"On Evaluating Compliance with Air Pollution Levels Not to Be Exceeded More than Once a Year"

Abstract

Neustadter and Sidik¹ have presented a critique of the computational procedures available for estimating expected ambient air pollutant concentrations from empirical data. In particular, the paper addresses two statistical considerations in using a mathematical model developed by Larsen and presented for use by the U. S. Environmental Protection Agency.² The authors have presented a numerical analysis of the errors introduced into Larsen’s estimation procedure by (i) the interchange of the order of applying exponentiation and expectation, as described by Patel³; and (ii) the use of the expected annual maximum pollutant concentration, instead of the expected annual second-highest level (SMP), for planning purposes.