Preparation of Meteorological Input for AERMOD using Malaysian ...

4 downloads 597 Views 315KB Size Report
meteorological stations located in Peninsular Malaysia. (Cameron Highlands, Subang, Sepang KLIA, and. Kuantan), replacement of the missing data and.
Preparation of Meteorological Input for AERMOD using Malaysian Meteorological Data 1

Tan Yen Chen1, 2, Luqman Chuah Abdullah1, 3, Tan Poh Aun4 Department of Chemical & Environmental, Faculty of Engineering, Universiti Putra Malaysia 43400 UPM Serdang, Selangor MALAYSIA 2

3

ERE Consulting Group Sdn. Bhd.

Institute of Tropical Forestry and Forest Products (INTROP), Universiti Putra Malaysia 4

SOx NOx Asia Sdn. Bhd. [email protected] / [email protected]

Abstract Gaussian plume dispersion model - AMS/EPA Regulatory Model (AERMOD) has been recognised as the preferred regulatory air dispersion model and has been proven to perform better than other available models. However, Malaysian meteorological data has limited parameters and the data recorded is inadequate to be used in AERMOD. Currently, processed meteorological data has to be bought from meteorological data service providers located overseas. The processed data does not represent the real conditions experienced at the site accurately. The study involves the identification of missing data in 4 meteorological stations located in Peninsular Malaysia (Cameron Highlands, Subang, Sepang KLIA, and Kuantan), replacement of the missing data and preparation of the data in accordance with the format that AERMOD requires. The study result in a methodology to replace missing data and calculation using bulk formulae which is developed based on certain assumptions that are practical and scientific.

1. Introduction In Malaysia, more than 200 industrial estates or parks and 18 Free Industrial Zones (FIZ) have been developed. The supportive government policy i.e. liberal equity policy, employment of expatriates and tax incentives has resulted in progressive growth of the industrial sector. More industries and factories to be set up and this will translate into more emissions to be released into the environment. The Environmental Quality (Clean Air) Regulations 1978 under the Environmental Quality Act 1974, is enforced by the Malaysian Government and under the regulations, project initiators need to predict the amount of pollutants generated from their proposed projects and put forward solutions to mitigate and reduce the impact on air quality to the surrounding areas. Air dispersion

978-1-4577-0005-7/11/$26.00 ©2011 IEEE

model is usually used to predict the dispersion of pollutants to the surrounding areas. Currently, in the United States, AERMOD has been introduced as a preferred regulatory model for air dispersion in year 2005 to replace the Industrial Source Complex Short Term Model version 3 (ISCST3) [1]. AERMOD consists of the model itself and two (2) preprocessors, i.e. meteorological preprocessor, AERMET and terrain preprocessor, AERMAP. AERMOD air dispersion model software is still not widely used in Malaysia as AERMET has been designed and functions well with the U.S. meteorological data only. As Malaysian meteorological data collected by the Malaysian Meteorological Department (MMD) are mainly for aviation industry and weather forecasting purposes, thus the data contained limited parameters for air dispersion modelling purpose. Besides that, the data itself contains some defective values (missing data) due to disruptions in the operation of the meteorological instruments. Only data which is at least 90% complete should be used for air modelling purposes [2]. At this moment, modification of data is done overseas by two service providers, Lakes Environmental and Trinity Consultants (Breeze) and a service fee is chargeable. The processed data created by these two providers does not truly represent the conditions at the project site as the meteorological data used by the providers comes from a public domain and are based on certain assumptions from surface data. Therefore, modifications and improvements to suit to local meteorological data are needed. The solution on solving the problem of missing data and production of a practical and scientifically acceptable meteorological data input for AERMOD using Malaysian meteorological data is discussed in this paper.

2. Atmospheric Pollution Components involved in atmospheric pollution are meteorological, emission concentration, and receptors (location in the impact zone). Atmospheric processes, such as movement of air, stability and exchange of heat dictate the fate of pollutants in the air. There are a few factors that affect the air pollution in the atmospheric environment. 2.1 Role of wind The wind speed plays an important role in diluting the pollutants in the atmosphere. When a source of pollutants is emitted into the air, the wind speed will determine how fast the pollutants will mix with the surrounding air and move from the source. The concentration of pollutants will be greatly reduced when strong wind occurs as the turbulence of air promotes mixing of pollutants with the cleaner surrounding air. 2.2 Stability of air Changes of atmospheric stability can cause smoke plumes from chimneys to change during the course of day. These changes due to heat radiation influence the dispersion of pollutants near ground and affect the behavior of smoke leaving a chimney. The pattern of smoke observed across the day till night is fanning, fumigation, looping, coning, and lofting [3,4]. 2.3 Inversion Inversion layer prevents pollutants from escaping into the air above it. If the inversion rises, the mixing depth increases and pollutants can be dispersed through a greater volume which will subsequently reduce the concentration of pollutants. The greatest mixing depth is usually found in the afternoon while the shallowest occurs in the early morning. 2.4 Topography (terrain) The shape of landscape plays an important part in trapping pollutants. Usually, valleys are prone to pollution due to poorly ventilated cold air and movements of pollutants downhill from the surrounding hillsides.

