|
|
|
SIP suit reportATTAINMENT DEMONSTRATIONSection 182(c)(2)(A) of the Act requires that serious ozone nonattainment areas submit an attainment demonstration to demonstrate that attainment will be achieved by the November 15, 1999 attainment date. The EPA recognizes this requirement in 64 FR 35930, Approval and Promulgation of Air Quality Implementation Plans, Louisiana, and in the technical support document (TSD) for 64 FR 35930. Both 64 FR 35930 and the technical support document state that the attainment demonstration must be done in conformance with the EPA's guidance document titled, "Guidance on Use of Modeled Results to Demonstrate Attainment of the Ozone NAAQS" (attainment guide). The State of Louisiana submitted the required attainment demonstration to the EPA as part of the State Implementation Plan. The attainment demonstration consists of the computer modeling for three ozone exceedances that occurred in the Baton Rouge area. The exceedances are first modeled to determine if the computer model can accurately predict the observed ozone levels. This is called the base case. The exceedances are then modeled using emissions estimates for the attainment date in 1999. The objective is to have the 1999 models demonstrate that the area will not exceed the ozone standard once the emissions reductions in the State Implementation Plan (SIP) has been put into effect. Attainment Demonstration Results The attainment guide gives two methods for demonstrating attainment, the Deterministic Approach and the Statistical Approach. The attainment demonstration submitted by the State of Louisiana could not pass the Deterministic Approach, as is stated in 64 FR 35930 and the TSD. The Deterministic Approach allows a demonstration to pass if the demonstration results show that the ozone levels are expected to be below the ozone standard of 120 parts per billion (ppb) by the attainment date. The Louisiana demonstration results showed that this is not possible and therefore Louisiana chose the Statistical Approach to passing the attainment demonstration. The Statistical Approach requires that attainment demonstration pass three "Benchmark" tests. The Louisiana demonstration did not pass any of the three tests. 64 FR 35930, the TSD and the Louisiana SIP all agree that Benchmark #2 and Benchmark #3 were not passed. Benchmark #1 was also not passed, though 64 FR 35930, the TSD and the Louisiana SIP state that it was passed. This discrepancy is due to the failure of Louisiana to correctly apply Benchmark #1. When Benchmark #1 is correctly applied it does not pass the requirements in the attainment guide. This means that none of the required Benchmarks were passed in Louisiana's attainment demonstration and therefore the attainment demonstration does not meet the requirements of section 182(c)(2)(A) of the Act. The primary requirement of section 182(c)(2)(A) is that the attainment demonstration must show that the SIP will provide for attainment of the ozone standard by the attainment date. Louisiana's attainment demonstration fails the EPA's tests in the attainment guide and instead shows that the current SIP will never provide for attainment of the ozone standard. The application of the attainment demonstration to Benchmark #1 is clearly described in the attainment guide. Louisiana failed to correctly implement the attainment demonstration as required in Section 4.3.3 and 4.3.4 of the attainment guide. Neither 64 FR 35930 nor the TSD took this failure into account. When the Louisiana demonstration is correctly applied under Section 4.3.3 and 4.3.4 Benchmark #1 is not passed. Weight of Evidence The attainment guide has provisions for applying a Weight of Evidence determination when the statistical test is not passed. The EPA used this provision to pass the Louisiana attainment demonstration. However, passing this attainment demonstration based on a Weight of Evidence determination is a serious breech of the Weight of Evidence guidelines in the attainment guide and should never have been allowed. The attainment guide goes to great length to make the point that Weight of Evidence determinations are of greater importance in the deterministic approach and should only be used in the statistical approach when "minor exceptions to the benchmarks exist." (attainment guide, page 19) This is because the Statistical Approach is a much more rigorous test than the Deterministic Approach and "already includes a substantial effort to consider corroborative evidence." (attainment guide, page 19) The Louisiana demonstration's exceptions to the benchmarks are major, not minor. In fact none of the benchmarks are passed with only a few of the failed tests coming close to passing. Further, the Weight of Evidence arguments used by the EPA to pass the Louisiana demonstration instead show that the ozone problems are not getting better and that the SIP will not provide for compliance with the ozone standard by the attainment date. Benchmark #1 Results. Benchmark #1 was incorrectly applied by Louisiana and incorrectly reviewed by the EPA. The failures of Louisiana and the EPA center around the fact that all three of the modeled episodes failed to meet the ozone standard and that two of these showed the ozone standard being exceeded in the same "Subregion of the Model Domain". These failures of the model are acceptable under Benchmark #1 so long as the modeled days were days that had weather conditions very conducive to ozone formation. The severity of the ozone forming potential of any day is discussed on pages 12 through 23 of the attainment guide. Using this standard, two of the modeled days are not