REGULATORY REQUIREMENTS FOR DISPERSION MODELLING IN TAIWAN INCLUDING THE USE OF ADMS David Carruthers 1, Xiangyu Sheng, Eilie Vanvyve 1 1 CERC, 3 King s Parade, Cabridge CB 1SJ, UK Scott Wilson, Michael Young Centre, Cabridge, CB QL Abstract: The ADMS range of odels (ADMS for industrial eissions, ADMS-Screen, ADMS-Urban and ADMS-Roads) are currently used across the world and have been widely used in Asia, e.g. Mainland China, Hong Kong, and India. This paper considers the regulatory requireents for dispersion odelling and the application of ADMS in Taiwan. Until recently ISCPrie has been the only short range dispersion odel acceptable for regulatory purposes in Taiwan. For the purposes of regulatory dispersion odelling, Taiwan is divided into 1 different regions each region containing a significant urban area; for exaple there are areas centred on the ajor cities of Taipei and Kaoshung. Each region has assigned to it a particular representative eteorological dataset (hourly data for a full year) and surface conditions including surface roughness; because of the ountainous nature of Taiwan the ipact of coplex terrain needs to be considered in all cases. Previously as part of the regulatory requireents odelling test cases for single point sources were set up with ISCPrie for each of the regions and these datasets are available on the Taiwanese EPA website. In order for ADMS to be approved for use in Taiwan the test cases were required to be run for ADMS. Model output required included contour aps of the annual 1 h ean concentration, and 1 hour and hour average percentiles and a tie series of hour averages at specified receptor points tie. This paper presents the results of using those datasets for ADMS, and discusses the perforances of ADMS in a Taiwan context including where appropriate coparison with ISCPrie and discussion of the ipacts of coplex terrain and surface roughness in the different regions. 1. INTRODUCTION This report describes odelling studies perfored as part of the subission of ADMS Carruthers et al (1993) to the Taiwanese Environental Protection Agency (EPA) for approval for use for regulatory purposes in Taiwan. The overall requireents of the Taiwanese authorities consisted of the following: (i) coparisons between ADMS and the current regulatory odel ISC3P, in particular coparisons of the th, th and 9 th percentiles of 1-hour and -hour averages and the correlation at a key receptor points, contours of pollutant concentrations over the odelling doain and tie series of concentrations at the receptor point; (ii) consideration and explanation of the differences between ADMS and ISC3P; there is currently a predisposition on the part of the regulatory authorities to assue that ISC3P is the odel by which other odels are judged. Inforing the authorities about the scientific advantages of other odels is an iportant part of enabling this situation to change. (iii) indication of the input paraeters required by ADMS including eteorological input and also paraeters used to represent the surface properties at each of the sixteen areas in Taiwan (surface roughness, albedo and Taylor-Priestley paraeter); (iv) discussion of basic ADMS odelling approach and pollutant types considered by the odel; (v) discussion of the treatent of coplex terrain, ipacts of buildings and treatent of cal conditions; (vi) discussion of treatent of deposition and cheical transforation; (vii) references for odel. With regard to the odelling requireents ADMS was required to be run for sixteen different areas in Taiwan each containing a significant city. These areas were Chahwa, Chaji, Hwalein, Ilan, Inlin, Kaoshung, Keelung, Miauli, Nantou, Pintung, Shinchu, Taichung, Taidong, Tainan, Taipei and Taoyuan and are shown in Figure 1. The sae locations have previously been odelled with ISC3P and thus as stated in (ii) above a key aspect of this work is to copare the pollutant concentrations calculated by the two different odels. Figure 1. The sixteen site locations and topography of the island. 1
