Carbon 13 composition of tropospheric CO in Brazil: A model scenario during the biomass burn season

The stable isotopes of carbon and oxygen are potentially powerful tools for distinguishing sources of CO in the troposphere due to isotopic differences among source emissions that are caused by isotope fractionation in formation or reaction. It is incorrect, however, to assume that the CO source strengths estimated using isotopic measurements on single‐day air samples truly represent a season and region. Atmospheric transport and dispersion models are useful for selecting representative sampling locations, dates, and duration to adequately reflect isotopic variation. Here a three‐dimensional transport and dispersion model was used to predict surface‐level 13 CO/ 12 CO ratios at four remote sites in the rain forest and savanna of Brazil during the 1992 burn season. The purpose was to demonstrate the scope of surface‐level CO isotopic variation due to isotopically distinct source emissions and changing meteorology. The model included 13 C signatures of four classes of CO sources: biomass burning, oxidized vegetative nonmethane hydrocarbon (NMHC) emissions, atmospheric methane oxidation, and fossil fuel combustion. Among the four model locations, sites 1 and 2 were well within the burn region, site 3 was at the edge of it, and site 4 was well north of it. The model employed the program HY‐SPLIT to track air masses and calculate CO concentrations from emissions at satellite‐detected burn sites which were mainly in the Brazilian savanna. An average CO δ 13 C value for burned biomass (−21.3‰ versus PDB) was determined from our δ 13 C measurements of savanna biomass, reported fuel loadings, and the distribution of savanna plant communities in Brazil. Two model scenarios were created, based mainly on the level of CO from fossil fuel combustion. Scenario A had a low CO contribution from this source (15 ppbv), and scenario B had a higher CO contribution (100.1 ppbv). Both model scenarios used −32.2, −48.3, and −25‰ for CO δ 13 C values for oxidized vegetative NMHC emissions, CH 4 oxidation, and fossil fuel combustion, respectively, based on data reported by others. Sensitivity studies showed that at sites closest to the burn region the model was influenced largely by the 13 C composition of burned biomass for both scenarios. At the site farthest from the burn region the model was influenced moderately by the amount of CO emitted per fire, a greater rate of CH 4 oxidation, and a higher 13 CO/ 12 CO ratio for fossil fuel combustion, particularly for scenario B. For the model scenario with minimal CO from fossil fuel combustion (scenario A), results showed surface‐level δ 13 C values for August 5, 1992, averaging about −23‰, close to the average δ 13 C value for biomass burning CO. Model results for August 11, 1992, showed 13 CO/ 12 CO that ratios at sites 1–3 were, again, close to the ratio for biomass burning CO (δ 13 C=−22.6‰ to −24.7‰). The more 13 C‐enriched values match closely with the most 13 C‐enriched measurements that have been reported for July/August in the tropics and southern hemisphere when elevated CO levels are driven by emissions from large‐scale biomass burning. At site 4 for August 11, 1992, the calculated surface‐level δ 13 C average was −32.6‰. Thus results indicate that 13 CO/ 12 CO ratios may be highly variable from week to week in the Amazon region during the biomass burn season. Model results suggest that on August 5, 1992, fossil fuel combustion probably did not alter significantly the 13 CO/ 12 CO ratio in surface‐level air at sites 1–4, given the low and high levels of CO from fossil fuel combustion defined in the two model scenarios. In addition, measurements taken at sites 1–3 probably would have been indistinguishable from the 13 C composition of the biomass burning source. At site 4 on August 11, however, other CO sources probably altered significantly the 13 CO/ 12 CO ratio in surface air from that of CO from biomass burning.