China Map

      This report covers the ChinaMap activities related to Emissions which includes the efforts of David Streets at Argonne National Laboratory and Chip Levy at GFDL. This is an update to our report submitted prior to last November's ChinaMap Workshop held in Atlanta. The specific aspects of this report are: (1) the delivery of the newest versions of the NOx and SOx emission estimates for the years 1990, 1992/93, 1995 and 2000 on the ChinaMap grid; (2) the summary of the work on comparing the nitrate and sulfate wet deposition fields with measurements taken in China in 1992/93; and (3) a discussion of the gridded SO2 and NOx biofuel emissions consistent with other parts of the ChinaMap project [1]. Details are presented below.

1. SO₂ and NO₂ Emission Estimates

      1.1. Methodology

      A need was identified in the ChinaMap project to develop emissions profiles for China for the years 1992/93 and 1995, for the purpose of comparing calculated deposition with observed data for sulfate and nitrate deposition in Chinese monitoring data sets. The RAINS-Asia model generates sulfur dioxide emissions estimates for the years 1990 and 2000 (nitrogen oxide emissions estimates were determined by multiplying energy data from the RAINS-Asia model by appropriate emissions factors). It was decided that interpolating between the RAINS-Asia emission values for 1990 and 2000 would provide results consistent with other parts of the ChinaMap project. A methodology is being developed for a different project which would involve extrapolation from 1990 to 1995 using fuel-use proxies. For the present, however, interpolation is considered a reasonable method that relies on real data for 1990 and anticipated growth rates by region and sector for the period 1990-2000.

      Five sectors were examined in this project: industry, power, domestic, transportation, and non-commercial (includes biofuels that are collected and burned for energy, such as wood, crop residues, and animal waste). The two sectors that contributed the most sulfur dioxide and nitrogen oxides in 1990 were the industrial and power sectors, which were responsible for 50.8% and 28.1% of the SO₂ emissions and 43.6% and 35.5% of the NO_X emissions, respectively. By the year 2000, the industrial sector will be contributing more than one-half of the SO₂ emissions in China (54.5%). The percent share of NO_X emissions will decrease in all sectors over this period except for the transportation sector, which will double its contribution, from 6.6% in 1990 to 13.6% in 2000.

      For each of the 27 regions in China (the regions are shown in Figure 1), a fixed rate of growth between 1990 and 2000 was assumed for each sector. This allowed the interpolation of values for 1992/93 and 1995 from the 1990 and 2000 values generated by the RAINS-Asia model. The formula for fixed annual growth in this period is:

where E represents the emissions, in Gg, of pollutant and r is the annual rate of growth. By manipulating this function algebraically, we find that:

Values for 1992/93 were calculated in a similar manner:

For three of the sectors–industry, domestic, and transportation–1992/93 and 1995 values were interpolated by applying this formula. Slightly different methods were used for the power and non-commercial sectors.

      The power sector was examined in greater detail. Because RAINS-Asia makes available information about two parts of the power sector, small utility generators (< 500 MW) and individual large point sources (LPS >= 500 MW), it was decided that it would be more accurate to estimate values for individual LPS rather than to interpolate the total power sector values. (This is a more accurate method because we know that the square of the sums is not necessarily equal to the sum of the squares, or (P₁ + P₂ + P₃ ... +P_N)^1/2 <> (P₁^1/2 + P₂^1/2 + P₃^1/2 ... +P_N^1/2) .) Values for 1992/93 and 1995 were interpolated for each LPS as above. The 1992/93 and 1995 values for the total power sector were similarly calculated, the LPS totals were subtracted to yield values for the small sources. In some cases, however, the LPS was not yet on line in 1990, but is predicted to be in operation by 2000. This presents a problem for the above method, because assuming a fixed rate of growth does not work if the original value is zero (i.e., let E₁₉₉₀ = 0 E₂₀₀₀ = 0 because 0 (1 + r)¹⁰ = 0). For these LPS’s, literature sources [2,3,4,5] were consulted to determine when the power plant went on line. If the LPS went on line in 1991, the emissions from this date were used in place of the zero value for 1990 and the years 1992/93 and 1995 were interpolated accordingly. If it was found that the LPS was not yet on line in 1995, it was assigned a zero value for the years of 1992/93 and 1995. Getting the emissions from LPS’s right is important for accurately representing the geographical patterns of emissions, against which monitored deposition is to be compared.