considered severe. In this case, neither of the non-severe modeled days can show an exceedance of the ozone standard and no two of the three modeled days can show an exceedance in the same Subregion. Both of the non-severe days showed exceedances of the ozone standard and two of the three days showed exceedances in the same Subregion. Therefore, the Louisiana demonstration fails Benchmark #1 by a wide margin. The attainment guide allows a state to use a different method to determine when days have a severe ozone forming potential than that used in the attainment guide. In order to use a different method, as Louisiana did, the state must meet the criteria in Sections 4.3.3 and 4.3.4 of the attainment guide. Louisiana did not use a method that met these criteria nor did the EPA check to see if the method proposed by Louisiana met the criteria in the attainment guide. Instead, the method used by Louisiana is so lax that it doesn't produce a statistical mean, an average, a standard deviation or the R2 value required in criteria 4) of Section 4.3.4. Also, the method used by Louisiana can't calculate the variance described in criteria 3) and the Louisiana method can't be compared to the EPA proposed method as required in criteria 3). In addition, the requirements of criteria 1) and 2) were not met by Louisiana nor were any of the requirements of Section 4.3.3 and 4.3.4 discussed or applied by the EPA when evaluating or approving the demonstration plan. This is a blatant failure of the EPA to adhere to their own requirements. The attainment guide requires that Louisiana have an R2 value >0.65, Section 4.3.3. Louisiana didn't give the R2 value derived from the recommended ranking procedure in the SIP or any subsequent documentation. Louisiana also didn't give the R2 value for the alternative procedure, as required. The attainment guide states that if the R2 value is below 0.65 "the Deterministic Approach be used in the attainment demonstration", with the results from the statistical tests used as Weight of Evidence for the Deterministic Approach. This procedure was not followed by Louisiana nor was it discussed by the EPA. The lax method Louisiana chose for assigning ozone forming potential to the modeled days is a very poor choice. This method isn't anything like the EPA recommended method, it doesn't meet the EPA's standards for Alternative Ranking Procedures in the attainment guide and it should never have been accepted by the EPA as an alternative procedure. Benchmark #2 Results. All parties agree that none of three modeled days meets the requirements of Benchmark #2. Two of the days are serious failures and are far removed from meeting the benchmark. These two flagrant failures are more pronounced when the EPA's method of ranking days based on ozone forming potential is used. Using the lax Louisiana method does move one of these serious failures closer to meeting the Benchmark, but even using this inaccurate method the day still fails Benchmark #2. The Louisiana demonstration fails Benchmark #2 so completely and by such a large margin that it becomes incredulous when the EPA considers these failures to be minor exceptions in their the Weight of Evidence determination. Considering these major failures of the benchmark as minor exceptions was the only way the EPA could pass the Louisiana attainment demonstration, so minor exceptions they became despite the major failure to meet the Benchmark. Benchmark #3 Results. Benchmark #3 was failed by one of the modeled days, August 19, 1993. The EPA deftly fails to consider this a failure because the attainment guide states that Benchmark #3 doesn't have to be applied unless the model underpredicts the actual ozone measured on that day. According to Louisiana none of the modeled days were underpredicted. Again, Louisiana failed to properly apply the EPA's attainment guide and the EPA failed to correct Louisiana in its use of their own guidelines. Louisiana didn't check for underperdictions in accordance with the attainment guide, page 18, Section 4.1.3. This section requires that measured ozone values be compared to predicted ozone values in the model. Instead of following the guidelines, Louisiana compared the maximum predicted ozone value to the maximum measured ozone value. This value is called the Mean Unsigned Error, see the Louisiana SIP, Appendix M, page 3-1. The proper way to evaluate benchmark #3 is to use the Average Accuracy of the Peak Concentrations, see Appendix M, page 3-2. This showed that maximum ozone was underpredicted for both August 16, 1989 and May 25, 1990, see pages 3-6 and 3-8. Louisiana submitted no technical analysis for the August 19, 1993 case. The problem with measuring underprediction in the manner Louisiana used is that the maximum predicted ozone value will seldom if ever be at or near an ozone monitor. If the model predicts ozone values correctly then the maximum measured value would be at the same place the maximum predicted value is. However, if there is no monitor there, as will almost always be the case, the model will appear to overpredict ozone every time. In this way, predicted ozone is being compared to measured ozone values at a location far removed from where the maximum ozone reading would have occurred. As any good scientist would, the EPA chose a much more reasonable method for comparing actual to predicted ozone values. Section 4.1.3 of the attainment guide states that the comparisons should be used to determine if the model underpredicts "observed ozone concentrations". This can only be done by comparing the observed values to the predicted values at the same location. Louisiana