3.1 Derived Parameter Albedo is the fraction of solar radiation reflected by the surface back to space without absorption while the Bowen ratio is an indicator of surface moisture or ratio of sensible heat flux to the latent heat flux. Bowen ratio is used for determining PBL parameters for convective conditions. The surface roughness length is related to the height of obstacles to the wind flow where it is the height at which the mean horizontal wind speed is zero. The selection of values of these 3 characteristics can be found in AERMET user guide [7,8]. Heat flux, H can be computed using Eqs.1 to 3 :-

H =

0.9 Rn

(Eq. 1)

1 + (1 / Bo ) 6

Rn =

4

(1 − r (φ )) R + c1T − σ SBT + c2 n

(Eq. 2)

1 + c3 r (φ ) = r '+ (1 + r ') exp( −0.1φ + b)

(Eq. 3)

Where σSB =Stefan-Boltzmann constant (5.67 x 10-8 Wm-2K-4) R = Solar Radiation (Wm-2) c1 = Empirical constants (5.31 x 10-13 Wm-2K-6) c2 = Empirical constants (60 Wm-2) c3 = Empirical constants (0.12) r(ø)= Albedo n= Cloud cover (0-1) r’ = r(ø=90o) and b=-0.5(1-r’)2. Calculation formulae of surface friction velocity, Monin-Obukhov Length, potential temperature gradient, missing height (mechanical and convective), and turbulent velocity scale can be found in User Guide of AERMOD and AERMET [7,8,9].

4. Malaysian Meteorological Data 3. AERMOD AERMOD is a steady-state plume air dispersion model. In the stable boundary layer (SBL), it assumes the concentration distribution to be Gaussian in both, the vertical and horizontal. In the convective boundary layer (CBL), the horizontal distribution is assumed to be Gaussian, but the vertical distribution is described as a bi-Gaussian probability density function (pdf) [5,6]. AERMOD are able to characterize the PBL through both surface and mixed layer scaling. It requires hourly data such as: (1) wind speed, (2) wind direction, (3) ambient temperature, (4) solar radiation (5) cloud cover values, (6) quantification of surface characteristics (surface roughness, albedo, Bowen ratio), and (7) twice-daily upper air soundings. 2 output files are needed in the AERMOD. The profile output (*.pfl) contains the observations made at each level of a sitespecific tower, one record per level per hour while the surface output (*.sfc) contains observed and calculated surface variable, one record per hour. The default output formats for both files are outlined in Tables 1 & 2 [7,8].

Currently, there are 384 meteorological stations across the whole of Malaysia with 36 of them being principal stations. The principal stations provide monitoring of weather conditions such as rainfall, air temperature, relative humidity, sunshine hour, wind speed and direction, solar radiation and atmospheric pressure. For upper air data observation, MMD currently has 8 stations i.e. Bayan Lepas, Kota Bharu, KLIA (Sepang), Kuantan, Kuching, Bintulu, Kota Kinabalu and Tawau.

5. Result Hourly data for the year of 2009 of 4 meteorological stations (Cameron, Subang, Sepang KLIA and Kuantan) was obtained. It was analysed and missing values were discovered (Table 3). Missing data will affect the accuracy of air dispersion prediction and most of the air modelling tools require 100% complete data, so the missing data must be substituted with reasonable values. The next section describes the assumptions made and their justifications.

