Isoprene

The important role of isoprene in atmospheric chemistry was already introduced in Section 1.2. Its source strength exceeds those of any other biogenically emitted compound known so far. The most important sink of isoprene is reaction with OH (85%), but reactions with ozone (11%) and NO₃ (4%) are also non-negligible. It is also noted, that biologically mediated loss of isoprene to different types of soil was also observed (Cleveland and Yavitt (1997)), but a first extrapolation to the global scale resulted in only about 20 Tg/yr compared to 300-500 Tg/yr of direct emissions. Recently, Sanhueza et al. (2001) could infer from surface measurements made at a savanna site in Venezuela and box model calculations that deposition of isoprene and its oxidation products methyl-vinyl-ketone and/or methacrolein, not accounted for in present models, must be a significant loss process for these compounds at this site.

The monthly (24h-averaged) surface distribution of isoprene is shown in Figure 3.15 for January and July. Due to its short lifetime of a few hours the distribution closely follows the emissions shown in Figure 2.4. Maximum values of 13 nmol/mol are shown in the central Amazon basin. Even higher maximum values of up to 27 nmol/mol are found in both runs for other months. It is noted that these results are obtained with a reduced source strength of isoprene of 350 Tg(C)/yr compared to Guenther et al. (1995) (500 Tg(C)/yr) and the procedure to emit isoprene with an even mixing ratio over the PBL height.

Even more so than for the other alkenes, comparison of measurements of isoprene at single locations is problematic due to the coarse model grid and small scale horizontal variations in its concentration. Nevertheless, the ensemble of model/measurement comparisons can give indications on general tendencies of the model to over or under-predict concentrations in certain ecosystems. A compilation of available surface measurements along with model predictions is listed in Table 3.5. It is noted that the model values are 24h averages, whereas the measurements are usually made during daytime. From box-model simulation one can estimate that the 24-average value is about 50-75% of mid-day concentrations.

Figure 3.15: Mean (24h-averaged) isoprene surface mixing ratios for January and July. White spots over Amazonia are values above 10 nmol/mol.

The model appears to be in basic agreement with measurements made in North America. Measurements in Japan are expected to be higher, since the model grid-cells actually also contain an oceanic fraction. The values in the Georgian Republic are obtained in a coniferous forest on the southern slopes of the Caucasus Mountains, which might not be representative of the larger area. The model significantly overestimates mixing ratios found in the central Amazon region, sometimes by a factor of 3 or more. The Amazonia sites are probably quite representative of the region, since no major inhomogeneities in the vegetation are expected. The situation is different for the Peru site. It is located in a region where a strong horizontal gradient in isoprene is calculated by the model with mixing ratios below 1 nmol/mol to the west to more than 10 nmol/mol east of the site (within 2-3 grid cells even at the higher resolution). If the model value is obtained from interpolation of adjacent grids to the exact position of the site a smaller value (5.7 and 5.2 nmol/mol for T63 and T21 run, respectively) is found than given in Table 3.5. Since the measurement site is well within the primary rain forest the value of the next grid cell to the east is used here instead. Both methods yield a significant overestimation compared to the daytime values of Helmig et al. (1998a).

Vertical profiles of isoprene have been reported by Andronache et al. (1994); Helmig et al. (1998a), and Warneke et al. (2001). The modeled steep gradient with values of less than 5% of the surface mixing ratio at 1.5 km altitude are consistent with the first two studies. Over Surinam (LBA-Claire campaign), however, the decrease in the model is much too fast and only about 100 pmol/mol are left at about 2.5 km, whereas the observations are still above 1 nmol/mol up to 4 km (Warneke et al. (2001) during the LBA-CLAIRE^3.12 campaign). These measurements, made with the PTR-MS^3.13 technique, however, only give information on the mass of the molecules. The mass signals then have to be attributed to molecules that can be expected under those condition (see e.g. Williams et al. (2001)). Although the full signal at mass 69 was assumed to be isoprene, other minor contributors may be possible. Therefore, an uncertainty of 0.5 nmol/mol has been assigned to these measurements (Warneke et al. (2001)). Even with this uncertainty, it appears that the shallow convection reaching up to about 4 km altitude often encountered during the campaign is not reproduced by the model, which can be seen in the stronger than observed decrease with height (not shown).

Mixing ratios of isoprene sometimes well above 1 nmol/mol have been observed during the LBA-Claire campaign at about 11 km altitude, provided no other substance is contributing significantly to mass 69. The model only reaches maximum mixing ratios of up to 120 pmol/mol over Surinam (at T63), but can reach nmol/mol levels over the central amazon basin. Note that the T21 run produces much smaller isoprene mixing ratios in the upper troposphere (about 100-300 pmol/mol over the same region) due to the difference in convective pumping between the two runs.

Despite the problems mentioned with the comparison of such a short lived compounds, we conclude that calculated isoprene mixing ratios are too high over the rain forest regions of South America. More vigorous vertical mixing in the boundary layer could help to reduce the discrepancy but it would probably also produce even higher concentrations in the upper troposphere, which appear to be unrealistic, at least for the high resolution run. Especially considering that Jacob and Wofsy (1990) assumed an uptake of isoprene through the leaf stomata, but no experimental confirmation of this is available.

The overestimated isoprene concentrations over the Amazon region has also been found in other studies (Granier et al. (2000); Houweling et al. (1998)) and appears to be an unsolved problem in current research of biosphere-atmosphere interactions. The uncertainties regarding this compound are still very large and more observations are clearly needed to resolve the problem.

Note that most of this has been published in two papers: von Kuhlmann et al. 2003a+b.