turned this logic around and tested to see whether the maximum predicted ozone concentration was higher than the ozone concentrations measured at the distant ozone monitoring stations. This method will inevitably lead to the appearance that the models overpredict ozone. The application of Benchmark #3 also brings up a problem with the EPA's analysis of the Louisiana demonstration. Louisiana didn't include a technical report for the August 18-19, 1993 episode. Therefore, the EPA couldn't have had sufficient data to independently evaluate the August 18-19 demonstration results. The EPA admits that they didn't review the technical report in its Errata to the 64 FR 35930 TSD. The technical support documents for the August 15-16, 1989 and May 24-25, 1990 demonstrations are included in the SIP submittal. These show that the model actually underpredicts ozone for the two episode days August 16 and May 25. Without the technical document for the August 19 episode it is impossible to tell whether ozone is underpredicted or not. Regardless, Benchmark #3 was not met for the August 19, 1993 episode. Application of the Weight of Evidence Despite the obvious failure of the Louisiana attainment demonstration to pass the statistical test the EPA approved the attainment demonstration based on a Weight of Evidence argument. This argument is presented in the EPA's TSD to 64 FR 35930. The TSD lists four specific factors that led the EPA to conclude that "attainment is likely despite model results which may not pass the statistical test". (TSD page 38) These four factors are stated on page 38 following the phrase "Specifically, EPA considered the following factors:". Factor 1 is, "Air quality in the Baton Rogue ozone nonattainment area for the last several years has shown a steady improvement toward attaining the ozone NAAQS." This statement is not accurate and the EPA presented no documentation to support this statement. In fact, there has been no improvement in the ozone problem through the 1990's. Figure 1 shows the number of ozone exceedances in each year since 1991. The line is a curve-fit of the data, which shows the number of exceedances has been rising throughout the 1990's. Figure 1 shows there has been no improvement in the ozone problem and that attainment by November 15, 1999 was highly unlikely. Figure 1. The number of ozone exceedances in the Baton Rouge nonattainment area from 1991 to 1998. The line is a power law fit of the data. Figure 2 shows the maximum annual ozone reading in the Baton Rouge nonattainment area since 1991. This shows that the magnitude of the problem has also not improved and that in addition to no improvement in the number of ozone days there has been no improvement in the maximum ozone readings either. Figure 2. The maximum ozone reading in the Baton Rouge nonattainment area from 1991 to 1998. The line is a power law fit of the data. The EPA's factor 1 in their Weight of Evidence demonstration does not show the ozone problem improving and it gives no indication that attainment by the applicable date is probable. Factor 2 states, "The ozone design value has dropped significantly from the 1990 design level of 168 ppb." However, the EPA TSD then contradicts factor 2 by stating "Although, it has been as low as 128 ppb in 1993, the design value currently stands at 139 ppb (based on 1997 air quality data)." The EPA admits that the ozone problem became considerably worse from 1993 to 1997, and with this we agree. The rise in the design value throughout the mid-1990's shows that attainment by the November 1999 attainment date was not expected nor probable. This alone should have been ample proof that a Weight of Evidence demonstration does not support attainment by the attainment date and that the attainment demonstration should never have been approved. Factor 3 sites the results of ozone precursor emission studies in the summer of 1996. Comparisons are made to precursor emissions from 1985. There is no doubt that precursor emissions dropped from 1985 to 1996 in a manner similar to the drop in ozone from 1985 to 1996. However, the EPA gives no evidence that 1996 precursor emissions levels are low enough that attainment of the ozone standard would be expected. In fact, ozone data from 1996 and 1997 show that attainment is not likely despite the drop in precursor emissions. Precursor emissions are not as important as the ozone data discussed in factors 1 and 2. There is also more ozone data available than precursor data. The EPA's arguments concerning precursor data merely shows how far they are willing to reach to approve the Louisiana SIP. Factor 4 states that "the number of exceedance days has also been on a generally downward trend, going from as many as 15 exceedance days in 1990 to as low as 4 exceedance days in 1996." The EPA next contradicts their position by discussing why the "spike of 11 exceedance days that occurred in 1995" should not be considered important. The EPA also failed to explain that the number of exceedance days increased to 8 in both 1997 and 1998. This should have been a clear indication that the ozone problem was heading in the wrong direction and that attainment by 1999 was highly unlikely. Louisiana also gives some factors that support the Weight of Evidence argument in the SIP. These include uncertainty in the boundary conditions. What they didn't mention is that the uncertainty could be in either direction, higher boundary conditions or lower boundary conditions. Instead, the State only tested the case of lower boundary conditions and ignored the possibility of higher boundary conditions. End Back to Air Projects page.
|