Table 1. Default format of profile output for AERMOD Data records: Variable Year (last 2 digits) Month Day Hour Measurement Height (m) Top – 1, if this is the last (highest) level for this hour, or 0 otherwise Wind direction at the current level, WDnn (degrees) Wind speed at the current level, WSnn (m/s) Temperature at the current level, TTnn (oC) Standard deviation of the wind direction fluctuations, SAnn σθ (degrees) Standard deviation of the vertical wind speed fluctuation, SWnn σw (m/s)

Fortran Format I2 I2 I2 I2 F6.1 I1

Columns 1-2 4-5 7-8 10-11 13-18 20

F5.0 F7.2 F7.1 F6.1

22-26 28-34 36-42 44-49

F7.2

51-57

Table 2. Default format of surface output for AERMOD Header record: latitude, longtitude, UA identifier, SF identifier, OS identifier, AERMET version date (format:2(2X,A8),8X,’UA_ID:’,A8,’SF_ID:’,A8,’OS_ID:’,A8,T85,’VERSION:’,A6) Data records: Variable Year (last 2 digits) Month Day Julian date Hour Sensible Heat flux, H (W/m2) Surface friction velocity, u* (m/s) Convective velocity scale, w* (m/s) Vertical potential temperature gradient, VPTG Height of convectively boundary layer, PBL (m) Height of mechanically boundary layer, SBL (m) Monin-Obukhov length, L (m) Surface roughness length, zo (m) Bowen ratio, Bo Albedo, r Wind speed, Ws (m) Wind direction, Wd (o) Reference height for Ws and Wd, zref (m) Temperature, T (K) Reference height for temperature, ztemp (m) 5.1 Upper air data Upper air data collected by instrument called radiosonde contains 2 soundings per day, i.e. 0Z sounding and 12Z sounding. Only data for morning sounding (0Z) is required. Local soundings are only recorded for a maximum of 15 levels as compared to U.S. data which is 79 levels. AERMOD requires a level of up to 5000 m, therefore, only 5 levels for each day are available from the local meteorological data. If there is any missing value to the 5 levels data, it is assumed that records of the previous day will be used to replace it.

Fortran Format I2 I2 I2 I3 I2 F6.1 F6.3 F6.3 F6.3 F5.0 F5.0 F8.1 F6.3 F6.2 F6.2 F7.2 F5.0 F6.1 F6.1 F6.1

Columns 1-2 4-5 7-8 10-12 14-15 17-22 24-29 31-36 38-43 45-49 51-55 57-64 66-71 73-78 80-85 87-93 95-99 101-106 10896-101

The levels of sounding must be transformed into hour scale. According to MMD, radiosonde is released at 8 a.m. every day. The 5 levels data available are assumed at hour 8, hour 11, hour 14, hour 20 and hour 6. It is also assumed that hour 7 for temperature, wind direction and wind speed is taken from surface data and is used as the start for each day. For the balance 19 hours, linear interpolation for parameter temperature and height can be used as theoretically, temperature decreases with height, from 0 km to 11 km at troposphere layers. No interpolation was carried out for wind speed and wind direction for the 19

levels and it is replaced with values 999 (wind direction) and 99.90 (wind speed). The standard deviation for wind direction fluctuations and vertical wind speed fluctuations is assumed and replace with 99.0 and 99.00. MMD have only 8 stations that record upper air data. Therefore, it is assumed that the upper air data of the nearest station to the site is to be used based on the fact that, generally, the weather patterns in Malaysia which located near the equator and in the doldrums are quite similar throughout the years. Table 3. Percentage of Missing Data Missing data (%) Parameter

Cameron

Subang

Sepang KLIA

Kuantan

(a) Surface data Pressure N/A Dry Bulb Temp. 1.04 Wind Direction 0.16 Wind Speed 0.16 Cloud Cover 45.83 Solar Radiation N/A 0.33 6.28 1.72 (b) Upper Air data Pressure 10.78 15.64 Height 12.48 17.23 Dry Temperature 12.62 17.76 N/S N/S Dew point Temp. 19.84 24.57 Wind Direction 10.78 15.64 Wind Speed 10.78 15.64 Notes: “-” means no missing data N/A means data was not supplied by MMD N/S means no upper air data was collected at the station

5.2 Surface Data Prior to further processing of surface data, all the missing indicators for all hours are replaced. For every single missing data, the average value is calculated from the previous and subsequent hours. For 2 or more sequences of missing data, replacement is taken from the previous day’s value. For wind speed and wind direction data that is missing, wind speed is assumed as 0 ms-1 and 0o for wind direction [7]. For cloud cover, the missing data is replaced with value 4. Calculation of heat flux for CBL is dependent on solar radiation, temperature, cloud cover, albedo and Bowen ratio. In order to fit into the equations, the measurement unit of solar radiation needs to be changed from MJm-2 to Wm-2h-1. This also applies to temperature (from oC to K), and cloud cover (from basis of 8 to basis of 10). Albedo and Bowen ratio values are characterized from the land use surrounding the site. It is assumed that the sun rises in Peninsular Malaysia at 7.00 in the morning and the sun sets at 7.00 in the evening. Albedo value changes across the daytime and according to the location of the sun. The angle of sunrise and sunset is assumed at 30o and is at 90o at 1.00 p.m. (Fig. 1). As neither sun radiation is received from nor transmitted out to the atmosphere at night time hours, an albedo of 1 is assumed.

