Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection

4.4.3 Testing CTM Simulation of the Current (Y2000) Atmosphere

The OxComp workshop defined a series of atmospheres and emission scenarios. These included Y2000, a new reference atmosphere meant to represent year 2000 that provides a baseline from which all changes in greenhouse gases were calculated. For Y2000, abundances of long-lived gases were prescribed by 1998 measurements (Table 4.1a), and emissions of short-lived pollutants, NOx, CO and VOC, were based primarily on projections to the year 2000 of GEIA/EDGAR emissions for 1990 (Olivier et al., 1998, 1999), see Section 4.3.1. Stratospheric O3 was calculated in some models and prescribed by current observation in others. The predicted atmospheric quantities in all these simulations are therefore short-lived tropospheric gases: O3, CO, NOx, VOC, OH and other radicals. Following the GIM model study (Kanakidou et al., 1999), we use atmospheric measurement of O3 and CO to test the model simulations of the current atmosphere. The Y2000 atmosphere was chosen because of the need for an IPCC baseline, and it does not try to match conditions over the l980s and 1990s from which the measurements come. Although the observed trends in tropospheric O3 and CO are not particularly large over this period and thus justify the present approach, a more thorough comparison of model results and measurements would need to use the regional distribution of the pollutant emissions for the observation period.

The seasonal cycle of O3 in the free troposphere (700, 500, and 300 hPa) has been observed over the past decade from more than thirty ozone sonde stations (Logan, 1999). These measurements are compared with the OxComp Y2000 simulations for Resolute (75°N), Hohenpeissenberg (48°N), Boulder (40°N), Tateno (36°N), and Hilo (20°N) in Figure 4.10. Surface measurements from Cape Grim (40°S), representative of the marine boundary layer in southern mid-latitudes, are also compared with the models in Figure 4.10. With the exception of a few outliers, the model simulations are within ±30% of observed tropospheric O3 abundance, and they generally show a maximum in spring to early summer as observed, although they often miss the month of maximum O3. At 300 hPa the large springtime variation at many stations is due to the influence of stratospheric air that is approximately simulated at Resolute, but, usually overestimated at the other stations. The CTM simulations in the tropics (Hilo) at 700 to 500 hPa show much greater spread and hence generally worse agreement with observations. The mean concentration of surface O3 observed at Cape Grim is well matched by most models, but the seasonality is underestimated.

Observed CO abundances are compared with the Y2000 model simulations in Figure 4.11 for surface sites at various altitudes and latitudes: Cape Grim (CGA, 94 m), Tae Ahn (KOR, 20 m), Mauna Loa (MLH, 3397 m), Alert (ALT, 210 m), and Niwot Ridge (NWR, 3475 m). The Alert abundances are well matched by most but not all models. Niwot Ridge and Mauna Loa are reasonably well modelled except for the February to March maximum. At Tae Ahn, the models miss the deep minimum in late summer, but do predict the much larger abundances downwind of Asian sources. At Cape Grim the seasonal cycle is matched, but the CO abundance is uniformly overestimated (30 to 50%) by all the models, probably indicating an error in Southern Hemisphere emissions of CO.

Overall, this comparison with CO and O3 observations shows good simulations by the OxComp models of the global scale chemical features of the current troposphere as evidenced by CO and O3; however, the critical NOx chemistry emphasises variability on much smaller scales, such as biomass burning plumes and lightning storms, that are not well represented by the global models. With this large variability and small scales, the database of NOx measurements needed to provide a test for the global models, equivalent to CO and O3, would need to be much larger.
The current NOx database (e.g., Emmons et al., 1997; Thakur et al., 1999) does not provide critical tests of CTM treatment of these sub-grid scales.


Figure 4.10: Observed monthly mean O3 abundance (ppb) from sondes at 700 hPa (left column), 500 hPa (centre) and 300 hPa (right) from a sample of stations (thick black line) compared with Y2000 model simulations from OxComp (thin coloured lines, see model key in legend and Table 4.10). The sonde stations include RESolute (75°N, 95°W), HOHenpeissenberg (48°N, 11°E), BOUlder (40°N, 105°W), TATeno (36°N, 140°E), and HILo (20°N, 155°W). Surface monthly O3 observations (thick black line) at Cape Grim Observatory (CGA, 40°S, 144°E, 94 m above mean sea level) are also compared with the models.

Figure 4.11: Observed seasonal surface CO abundance (ppb, thick black lines) at Cape Grim (CGA: 40°S, 144°E, 94 m above mean sea level), Tae Ahn (KOR: 36°N, 126°E, 20 m), Mauna Loa (MLH: 19°N, 155°W, 3397 m), Alert (ALT: 82°N, 62°W, 210 m), and Niwot Ridge (NWR: 40°N, 105°W, 3475 m) are compared with the OxComp model simulations from Y2000, see Figure 4.10.


Other reports in this collection