Chapter 4: The LOTUS regression model

One of the primary motivations of the LOTUS effort is to attempt to reconcile the discrepancies in ozone trend results from the wealth of literature on the subject. Doing so requires investigating the various methodologies employed to derive long-term trends in ozone as well as to examine the large array of possible variables that feed into those methodologies and analyse their impacts on potential trend results. Given the limited amount of time, the LOTUS group focused on the most common methodology of multiple linear regression and performed a number of sensitivity tests with the goal of trying to establish best practices and come to a consensus on a single regression model to use for this study. This chapter discusses the details and results of the sensitivity tests before describing the components of the final single model that was chosen and the reasons for that choice.