      Data for the non-commercial (biofuel) sector are not part of the RAINS-Asia model. Data for 1990 were collected for a different project [6] and will be added to future versions of RAINS-Asia. Because only 1990 data were available, a different method, extrapolation for future years based on trend data, had to be developed for this sector. A few literature sources [7,8] had data for several years on per capita rates of energy consumption for different non-commercial fuels. Trends were extrapolated from these data for each type of biofuel (wood, crop residue, and animal waste) and were assumed to remain constant through 2000. In each case it was found that the per capita rate of consumption was decreasing over time. However, the rural population (the primary consumers of non-commercial energy) was growing at a faster rate, so total use of non-commercial energy increased slightly over this period. Appropriate conversion factors for each fuel and pollutant were then applied to the energy values projected for 1992/93, 1995, and 2000.

      Table 1 and Table 2 show the SO₂ and NO₂ emissions by sector for each of the 27 regions in China. For the industrial, power, domestic, and transportation sectors, the years 1990 and 2000 represent data taken from the RAINS-Asia model. The emissions values for years 1992/93 and 1995 are derived using the methodology described above. Table 3 lists each LPS in China, by region, name, and location (longitude and latitude). SO₂ and NO_X emissions are presented for each LPS, with values for the years 1990 and 2000 taken from the RAINS-Asia model. Emissions values for the years 1992/93 and 1995 were interpolated as explained above. Figure 2 and Figure 3 illustrate the total SO₂ and NO₂ emissions in China by sector and year.

1.2 Gridded Inventories

      The emissions estimated by David Streets and co-workers are provided at the regional level and are based on the RAINS-Asia region definitions as defined in Figure 1 and Table 4. This data is then assigned to ChinaMap grids using the following algorithm: (1) point source values are assigned to the grid containing the explicit point-source location; (2) area sources are first distributed to 1° by 1° grids belonging to the RAINS-Asia regions (every grid is assigned to only one region) using population as the weighting factor, and then mapped to the ChinaMap grid. The area sources are further limited by allowing distribution only to grids which are classified as "land" grids. The land grids are defined from the regional-climate model "land-sea" mask provided by NCAR.

      The SO₂ and NO₂ emissions are presented in Figure 4 and Figure 5, respectively. The data files are available for download.

2. Comparison with China Deposition Data

      The SO₂ and NO_x gridded emissions are the result of many steps, each including uncertainties. These include: regional energy use estimates by fuel and by sector; emission factors by fuel, region and activity; and allocation of the emissions to each grid using population as a surrogate for industrial/energy utilization. One step in the process of evaluating the estimated emissions involves the comparison of model predictions of sulfate and nitrate deposition which utilize these emissions, with observed values.

      China has been monitoring acid deposition since the late 1970's. In 1991 the Chinese EPA established a network of precipitation chemistry stations throughout China. In 1992/93 81 stations sampled precipitation chemistry. The locations of the stations are shown in Figure 6 and the locations are presented in Table 5. (Further background information on China precipitation measurements are presented in Appendix 1.) For the purpose of emissions evaluations, data on sulfate and nitrate wet deposition were made available for use in the ChinaMap project. The data were reported as wet deposition of sulfate and nitrate, by each of four seasons and an annual total. The observational data are summarized in Table 6 and Table 7. Unfortunately, details regarding the measurement, analysis and site description (other than location) did not accompany the data set. We have pieced together supplemental information. The network is largely located in the suburban areas of large cities (see Appendix 1for city names). Measurements were designed as wet only, but did not include automatic event only samplers, and samples were probably aggregated on multi-day or weekly basis. The data was reported to us as g-sulfur and g-nitrogen. However, we compared other references to this data, and to independent measurements at near-by locations, and we believe the data is reported as grams as sulfate and grams as nitrate. In addition, the nitrate deposition data was missing one season (i.e., it had data for 3 seasons and annual total).

     2.1 Sulfate Deposition

      The measured sulfate deposition was compared to results calculated using the RAINS-Asia model. In the RAINS-Asia model sulfur deposition is calculated using a three-layer lagrangian puff model. The model predictions are performed on a 1o by 1o grid system using observed meteorology for the base year 1990, and emission estimates for 1992/93. The total predicted sulfur deposition and wet deposition maps are presented in Figure 7. Super-imposed on these maps are the measured values. The total and wet deposition maps have the same general structure, and both capture the large-scale features in the observed deposition patterns. Specifically, the highest deposition values are observed and predicted to be in the Chongqing, Beijing, Shanghai, Guiyang, Guangzhou regions, with a very distinct east-west gradient. In general the observations are higher than the wet deposition predictions, and closer to the total predicted values.