10 9

8 7

60o 50o 40o

30

11 70

o

13

12 80

o

90

14

15

o

80

16

o

70

o

60o o

17 18

50 40o

o

30

19 o

Fig. 1: Angle of sun related to time The heat flux and u* for the SBL are computed using bulk formulae. These calculations need surface roughness value, wind speed, and the height of wind measurements. Surface roughness depends on land use while the height of wind measurement is 10 m (the meteorological instrument is located at this level). Wind speed of below 0.2 ms-1 is replaced to 0.2 ms-1 to avoid mathematical error. The value of heat flux for SBL is taken for night time while the value of heat flux of CBL is taken for day time. Monin-Obukhov value for SBL is calculated using the heat flux of SBL. For u* and L of CBL, the calculation is applied to the positive value of heat flux. As AERMET did not specify the formula for calculating the potential temperature gradient, it is assumed that the value is calculated from the upper air profile data using the equation below: d θ θ 2 − θ1 = (Eq. 4) dz z 2 − z1

g z  cp 

θ =T +

[10]

(Eq.5)

The minimum value of potential temperature gradient is fixed at 0.005 Km-1. For the convective mixing height, the time of early sunrise is assumed to be at 6.00 a.m. and value of potential gradient from upper air data is used instead. The growth of zic and zim are restricted to 4000 m. Table 4 summarises all the assumptions used in the preparation of meteorological data for AERMOD. 5.3 AERMOD versus ISCST3 A complementary test has been carried out to compare the air dispersion between AERMOD and ISCST3 by using the processed meteorological output derived from the above methodology (for AERMOD) and methodology developed by air modellers (for ISCST3) [8]. The test consists of 100 runs (single source) with random samples with different emission rates, pollutant types, stack elevation, gas exit temperature, plume velocity, stack diameter, urban or rural environment. 70% of the samples were allocated as point source with normal emission, 20% as point source abnormal emission and the balance 10% as area source. All the 100 emissions were modelled with the same condition of flat terrain, default regulatory model at the 10 km x 10 km Cartesian grid. 10 discrete receptors were randomly chosen to obtain the maximum average

incremental concentration (MAIC) at the particular location. The results using Cameron meteorological data were plotted for a graphic view as shown in Fig. 2. Other 3 meteorological data has similar result pattern with Cameron. The result was arranged in group of

pollutants scenario (CO, NO2, PM10, SO2, TSP, and area source) and according to rural and urban setting. For every emission source scenario, the MAIC is presented in 2 time scales which are according to the time scale available in the Recommended Malaysia Ambient Air Quality Guidelines (RMAQG).

Table 4. Assumptions used in Meteorological Data Preparation for AERMOD Parameters (a) Profile data Sounding Levels, temperature, wind speed, wind direction

Assumptions

• • • •

0Z sounding to be used 5 levels at the range of 50 m to 6000 m to be used Missing data in the 5 levels is replaced from the previous day record and are assumed at hour 8, hour 11, hour 14, hour 20 and hour 6 Hour 7 data is obtained from the surface data record for temperature, wind speed and wind direction Linear interpolation to replace the 19 levels Wind speed and direction in the 19 levels are replaced with 99.0 and 999. Missing indicator of 99.0 and 99.00 for all hours Nearest to the site to be used

• • • • • •

Nearest to the site to be used Averaging for single missing data Replaced with the previous day’s record for 2 or more sequences of missing data 0 ms-1 and 0 o for missing wind speed and temperature Value of 4 for missing cloud cover Values taken based on land use surrounding site

• • • • • • •

Change to basis of 10 Sunrise at 7.00 a.m.; Sunset at 7.00 p.m. Angle of sunrise and sunset are at 30o while 90o at 1 p.m Wind speed below 0.2 ms-1 is changed to 0.2 ms-1 Iteration only applies to positive heat flux Values taken from upper air data; Minimum of 0.005 Km-1 Early sunrise at 6 a.m.; Maximum of 4000 m

