fect on algal populations. A die-off of blue-green algal populations was observed in Liberty Lake in eastern Washington following the ashfall ( 8 ) , whereas no significant changes in blue-green algal populations have been observed in Lake Lenore and Moses Lake in the center of the ashfall area in central Washington (9).
(4) Guillard, R. R. L. “Culture of Marine Invertebrate Animals”; Smith, W. L., Chanley, M. H., Eds.; Plenum Press: New York, 1975. (5) Morel, F. M. M.; Rugter, J.G: Jr.; Anderson, D. M.; Guillard, R. R. J . Phycol. 1979,15, 135. (6) Strickland, J. D.; Parsons, T. R. “A Practical Handbook of Seawater Analysis”; Bull., Fish. Res. Board Can. 1972,167, 311. ( 7 ) McKnight, D. M.; Morel, F. M. M. Limnol. Oceanogr. 1980,25, 62. (8) Funk, W. H., Washington State University, personal communication, 1980. (9) Edmondson, W. T., University of Washington, personal communications, 1980.
Literature Cited (1) Taylor, H. E.; Lichte, F. E. Geophys. Res. Lett., in press. (2) Pereira, W. E.; Rostad, C. E.; Taylor, H. E. Geophys. Res. Lett., in press. (3) Kurenkov, I. I. Limnol. Oceanogr. 1966,II, 426.
Received for review August 4,1980. Accepted November 20,1980.
CORRESPONDENCE
SIR: In their paper, “Determination of Vehicle Emission Rates from Roadways by Mass Balance Techniques” [ES&T 1980,14, 7001, Bullin et al. have attempted to determine the emission rates of pollutants from vehicles by a simple scheme and have compared the results of their “validated” method to the results from the EPA emission models, AP-42 and MOBILE 1 (I, 2 ) . While the authors accurately point out the shortcomings of the methodology used in AP-42 and MOBILE 1, the mass balance scheme is no panacea. It is unfortunate that the authors did not expound on the obvious and not so obvious limitations of this scheme and its applications, which are serious enough to invalidate their conclusions. The following discussion will bring out the major limitations of their methodology. For a continuous source, if the x coordinate is taken along the downwind direction, the equation for a mass balance is
Q=
Jms-~
CU dz dy
(1)
where Q is the emission rate, C is the measured concentration, and U is the wind velocity. The equation used by the authors
Q=
sm 0
CU, dz
(2)
where U , is the crossroad wind speed, is a valid simplification of the general equation if there is no concentration gradient in the direction parallel to the road. Clearly, this simplification is strictly valid for winds exactly perpendicular to the roadway. In practice, however, the method could be valid for situations where the cross-wind dispersion can be ignored. This assumption requires that the line source have no edge effects. As the wind direction deviates from normality to the roadway segment, the use of eq 2 will result in inaccuracies in the emission rate estimation. This equation predicts zero emission rate when the wind blows parallel to the roadway! These considerations also suggest that the relative distance of the monitor from the roadway will be equally important. Before discussing the applicability of this approach to parallel wind-road orientation angles, it is informative to see how the method performs under perpendicular cases. Even though one can expect eq 2 to hold under perpendicular wind-road orientation cases, the modifications of the flow field adjacent to the roadway, due to the moving traffic, make the application of this method subjective. For example, the authors seem to have used the wind speeds at each of the downwind locations where concentrations were measured, although they do not clearly state this in the paper. If it is so, these wind speeds are significantly affected at the lower levels by the moving traffic as shown by Rao et al. ( 3 , 4 )and Sedefian 364
Environmental Science & Technology
et al. ( 5 ) from the New York and GM experiments. Furthermore, if the winds at the nearest roadside tower are used to estimate the emission rate, the wind-road angle becomes a difficult parameter to determine since the above studies also showed a wind direction shift up to 90° to the wind direction recorded upwind. The authors do not even address these problems. How is one to determine whether the plume is entirely “defined” within the heights of observation as required by the proposed method? The observed concentration profile shape changes from an approximately exponential one to an approximately linear one within 10-m distance from tower 1to tower 2 of the GM data. Does one assume 10%of the maximum concentration to indicate an upper bound to the plume? If there are only two measurements within this height, should these two points be connected by a straight line or should a curve fitting be used? The observed plume dimensions can vary in a nonconservative manner (with respect to the proposed mass balance equation) due to the complex turbulent and mean flow structure adjacent to the highway. The above considerations are even more important in situations where a background concentration of a pollutant, such as CO, exists. Even if this background is assumed to be vertically and horizontally uniform on the upwind side (one needs to assume this if only one monitor exists on the upwind side), it can be easily shown that JmCzL1,2 dz - JmCili,i dz Z Jm(Cz - CdU, dz
(3)
where subseripts 1 and 2 refer to the upwind and downwind locations. The above relation is true regardless of which wind speed is used in the integral on the right-hand side. All of the above considerations suggest a tremendous amount of subjectivity in estimating the emission strength from the proposed methodology. We used a few cases of data presented in their Table I to determine the emission rates graphically from eq 2 by making assumptions such as (a) exponential or linear profile of CU,l, (b) upwind or downwind wind speeds at all heights or at the height of plume release, (c) correcting and not correcting for the angle of the tower line with respect to the roadway (since it is assumed in the method that there are no concentration gradients perpendicular to the flow), and (d) certain combinations of the above assumptions. The results indicated a factor of more than 2 a t the extremes of the estimated emission rates. It is clear that their method, even if applicable, is limited to situations where the emissions are uniform along the roadway so that there are no cross-wind gradients of concentration and, thus, the calculated emission rate per unit length of roadway will be meaningful. For example, if the mass bal0013-936X/81/0915-0364$01.25/0
@ 1981 American Chemical Society
