Rn-222 measurements have been made in over 10,000 ground-contact rooms
in 908 federal buildings. These data were examined statistically to c
ompare parametric distributions that might be useful in the design and
execution of future surveys of indoor Rn-222 in large buildings. In 1
52 of the 365 buildings with the most measurements per building, the l
ognormal distribution was acceptable. Many other distributions were ob
served in the other 213 buildings, including bi- and tri-modal distrib
utions. In contrast, when data from entire facilities were examined, t
he data were usually described by a lognormal distribution. Because of
the above observations, we propose that any future surveys of indoor
Rn-222 in large buildings be done in two phases, screening and assessm
ent. During the screening phase, all facilities may be surveyed with a
comparatively low sampling density, perhaps, one measurement per 465
m(2) (5,000 ft(2)) of floor space. Based on the results presented here
, it is expected that statistical analysis of the screening data will
reveal any facilities with substantial evidence of a high incidence of
rooms with elevated indoor Rn-222 concentrations (e.g., above 150 Bq
m(-3)). In the assessment phase, the identified facilities sill be sur
veyed with a sampling density of one measurement for every ground-cont
act roam or about one per 84 m(2) (900 ft(2)). Using numerical simulat
ion techniques, we have tested the proposed screening phase protocol a
gainst data from sixteen facilities. The results of the simulated scre
enings support the feasibility of a two-phased approach to the task of
identifying facilities having rooms with elevated Rn-222 concentratio
ns.