. SIMULATIONS Taiwan data were provided to CERC by the EPA in two series: (1) eteorological data for a nuber of sites, a topography database and data of eissions fro transport, industries, etc., and () the input, configuration and result files of the siulations perfored with ISC3P. The data provided were used to configure ADMS. In particular, the topographical data, wind easureent heights and weather station latitudes were extracted fro the first series, and the source location, odelling doain and eteorological data fro the second series. Each siulation covered a by k area with a - resolution (1x1 grid points). In each case an idealized eission fro a single stack was odelled with the following characteristics: height ; diaeter 1. ; continuous eission of total suspended particulates TSP of gs -1 with an exit velocity of. s -1 and at a teperature of 99. C. The siulations all included topography and one year of hourly sequential eteorological data. Note that the siulations for each area differ in the doain they cover (different locations), the topography (that corresponding to each site), the eteorological data (different weather stations) and the location of the receptor. 3. RESULTS Model calculations were perfored for each of the sixteen sites and the following and output as required were the following: Note that since the stack is not representative of a stack at a real site there was no option to copare odelled and easured concentrations 1) a wind rose of the wind directions and speeds ) a contour plot of the annual ean concentration of TSP calculated fro the hourly concentration values and presented superiposed on topography, 1 th percentile indicated by a dot; 3) a tie series of the daily concentration of TSP with the correlation concentrations are for a specific receptor located close to the source (within 1 k; ) a tie series of the hourly and daily concentration percentiles of TSP concentrations are at the receptor located nearby the source (within 1 k); ) a table of the th and 9 th percentiles of the hourly and daily concentrations of TSP these values ay not atch directly those shown in the plot of percentiles as an additional filter was applied to the data (the ISC3P values were only considered if an ADMS value was available for the corresponding hour). Note that since the stack is not representative of a stack at a real site there was no option to copare odelled and easured concentrations. A suary of the odel results are presented in the Table whilst ore detailed results for Chawa and Nantou are presented in Figures and 3. Table 1. 9 th percentiles of 1-hour and -hour averages at receptor points in each of the sixteen Taiwan areas. 1-hour average concentration -hour average concentration Site ADMS ISC3P Difference Difference ADMS ISC3P ADMS-ISC3P ADMS-ISC3P Correlation ( g -3 ) ( g -3 ) ( g -3 ) (%) ( g -3 ) ( g -3 ) ( g -3 ) (%) Chahwa 1.7 1.3 1.11 7..9..13..3 Chaji 1.9 1..31..3.1.9 1..9 Hwalein 1.1. 1.9 1... -1.1 -. -.3 Ilan.1 1.7-3. -...9 -. -.9.71 Inlin 1.9 1..1.1.7. -3. -..9 Kaoshung 1.3.7 1.93 1. 3.9. -1.39 -..1 Keelung 11.1.1-17.3 -..97 1. -1. -77..7 Miauli 11.9 3.7-11. -9..71. -.7 -.1. Nantou 11.31. -1.9-11. 3. 3.3 -.1 -.3.7 Pintung 1..9. 19..1. -. -..17 Shinchu...19.3.1. -.3-13.. Taichung..3 -. -3..71.9 -.3 -.7. Taidong 1.1 1. 1.1 7...7. 1..1 Tainan.7 19.3.9. 7.3 1.77-3. -3.9.1 Taipei 1. 17. 1..7 7.1. -.1-37.7. Taoyuan 1.39 1.1-7.71 -...9-7.1 -.3.7 It is seen that the differences between the odel predictions at the receptor point is relatively sall at ost sites, the notable exceptions being Keelung, Miauli and Taoyuan. The concentrations calculated by ISC3P at these points are likely to be unrealistically high due to the tendency of the ISC3P coplex odel soeties to overpredict by large factors in ISC3P the hilly or ountainous terrain does not affect the airflow so that the plue is transported directly into the hill or ountain side; this effect is illustrated well in the contour plots for Nantou where ISC3P shows high levels on hill top despite very little wind in the relevant directions. In reality it is well understood the air tends to flow 13