      The data is compared more quantitatively in Figure 8, where scatter-plots of the model predicted and observed values are presented. Also plotted are the factor of 2 lines. The observations are point values and the predictions are for the 1 by 1 grid cell containing the observation. The observed data are plotted separately against the RAINS-Asia predicted total, wet-only and dry-only deposition values. In general, the wet-deposition amounts are under predicted, while the total deposition estimates are systematically higher than the observed values. This behavior is consistent with the measurement values being more representative of urban areas, and not representative of the grid-cell average value. Other factors which could contribute to these differences between observations and predictions include model, measurement, and/or emissions deficiencies. On the measurement side, a sampling methodology which captured some portion of dry deposition or had some evaporative loses would be consistent with the model predictions. A major factor on the modeling side is the fact that the predictions utilized 1990 meteorology, while the measurements were performed in 1992/93. We do plan to re-run the RAINS-Asia model with multiple years of meteorology (including 1992 and 1993) (as part of RAINS-Asia Phase-II), and this will enable a direct evaluation of the contribution to differences in meteorological conditions. Observed values of precipitation at the measurement sites would be useful and could be compared to the 1990 levels to see if differences in precipitation amounts could be an important factor.

      Seasonal variations in sulfate deposition were also compared with the observations. Shown in Figure 9 are the seasonal wet deposition maps, along with the observed seasonal deposition data. Again the model predictions are qualitatively consistent with the observations. For example, the largest wet deposition occurs in the summertime, and a distinct decrease in wet deposition above about 30°N latitudes in the winter is shown in both the observed and predicted values. The seasonal scatter plots are also presented (Figure 10). As shown in the wet-deposition only plots, there is a large scatter for each season. However, the largest differences occur in the spring season, where the observed values are very high, and a systematic underprediction is apparent. Additional information on the sampling procedure, and observed precipitation fields are needed to develop explanations for differences in the spring comparisons.

      We also compared the observed wet deposition values directly with the gridded emission values. to see how well the observed fields are correlated with the estimated emissions. Results are summarized in Figure 11. In general rather poor results are found when the emissions in the grid cell containing the measurement site are regressed with the observed values. Similar results were found when the observations were regressed with the emissions shifted in a systematic way (e.g., all systematic combinations (eight surrounding grid cells) of nearest neighbor shifts were tried). The best results were found by selecting points from the nine surrounding grid cells. Most of the points were chosen from the actual grid containing the measurement point and the NE nearest neighbor. These results are perhaps not too surprising since the wet deposition fields are a function of both emissions and transport characteristics, and deposition fields of sulfate are smoother (and typically displaced hundreds of kilometers) than the emissions field (see for example Figure 4andFigure 7)

     2.2 Nitrate Deposition

      Nitrate deposition predictions using the GFDL GCTM, run with climatological winds and using the 1990 NO₂ emissions estimates, were also compared with the measured annual deposition data. The scatter plots are presented in Figure 12. (The predicted results using the 1992/92 emissions are on average 10% higher than those shown in Figure 12). The GFDL-GCTM utilizes a grid system which is roughly 3 degrees by 3 degrees. Therefore several observation points may fall within the model grid. To compare the data, multiple observation points were averaged within the grid. This reduced the number of comparison points from 81 to 62 . The results are similar to those for sulfate, in that the predicted wet deposition of nitrate is systematically lower than the observations, while the predicted total nitrate deposition is systematically higher than the observed value. NO₂ emissions were also regressed against observed nitrate deposition in the same manner as discussed for SO₂, and similar results were found.

3. Gridded Biofuels Emission Estimates

      One unique aspect of the emissions work has been the inclusion of emissions from non commercial sources. These biofuel emissions were previously discussed in Section 1.1 of this report. In 1990 biofuel use in China accounted for emissions of ~0.5 Gg-SO₂ and ~0.7 Gg-NO₂, which represent 2% and 8% of China's total SO₂ and NO₂ emissions, respectively. The usage of biofuels varies strongly with region. This is shown more clearly in Figure 13, where the gridded 1990 emissions are presented, along with plots indicating the percent contribution of befuel emissions for each region. In the western and southeastern regions of China biofuels account for more than 10% of the SO₂ emissions; and in some regions of eastern China biofuel sources account for between 20 to 30% of NO₂ emissions. Biofuel are a very significant source of emissions in Southeast Asia, while a small contributor in Japan and Korea.

