Because of the interest in the behavior of radon gas in indoor environments, this study samples houses in the city of Queretaro due to their aforementioned geological history and the ability are provided to evaluate the air quality that inhabitants are most likely exposed to, especially housewives and children who spend most of their time in these environments with the latter being the most vulnerable. The initial idea was to monitor houses with similar construction characteristics, and typical problems encountered were owners’ availability and the complex architecture of houses in the study areas.

Sampling was performed using an SARAD Radon Scout instrument. Its design and memory size allowed for real-time measurements with high accuracy. This instrument can exactly determine variations in radon-222, even at low concentrations. The instrument consists of sensors that measure the humidity, temperature, and barometric pressure using the SARAD GmbH software, version 5.0. The measurement periods varied from 7 to 15 days, and the integration time was one hour (one sample per hour) for all of the measurements.

Using the obtained results, a Spearman’s rank correlation analysis was performed to determine whether any statistically significant linear correlations existed with a confidence level of 95% among the analyzed variables. According to the analysis, no linear correlations existed (See Annex A). However, this does not necessarily imply that the variables are independent because they could be non-linearly related.

The following variables were observed at each site: temperature (˚C), relative humidity (%), atmospheric pressure, and radon concentration. Additionally, a multivariate statistical analysis was performed. A cluster analysis (CA) was applied by forming 3 groups: group 1 is comprised of 12 residences located in the central zone and

corresponds to 35% of the total of sampled houses; group 2 is comprised of 19 residences and corresponds to 56% of the samples; group 3 is comprised of 9% of the samples. In this last group, sites 12, 22, and 28 differ from the other sampled sites because they had radon concentrations above the norms permitted by the U.S. EPA (148 Bq/m³) for indoor environments (Table 2). In Figure 2, the distribution of the sites with respect to their grouping is presented. It is worth noting that the distribution of houses at each site depended upon the availability and help from the proprietor of the house. Although a similar number of samplings were planned for each site, this goal was not possible. The aforementioned groups differ in their variables as shown in Table 3, and one can observe the centroids of each variable in each group. This analysis indicates group 3 is influenced by humidity.

A principal component analysis or factor analysis was performed to determine the associations between the analyzed variables and estimate the possible origins of the radon concentrations in the indoor environments of the 34 sampled houses [7] [8] . The first two components of this principal component analysis that obtained an eigenvalue above 1 were extracted and represent 67.29% of the variability in the observed data (Table 4), which is assumed sufficient for the analysis performed here.

Table 5 presents the weights for each component variable. In it, one can observe that for component 1, rela-

tive humidity and temperature have the highest weights, 0.677 and 0.492, respectively; while the greatest weights for component 2 correspond to radon and temperature with values of 0.693 and 0.609, respectively. Using the obtained radon concentrations, we attempted to establish the spatial distribution of radon with respect to the geological faults. However, there was no evidence that the location of the houses and the faults were directly related, which was verified by the PCA results.

The purpose of the FA was to strengthen the PCA results and obtain a number of factors to explain most of the variability in the three meteorological variables measured (relative humidity, temperature, and atmospheric pressure) across two periods, March-June and September-November, which correspond to the dry-hot and wet periods, respectively. The relative humidity and temperature that exist within the houses determined the flow of radon towards the interior.

In the PCA, two factors were found to represent 67.29% of the variability in the original data (Table 4). The FA corroborates what was observed in the PCA and shows that the relative humidity and temperature have the greatest weights for factor 1 with values of −0.747 and 0.878, respectively, while the greatest weights for factor 2 correspond to radon and the pressure with values of 0.818 and 0.657, respectively (Table 6). These findings allow us to infer that component 1, or factor 1, is influenced by environmental conditions relating to the humidity and temperature because some periods of the year could be cold dry, cold wet, hot dry, or hot wet for example.

In contrast, component 2, or factor 2, favors certain mixing conditions or the presence of certain substances in the air. For these reasons, it is important to consider how wind acts on the indoor radon concentration in two opposing manners. On the one hand, the wind can favor the exchange of air between the interior and exterior of a dwelling depending on its architectural characteristics. On the other hand, the wind induces a local pressure difference between the interior and exterior, which favors the entry of radon into closed sites. These atmospheric variables do not affect the air of houses in the same way and a phase difference occurs over time until the conditions equilibrate.