Geostatistical characterization of nitrogen dioxide concentration in an urban area: Part II: Time component of the estimation error

The temporal component of the estimation error is usually neglected in calculation of the variance of the error associated with the mapping of annual nitrogen dioxide (NO2) concentration. However, under realistic hypotheses, adding a temporal term to the spatial estimation variance is convenient enough. In a survey using passive diffusion samplers, which was carried out in 2001 on the Mulhouse area, the two measurement phases do not allow calculation of variance of the temporal error. But this error is accessible by means of the few fixed analyzers that continuously measure NO2 concentrations. The statistics of the experimental temporal error are examined and the temporal variogram is calculated. Experimental quadratic error is then compared to the theoretical estimation variance, deduced from the temporal variogram. The fixed analyzers available in Alsace are then used to “optimize” empirically the duration and the date of the beginning of the temporary measurement phases. Without increasing the cost of the temporary survey, it is then possible to improve the precision of the maps of annual NO2 concentration, by combining the spatial optimization presented in Part I of the paper, and the temporal optimization presented here.