• • •



Standard deviation Stations (b) Surface data Stations Missing data

Albedo, Bowen ratio, surface roughness Cloud cover Heat flux u* and L VPTG Zic and Zim

Fig. 2: Percentage of different between AERMOD and ISCST3 for Cameron

It was observed that most of the dispersion for the urban scenario had a negative percentage and positive percentage of less than 50% as compared to the rural scenario. Meanwhile the entire area source scenario had a positive percentage of more than 50%. This shows that AERMOD prediction for MAIC is lower for urban setting and higher for rural scenario and area source dispersion when compared to ISCST3. Based on the results, AERMOD has a higher prediction of MAIC value compared to ISCST3 due to the lower convective and mechanical height calculated from the meteorological data preparation for AERMOD as compared to ISCST3. In ISCST3, the mixing heights were calculated from the wind speed that had been modified to 1 ms-1. Meanwhile, the lowest wind speed for AERMOD calculation was 0.2 ms-1. Thus, ISCST3 had a higher mixing value. Both mixing height which reflect the stability of the atmosphere, will affect the dispersion pattern of pollutants. AERMOD generally predicted lower MAIC than ISCST3 for urban dispersion as the anthropogenic heat flux from the urban population was integrated in the AERMOD algorithm. Therefore, heat flux will be higher in an urban area and create an unstable atmosphere for mixing.

7. Conclusions A methodology of missing data replacement and calculation using bulk formulae for meteorological data preparation in AERMOD using Malaysian meteorological data has been successfully developed. The methodology is only valid for air quality dispersion modelling without precipitation effect and is developed based on certain assumptions that are practical and scientific. Complimentary test have been carried out to compare the prediction of AERMOD as compared to ISCST3. AERMOD consistently predicted the MAIC lower in urban scenario while higher in rural and area source scenario when compared to ISCST3. It was also observed that AERMOD simulates in a way similar to the real meteorological condition, i.e. wind speed and wind direction.

Acknowledgement Authors would like to express sincere gratitude to World Federation of Scientists, WFS for the scholarship provided leading towards the successful completion of this research work. We appreciate the free access of meteorological data provided by the Malaysian Meteorological Department.

References [1] Perry, S.G., Cimorelli, A.J., Paine, R.J., Brode, R.W., Weil, J.C., Venkatram, A., Wilson, R.B., Russell F.L., Peters, W.D., AERMOD: A dispersion model for industrial source applications. Part i: General model formulation and boundary layer characterization, J. Appl. Meteor, Vol. 24, (2004), pp.682-693. [2] Dennis, A. and Russell, F.L., Procedures for Substituting Values for Missing NWS Meteorological

Data for Use in Regulatory Air Quality Models, (1992), pp.1-4. [3] Ahrens, C.D., Meteorology today: An Introduction to Weather, Climate, and the Environment, U.S.A, Brooks/Cole, 2000, Chap. 17, pp.441-467. [4] Pasquill, F., Atmospheric Diffusion: The Dispersion of Windborne Material from Industrial and other Sources, New York, Halted Press, 1974, Chap. 6, pp.329-400. [5] Willis, G.E. and Deardorff, J.W., A laboratory study of dispersion in the middle of the convectively mixed layer, Atmos.Environ.,Vol. 15, (1981), pp.109-117. [6] Briggs, G. A., Plume dispersion in the convective boundary layer. Part II: Analysis of CONDORS field experiment data, J. Appl. Meteor., Vol. 32, (1993), pp. 1388-1425. [7] U.S. EPA., User’s Guide for the AERMOD Meteorological Preprocessor (AERMET), EPA-454/B03-002, North Carolina, U.S. Environmental Protection Agency, 2004. [8] Tan, YC., Meteorological Data Preparation for Gaussian Plume Air Dispersion Model (AERMOD), Selangor, Universiti Putra Malaysia, 2010. [9] Cimorelli, A.J., Perry, S.G., Venkatram, A., Weil, J.C., Paine, R.J., Wilson, R.B., Russell, F.L., Peters, W.D., Brode, R.W., Paumier, J.O., AERMOD: Description of Model Formulation, EPA-454/R-03-004, North Carolina, U.S. Environmental Protection Agency, 2004. [10] Stull, R.B., An Introduction to Boundary Layer Meteorology, Netherlands, Kluwer Academic Publishers, 1988.

Suggest Documents