The AMCL for radon is defined as the concentration of radon in water that will contribute to indoor air a radon concentration equal to the national average ambient-air concentration. In effect, the AMCL is defined implicitly by the following relationship:
where TF represents the water-to-indoor-air transfer factor, and Mambient is the national average ambient-air concentration. Note that Mambient is a single number that is unknown but that can be estimated with some degree of uncertainty (see chapter 2). The AMCL is also a single number that is to be determined. The role of the transfer factor, TF, in this relationship is less clear, because it is not simply a single number. In fact, the TF is subject to both variability (that is, variation from dwelling to dwelling and over time for a given dwelling) and uncertainty (that is, the distribution of TF over the population of all dwellings is unknown). The legislation that mandates the derivation of an AMCL does not specify how the TF is to be derived. The committee chose to interpret TF as the mean transfer factor (MTF) that is, as a single numerical quantity that, like Mambient, is unknown but can be estimated with uncertainty:
The AMCL derived from this relationship has the property that water with radon at the AMCL will, on average over the population of dwellings (that is, over the distribution of TFs), contribute a concentration of Mambient to indoor air. It does not, however, imply that in any given dwelling, the contribution to the airborne radon concentration from water at the AMCL will equal Mambient.
The available data regarding ambient-air concentrations and transfer factors are limited but adequate for estimation of MTF and Mambient , and the committee determined that the best estimate of the AMCL that can be derived from the available data is the ratio of the estimated arithmetic means:
There are, of course, uncertainties in the estimates of both MTF and Mambient, so it is necessary to assess the magnitude of uncertainty that they induce in the estimation of the AMCL. If the uncertainties are relatively small, the value of AMCLest is unlikely to differ greatly from the true value defined implicitly by equation 9.2. In view of the limited nature of the data available for estimation of MTF and Mambient, the committee chose not to attempt to represent the uncertainties in those estimates with probability distributions. Instead, the committee defined upper and lower bounds that, in its opinion, are highly likely to contain the true values.