A pragmatic approach to estimate the number of days in exceedance of PM10 limit value. (June 2015)

• Record Type: Journal Article
• Authors: Beauchamp, Maxime; Malherbe, Laure; de Fouquet, Chantal
• Is Part Of: Atmospheric environment. Volume 111(2015)
• Page Start: 79
• Page End: 93
• Publication Date: 2015-06
• Subjects: Geostatistics -- Kriging -- Air pollution -- Exceedances of air quality limit values -- Number of exceedances; Air -- Pollution -- Periodicals; Air -- Pollution -- Meteorological aspects -- Periodicals; 551.51
• DOI: 10.1016/j.atmosenv.2015.03.062
• Languages: English
• ISSNs: 1352-2310

Abstract

Abstract: European legislation on ambient air quality requests that Member States report the annual number of exceedances of short-term concentration regulatory thresholds for PM10 and delimit the concerned areas. Measurements at the monitoring stations do not allow to fully describe those areas. We present a methodology to estimate the number of exceedances of the daily limit value over a year, that can be extended to any similar issue. This methodology is applied to PM10 concentrations in France for which the daily limit value is 50 μg m −3, not to be exceeded more that 35 days. A probabilistic model is built using preliminary mapping of daily mean concentrations. First, daily atmospheric concentration fields are estimated at 1 km resolution by external drift kriging, combining surface monitoring observations and outputs from the CHIMERE chemistry transport model. Setting a conventional Gaussian hypothesis for the estimation error, the kriging variance is used to compute the probability of exceeding the daily limit value and to identify three areas: those where we can suppose as certain that the concentrations exceed or not the daily limit value and those where the situation is indeterminate because of the estimation uncertainty. Then, from the set of 365 daily mappings of the probability to exceed the daily limit value, the parameters of a translated Poisson distribution is fitted on the annual number of exceedances of the daily limit value at each grid cell, which enables to compute the probability for this number to exceed 35. The methodology is tested for three years (2007, 2009 and 2011) which present numerous exceedances of the daily limit concentration at some monitoring stations. A cross-validation analysis is carried out to check the efficiency of the methodology. The way to interpret probability maps is discussed. A comparison is made with simpler kriging approaches using indicator kriging of exceedances. Lastly, estimation of the population exposed to PM10 exceedances is discussed. Highlights: We use a kriging model to combine surface observations and the CHIMERE model. Daily probabilities of exceedance are computed with a Gaussian hypothesis. A translated Poissonian model estimates annual probabilities of exceedance. Cross-validation and comparisons with simpler kriging approaches are performed.