4. Closure

      The emission estimates have been compared with other estimates in our last report, and with the Chinese EPA data on sulfate and nitrate deposition. When the emission estimates are used in deposition models, the resulting predicted regional distributions of sulfate and nitrate deposition are qualitatively consistent with the observed patterns. In both the sulfate and nitrate calculations the models underpredict the wet deposition amounts. The lack of detail regarding the observed data sets, and the differences in meteorological conditions used in the modeling analysis (1990 for sulfate deposition and climatological for nitrate deposition) make it difficult to draw any hard conclusions regarding whether the differences between the observations and predictions are due to errors in the emissions, deficiencies in the models, and/or due to the design of the monitoring network (i.e., consisting of sub-urban as opposed to regional stations). Additional information on the observations (methodology, etc.) and measured precipitation amounts would greatly aid in the analysis. (Please let us know to whom we can direct a request for additional information.)

      We are now working on finalization of the mineral aerosol emission algorithm, and on estimating emissions from CO. This work will be reported in our next interim report, which we plan to deliver in the next few months.

Figure 1.     Regions used in Rains-Asia. Names are given in Table 4.

Figure 2.     Chinese SO₂ Emissions by Sector.

Figure 3.     Chinese NO_x Emissions by Sector.

Figure 4.     Estimated SO₂ emissions in 1990, 1992/93, 1995 and 2000. Units: gm-SO₂/m²/yr.

Figure 5.     Estimated NO_x emissions in 1990, 1992/93, 1995 and 2000. Units: gm-NO₂/m²/yr.

Figure 6.     Locations of precipitation chemistry monitoring sites.

Figure 7.     Calculated total and wet-only annual sulfur deposition using the Rains-Asia model and emission estimates. Also shown are the measured values.

Figure 8.     Measured and predicted sulfur deposition in China. Comparisons are with observed wet deposition and predicted total, wet-only and dry-only deposition. Units: gm-SO₄^2-/m²/yr.

Figure 9.     Same info as Figure 7 but showing wet-only deposition by season, DJF, MAM, JJA, SON.

Figure 10.     Measured and predicted sulfur deposition in China. Comparisons are with observed wet deposition and predicted total, wet-only and dry-only deposition by season.

Figure 11.     Comparison of observed wet deposition with (1) estimated emissions in the same grid cell as observed; (2) estimated emissions in one of nearest neighbors. A plot for NO_x emissions for case (2) is also shown.

Figure 12.     Predicted and observed nitrate deposition. Predicted values are from the GFDL-GCTM using China Map emissions.

Figure 13.     Gridded 1990 emissions of SO₂ and NO_x from biofuels. Units: gm SO₂/m² or gm NO₂/m² per year. Also shown are the percent contribution of biofuels to the total estimated SO₂ and NO_x emissions.

Table 1.     Chinese SO₂ Emissions by Sector - 1990,1992/93,1995,2000.

Table 2.     Chinese NO_x Emissions by Sector - 1990,1992/93,1995,2000.

Table 3.     Large Point Sources of China.

Table 4.     Rains-Asia region names. See Figure 1 for locations.

Table 5.     China Wet Chemistry Monitoring Sites 1992/93.

Table 6.     Sulfate - Wet deposition, 3 month quantities or Annual, in China. Units: gm SO₄^2-/m²/year.

Table 7.     Nitrate - Wet deposition, 3 month quantities or Annual, in China. Units: gm NO₃^-/m²/year.

Sources:

(1) Arndt, R.L., Carmichael, G.R., Streets, D.G., and Bhatti, N., Atmospheric Environment, Vol 31. No. 10, 1553-1572, 1997.
(2) Energy in China, Ministry of Energy, People's Republic of China, Beijing, 1992.
(3) Maude, C.W., Kirchner, A.T., Daniel, M., and Montfort, O., World coal-fired power stations: Africa, Asia and Australia, International Energy Agency Coal Research, London, 1994.
(4) Electric Power, eds. J. Sun and C. Liu, Epoch Printing Co., 1995.
(5) Directory of Power Plants in Asia and Australia, ed. C.A.E. Bergesen, Utility Data Institute, Inc., UDI-051-92, Washington, D.C. 1992.
(6) Streets, D.G. and Morris, S.T., Energy The International Journal, 1998, in press.
(7) Lu, Y., Fueling One Billion, Washington Institute Press, WAshington, D.C., 1993. (8) China Energy Databook, ed. J.E. Sinton, Lawrence Berkeley National Laboratory, LBL - 32833 Rev. 4 UC-900, Berkeley, 1996.