ance scheme indicated that MOBILE 1overestimated emissions by 50% for a given traffic condition, how would the scheme be used to estimate the eI%ssions for a different traffic pattern? Although AP-42 or MOBILE 1 provides only approximate emission rates, they are capable of forecasting the changes in emission rates with changing traffic patterns. Furthermore, the approximate emission rate is not too critical in the model calibration procedure since model estimates are linearly proportional to the emissions. This can be incorporated in projecting future pollutant levels once a good correlation between measured and the corresponding observed concentrations is established. The correlation coefficient is related to the model’s handling of the physical processes of atmospheric transport and dispersion. The data presented in their Table I show that, although the average calculated and measured emissions agree well, the variability is anywhere from 39% underestimation to 32% overprediction at tower 1.The errors are even larger at tower 2. The calculations with the SRI data base show (Tables I1 and 111) considerable disagreement between calculated and measured tracer gas emission rates. There were significant wind directional shifts between many of the hours during which the SRI experiments were conducted (6). One consequence of using these periods without subdividing the hour into shorter intervals is that it gives the impression of having significant upwind concentrations when, in fact, part of the hour had those receptors downwind, while they were upwind for the balance of the hour. Further, there were several hours when small tracer concentrations were seen upwind of the highway. These consistently occur with light winds and reflect either the natural meandering of the wind under these conditions or the influence of the vehicle-induced drag flows. The authors do not indicate whether the data used in the paper excluded the above situations. The emission rates for CO data were calculated (see Tables IV-VII) from CO measurements at two heights only. In addition, if these towers are sufficiently far from or very near the highway, significant errors will occur in the estimation of the emission rate. In the computation of the emission factors from
SIR: We believe that the comments by Rao, Sedefian, and Peterson in their letter to you are largely due to misinterpretation of the method used and the intent of the paper. In fact, many of the items that the correspondents said were omitted from the paper are very clearly included. We did not present the method or the paper as a panacea. We point out in the paper that much more data are needed to adequately describe vehicular emission rates. This method is the only known means for checking MOBILE 1, and we think it deserves further consideration and work. In response to the second paragraph, a t no time did we suggest that the mass balance method be applied to parallel wind situations. For nonparallel wind situations, edge effects have no influence at all on the method if they are uniform. End effects would be very important, but the paper clearly states that the roadway must be effectively infinitely long and straight. Thus, the simplification shown as eq 2 in the correspondents’ letter is valid. The wind angle with respect to the roadway is included for each case in the paper. The wind angle had a small standard deviation and the speed was moderately strong for essentially all cases. In the Texas data, the background values at both locations on the upwind tower had to be within 1 ppm of each other. For the tracer gas data from 0013-936X/81/0915-0365$01.25/0 @ 1981 American Chemical Society
AP-42 or MOBILE 1,county-wide traffic information instead of on-site traffic information was used which could result in a considerable error in the emissions model predictions. It is unclear how such large differences in calculated emission/ vehicle could occur in Table IV. The mass balance technique as given in eq 1is a theoretically sound method to determine emissions from the roadways, but numerous problems arise in its application. The methodology proposed by the authors with eq 2 has very limited practical applicability. Because of the various limitations to the proposed methodology, and the way it has been applied to estimate source strengths for CO, the authors’ contention that the emission factors are underpredicted 43-60070 from AP-42 and 8-40070from MOBILE 1is unjustified. Literature Cited (1) U.S. Environmental Protection Agency, Office of Air Programs,
Research Triangle Park, NC, 1975, AP-42. (2) U.S. Environmental Protection Agency, Office of Transportation and Land Use Policy, Washington, D.C., 1978, Report No. EPA40019-78-006,MOBILE 1. (3) Rao, S. T.; Sedefian, L.; Czapski, U. H. J.App2 Meteorol. 1979, 18, 283. (4) Rao, S. T.; Keenan, M. T.; Sistla, G.; Wilson, J. S. “Atmospheric Turbulence and Pollutant Dispersion Near Roadways”; Final Report to EPA on Project R-806017-01. (5) Sedefian, L.; Rao, S. T.; Czapski, U. H. Atmos. Enoiron., in press. (6) Dabberdt, W., personal communication.
Leon Sedefian S. Trivikrama Rao Division of Air New York State Department of Environmental Conservation Albany, NY 12233
William B. Petersen Meteorology and Assessment Division U S . Environmental Protection Agency Research Triangle Park, NC 2771 1
GM or SRI, no cases were used where the far upwind monitors showed any of the tracer. In response to the third paragraph, if the upwind and downwind monitors are sufficiently far from the roadway, the modifications of the wind flow field will be negligible. In the Texas data, the upwind station was usually -150 ft from the upwind edge of the roadway and the downwind station used in the mass balance calculations was usually -50 f t from the downwind edge of the roadway. In the GM data, the towers were -50-100 ft from the respective road edge. We think the correspondents will agree that the modifications in the wind flow field due to the roadway will be negligible at this distance. However, in the truest sense, the wind angle shift across the roadway has no bearing on the problem, as long as the wind speeds are taken at the points of pollution measurements. The paper stated this quite clearly, just as i t stated that the meteorological instruments were located near the downwind pollutant monitors. We did not try to take a mass balance in the mixing cell, as the letter suggests, since the effective wind in this region is too highly turbulent to even properly assess wind direction. In response to the fourth paragraph, the mass flux profile, CU,, for all cases was such that a well-behaved curve through Volume 15, Number 3, March 1981
365
