ETC/ACM logoETC/ACM



Services

AirBase - the European Air quality dataBase

AirBase

 

Data retrieval Background info Country support for reporting

Data aggregation, calculation of statistics and NOx values in AirBase

AirBase provides statistical data on air pollutant parameters additional to the reported national measurement data.
It follows the 'criteria for the aggregation of data and the calculation of statistical parameters' of Annex IV in Commision Decision 2001/752/EC.

1. Hourly and daily values

Aggregation of data

The air quality statistics in AirBase are based on hourly values, daily (24-hour) average values and daily 8-hour maximum values. However, most of the reported measurement data are in hourly time episodes. To obtain the daily and 8-hour based statistical parameters the hourly values (if available) are aggregated to derive daily and 8-hourly values. If a country reports both hourly and daily values, the reported daily values will be ignored. The calculated daily values will be used instead for calculating the statistics. If 3-hourly data are delivered, these data are aggregated into daily values.

For the aggregation of hourly data to longer averaging periods (8 hourly, daily) a minimum data capture of 75% is required to calculate a valid aggregated value:

For the aggregation of 3hourly data to daily values we have the following rule:

Statistics calculation on annual basis

The following types of annual statistics are calculated depending on the component:

The annual statistical parameters of the table are routinely calculated and stored in AirBase. The statistical parameters are calculated irrespective of the proportion of valid data (data capture) with one exception: all hourly and daily statistics, which are based on one day or less, are excluded. So statistics with a data coverage lower than 0.275% are not calculated. For each statistic the data coverage(*) percentage is calculated. This is done as follows:

Data Coverage = (Nvalid / Nyear) · 100%

where Nvalid is the number of valid hourly/daily values and Nyear is the number of hours/days in the year.

(*) Definition of Coverage:
In the Air Quality Daughter Directives the terms Data Capture and Time Coverage have been defined.
Time Coverage is the percentage of measurement time in a given period.
Data Capture is the percentage of valid measurement values in a given data set.
For each yearly time series the so called Data Coverage has been calculated in AirBase and is defined as:
Data Coverage = Data Capture · Time Coverage
The data capture and time coverage, and hence the data coverage, include losses of data due to the regular calibration or the normal maintenance of the instrumentation. In the Directives these losses are excluded.

The following table gives an overview of the annual statistical parameters available in AirBase. The annual statistical parameters are calculated on basis of the hourly values, the daily average values and the daily 8-hour maximum values. Be aware that the same statistical parameter, but based on a different (averaging) time period, will lead in most cases to a different outcome. This has its cause in the difference in the definitions on how to derive the hourly, the daily average and the daily 8-hour maximum values.

For example:

When there is a peak in the 1-hour values on a day, this peak value will be appointed as the maximum 1-hour value of that day. The annual maximum is the maximum of all those daily maximum values of that year.

In deriving the daily average value, such peak will be leveled out by the other hourly values of that day. If the criterion "a daily averaged concentration is calculated when at least 18 valid hourly values are available" is not fulfilled, the daily average value will even not be calculated for that day. In that case the peak value will not be recorded and stay completely 'invisible' as maximum daily average for such a day.

This leads in most cases to large differences between an annual maximum based on hourly values (table column 1-hour values) and an annual maximum based on daily average values (table column daily average values).

top of page

Table: calculated annual statistics in AirBase

Component Annual statistical parameter based on
1-hour values daily average values daily 8-hour maximum values
Sulphur dioxide
(SO2)
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• 99.9 percentile
• maximum
• hours with conc. > 350 µg/m3
• 25th highest value
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• days with conc. > 125 µg/m3
• 4th highest value
-
Nitrogen dioxide
(NO2)
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• hours with conc. > 200 µg/m3
• 19th highest value
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
-
Nitrogen monoxide
(NO)
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
-
Nitrogen oxides
(NOx)
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
-
Ozone
(O3)
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• AOT40
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• days with conc. > 120 µg/m3
• 26th highest value
• SOMO35
Carbon monoxide
(CO)
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
Particulate matter
(PM10)
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• days with conc. > 50 µg/m3
• 8th highest value
• 36th highest value
-
Other • annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
• annual mean
• 50 percentile
• 95 percentile
• 98 percentile
• maximum
-
top of page

Calculation of specific aggregations and statistics

A. For all components

  • Annual mean
    The annual mean is calculated as follows:

    Annual mean = Σi Ci / Nvalid

    where Ci is the valid hourly/daily/day8hmax concentration and the summation is over all valid hourly/daily values measured in the year. Nvalid is total number of valid hourly/daily values in the year.