over (neutral/convective flow) or around hilly terrain reducing ground-level concentrations relative to those predicted by ISC3P. Note that ADMS explicitly calculates the flow over coplex terrain and hence the dispersion. There are of course any other factors which distinguish ADMS (a new generation odel taking account of the latest understanding of the physics of the atospheric boundary layer and dispersion) fro ISC3P (an old generation odel which is based on an outdated representation of the atospheric boundary layer). These can of course result in differences in the predicted concentrations between the two odels. Considering the results for Chawa and Nantou in ore detail, notable for Chawa is the uch broader area affected by elevated annual average concentrations especially to the north of the site despite the wind generally being fro the north; ADMS generally reflects the prevailing wind. Despite this the peak short ter concentrations as indicated by the table of th and 9 th percentiles near the point of greatest ipact are generally siilar. A siilar conclusion can be reached for the Nantou site with arkedly different spatial pattern for the annual average but quite siilar peak levels. 3 31 Chahwa 1 1 3 33 3 3 9 7 7 9 1 11 3 13 1 1 1 19 17 1 1 3 1 1 (knots) Wind speed 1.3.1.1. (/s) 7 3 9 3 31 Nantou 1 1 3 33 3 3 1 7 3 13 1 1 1 19 1 1 17 3 1 1 (knots) Wind speed 1.3.1.1. (/s) 9 1 11 Figure. Wind roses of Chahwa and Nantou. Chahwa - annual concentration of TSP (ug/³) ISC3P, terrain o = source, x = onitor Chahwa - annual concentration of TSP (ug/³) ADMS, terrain o = source, x = onitor ax:. ug/ 3 19 19 19 1..1 1. 1. 1..9..3.1 17 1 1 7 ax:. ug/ 3 19 19 19 1 Figure 3. Annual ean concentration for Chahwa of TSP odelled by ISP3P and ADMS. The receptor is arked close to the source...1 1. 1. 1..9..3.1 17 1 1 7 ISC3P ADMS conc. (ug/ 3 ) 1 correlation =.3 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1 Figure. Chahwa; Teporal series of ISC3P and ADMS odel results at the receptor. 1 conc. (ug/ 3 ) 1
ISC3P ± % ISC3P 1-h ADMS 1-h ISC3P -h ADMS -h concentration (ug/ 3 ) 3 1 1 3 1 1 concentration (ug/ 3 ) 1 3 7 9 1 percentiles (%) Figure. Percentiles plot of ISC3P and ADMS odel results at the receptor; for hourly and daily averages. Table. ISC3P and ADMS hourly and daily percentiles for Chahwa. Data 1-hour average -hour average ISC3P ADMS ISC3P ADMS th percentile....1 9 th percentile 1.3 1.7..9 3 3 3 3 Nantou - annual concentration of TSP (ug/³) ISC3P, terrain o = source, x = onitor ax: 1.3 ug/ 3 3 1 1 1 3..1 1. 1. 1..9..3.1 17 1 1 7 3 3 3 3 Nantou - annual concentration of TSP (ug/³) ADMS, terrain o = source, x = onitor ax:. ug/ 3 3 1 1 1 3 Figure. Annual ean concentration of TSP for Nantou odelled by ISC3P and ADMS of Nantou. The receptor is arked close to the source...1 1. 1. 1..9..3.1 17 1 1 7 ISC3P ADMS conc. (ug/ 3 ) 1 correlation =.7 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1 Figure 7. Teporal series of daily averages of ISC3P and ADMS odel results at the receptor. 1 conc. (ug/ 3 ) 1
ISC3P ± % ISC3P 1-h ADMS 1-h ISC3P -h ADMS -h concentration (ug/ 3 ) 3 1 1 3 1 1 concentration (ug/ 3 ) 1 3 7 9 1 percentiles (%) Figure. Percentiles of ISC3P and ADMS odel results at the receptor point hourly and daily averages. Table 3. ISC3P and ADMS hourly and daily percentiles for Nantou. Data 1-hour average -hour average ISC3P ADMS ISC3P ADMS th percentile.. 1. 1. 9 th percentile. 11.31 3.3 3.. CONCLUSION The regulations in Taiwan have until recently required the sole use of ISC3P for regulatory peritting. The regulators have now accepted the principle that other odels ay be used subject to satisfactory copliance with subission requireents. Use of new odels ay provide the basis for these regulations to evolve so that ISC3P is no longer seen as the standard. REFERENCES Carruthers D.J, Holroyd, Hunt J.C.R, Weng W-S, Robins A.G, Apsley D.D, Thoson D.J and Sith F.B., 199: UK- ADMS: A new approach to odelling dispersion in the earth s atospheric boundary layer. Journal of Wind Engineering and Industrial Aerodynaics,, 139-13. Taiwan EPA odel approval procedure: http://www.aqc.org.tw/, accessed at August. 1