Periodogram analysis has been used as an interpretative tool of long period
measurements of benzene concentration in urban atmosphere. By removing the
request of evenly sampled data, the periodogram allows to extend spectral
analysis to this kind of data, which usually suffer the "missing data" prob
lem due to instrument vacancies. In this way it is possible to identify the
periodic behaviour of data concentration and then reproduce it by means of
harmonic functions even for very incomplete experimental information allow
ing a good evaluation of some important statistical properties of data.