CarterAtmospheric Environmentin pressFebruary 22, 1996Statewide Air Pollution Research Center andCollege of Engineering, Center for Environmental Research and TechnologyUniversity of CaliforniaRiverside, CA 92521ABSTRACTTwo condensed mechanisms for the atmospheric reactions of isoprene, which differ in the numberof species used to represent isoprene’s reactive products, have been developed for use in ambient airquality modeling. They are based on a detailed isoprene mechanism that has recently been developed andextensively evaluated against environmental chamber data. The new condensed mechanisms give veryclose predictions to those of the detailed mechanism for ozone, OH radicals, nitric acid, H2O2,formaldehyde, total PANs, and for incremental effects of isoprene on for ozone formation in one daysimulations. The effects of the condensations become somewhat greater in multi-day simulations,particularly in cases where NO3reactions are important at nighttime, but the ozone predictions are stillvery close. On the other hand, the SAPRC-90, RADM-2, and Carbon Bond IV isoprene mechanisms givequite different predictions of these quantities. It is recommended that the new mechanisms replace thosecurrently used in airshed simulations where isoprene emissions are important.KeywordsIsoprene, Airshed Models, Chemical Mechanisms, Photochemical Smog, Ozone, Air Quality, Methacrolein,Methyl Vinyl Ketone, Biogenic Hydrocarbons.INTRODUCTIONIsoprene is emitted from certain types of vegetation, and is believed to play an important role inboth urban and rural ozone formation (Trainer et al, 1987; Chameides et al, 1988; Sillman et al, 1990).For this reason, its reactions are represented in most of the currently used urban or regional air qualitymodels. For example, the Carbon Bond IV (CB4) (Gery et al, 1988), RADM-2 (Stockwell et al (1990),or SAPRC-90 (Carter, 1990; Lurmann et al, 1991) chemical mechanisms, which are widely used in airshedmodels, all include separate reactions for isoprene. However, to avoid adding new species to the modelto represent speculative reactions of isoprene’s products, these isoprene mechanisms are all highlycondensed. In addition, in recent years there has been substantial improvements in our understanding ofthe atmospheric chemistry of isoprene (Paulson and Seinfeld, 1992, Carter and Atkinson, 1996, andreferences therein), and this new information is not reflected in these mechanisms.Recently, Carter and Atkinson (1996) developed a detailed mechanism for isoprene whichincorporates the recent progress in our understanding of isoprene’s atmospheric reactions. This was2evaluated using results of NOx-air irradiations of isoprene and its two major products, methacrolein andmethyl vinyl ketone (MVK), in five different environmental chambers at two different laboratories. Inmost cases the mechanism simulated the experimental data to within the uncertainty of the data and thechamber and run characterization model, although it tended to underpredict PAN yields in the isopreneruns, despite giving good simulations of this product in the methacrolein and MVK runs. This discrepancyfor PAN might be due to uncertainties in the mechanism developed to represent the reactions of the C5unsaturated carbonyl products, though the possibility that it is due to interferences in the experimentalmeasurements of PAN have not been ruled out.In any case, this new mechanism gives substantially better simulations of the data than themechanism of Paulson and Seinfeld (1992), the most up-to-date and comprehensive isoprene mechanismprior that work. In addition, as shown in Figure 1, the new mechanism also gives substantially betterpredictions of ozone formation and NO oxidation in representative environmental chamber experimentsthan do the condensed mechanisms currently used in airshed models. [Ozone formation and NO oxidationis measured by the quantity d(O3-NO), the change in [O3]-[NO], since the start of the experiment. SeeCarter and Atkinson (1996), for a discussion of the chamber modeling approach, and results of simulationsof other experiments and measurements.] While the d(O3-NO) data from some of the runs are reasonablywell predicted by some of the condensed mechanisms, the new mechanism consistently gives the bestpredictions for the largest number of experiments. Therefore, this can be considered to represent anadvance in our ability to model the atmospheric reactions of isoprene.In view of this, the Carter and Atkinson (1996) isoprene mechanism ideally should be used inairshed model applications where the reactions of this compound might be important. However, it is muchmore detailed than most would consider to be necessary or appropriate for current airshed modelapplications. In particular, it requires adding to the general mechanism a total of 19 new species, listedin Table 1, to represent isoprene’s various primary and secondary products. This is far greater than thenumber of species currently used for any of the other VOCs present in the atmosphere, and this level ofdetail is not necessary in most current applications cases where the primary interest is in simulating themajor air quality features such as ozone, overall radical levels, total nitrate or oxidant formation, etc.In this paper we present two condensed versions of the Carter and Atkinson (1996) isoprenemechanism that might be more suitable to current model applications. Since isoprene’s products aresufficiently different in reactivity characteristics from other product species already in the mechanisms,a minimum level of chemical realism requires the addition of at least one new species to the model to3represent these compounds. However, a mechanism where all isoprene’s products are lumped togetherwould not be useful for applications where isoprene product data are available for comparison with modelpredictions. For example, atmospheric measurements of methacrolein and MVK, the major isopreneoxidation products whose yields have been quantified (Carter and Atkinson 1996, and references therein)have been reported in certain ambient air studies (Pierotti, et al., 1990; Martin et al., 1991; Montzka etal., 1993, 1995; Yokouchi, 1994). Therefore, we also developed a second version of the mechanism wheremethacrolein and MVK are represented explicitly.DESCRIPTION OF MECHANISMSThe starting point for this work is a version of the detailed SAPRC mechanism