  • Percentiles
    The yth percentile should be selected from the measurement values (valid hourly/daily/day8hmax concentrations). All the values should be listed in increasing order:

    X1 ≤ X2 ≤ X3 ≤ … ≤ Xk ≤ … ≤ XN-1 ≤ XN

    The yth percentile is the concentration Xk, where the value of k is calculated as follows:

    k = (q · N)

    with q = y/100 and N the number of values. The value of (q · N) should be rounded off to the nearest whole number (values < 0.499999… are rounded to 0, values = 0.5 are rounded to 1).

  • Maximum
    The (annual) maximum is calculated as follows:

    Maximum = max(Ci)

    where Ci are the valid hourly/daily/day8hmax concentrations and i is running over all valid hourly/daily/day8hmax values measured in the year.

B. Only for SO2, NO2, PM10 and O3

  • kth highest value
    The kth highest value should be selected from the measurement values. All the values should be listed in decreasing order:

    X1 ≥ X2 ≥ X3 ≥ … ≥ Xk ≥ … ≥ XN-1 ≥ XN

    The kth highest value is the concentration Xk.

    Example: the limit value for the protection of human health for PM10 is that the daily average of 50 µg/m3 will not be exceeded on more than 35 days per year. If the 36th highest value is more than 50 µg/m3, the limit value for PM10 has been exceeded.

  • Number of hours/days with concentration > y µg/m3
    The n number of hours/days with concentration > y µg/m3 (with y = limit or threshold value) can be calculated from the measurement values:

    X1 , X2 , X3 , … , Xk , … , XN-1 , XN

    N is the number of Xk-values for which Xk > y µg/m3. If n > 35 in the example on PM10 at the previous bullet, the limit value for PM10 has been exceeded.

C. Only for O3 and CO

  • 8-hour running mean
    The 8-hour running averaged value for each hour is calculated as the average of the values for that hour and the 7 foregoing hours (averaging period). So, the averaging period of hour1 of dayn is hour17 of dayn-1 until hour1 of dayn. The averaging period of hour24 of dayn is hour16 of dayn until hour24 of dayn.

  • Maximum daily 8-hour mean
    The maximum daily 8-hour for a day is the maximum of all the 8-hours running averages for that day.

D. Only for O3

  • AOT40 (crops)
    (Accumulated dose of ozone Over a Threshold of 40 ppb)
    AOT40 means the sum of the differences between hourly concentrations greater than 80 µg/m3 (= 40 parts per billion) and 80 µg/m3:

    AOT40measured = Σi max(0, (Ci - 80))

    where Ci is the hourly mean ozone concentration in µg/m3 and the summation is over all hourly values measured between 8.00 – 20.00 Central European Time(**) each day and for days in the 3 month growing season crops from 1 May to 31 July.

    (**) In AirBase the time zone was disregarded. So the values between 8.00 - 12.00 of the reported time have been taken.

    AOT40 has a dimension of (µg/m3)·hours. AOT40 is sensitive to missing values and a correction to full time coverage has been applied:

    AOT40estimate = (AOT40measured · Nperiod) / Nvalid

    where Nvalid is the number of valid hourly values and Nperiod is the number of hours in the period.

  • SOMO35
    (Sum of Ozone Means Over 35 ppb)
    For quantification of the health impacts the World Health Organisation recommends the use of the SOMO35 indicator. SOMO35 stands for the sum of the differences between maximum daily 8-hour running mean concentrations greater than 70 µg/m3 (= 35 parts per billion) and 70 µg/m3:

    SOMO35measured = Σi max(0, (Ci - 70))

    where Ci is the maximum daily 8-hour running mean ozone concentration in µg/m3 and the summation is over all days per calendar year.

    SOMO35 has a dimension of (µg/m3)·days. SOMO35 is sensitive to missing values and a correction to full time coverage has been applied:

    SOMO35estimate = (SOMO35measured · Nperiod) / Nvalid

    where Nvalid is the number of valid daily values and Nperiod is the number of days per year.

top of page

2. Other than hourly and daily values: n-day (n>1), n-week, n-month, year and var

Non automatic measured components (e.g., the components from the 4th Daughter Directive (Heavy Metals and PAHs) have also other averaging times than hour and day: week, 2-week, 4-week, month, 3-month, year, etc.).
These measurements consist of samples with a start date/time and an end date/time.

The averaging time is the period of the sample (end date/time minus start date/time). If the sample periods of a component differ 25% or more from a constant averaging time, the averaging time has been defined as "var".
For Example: if all periods of 4-week samples are within 21 and 35 days, the averaging time is not "var" but still 4-week.

The 100% period for a n-month sample has been defined as the period starting from the start date/time of the sample and ending on the same day number and time n months later.
Some Examples: (1) the sample starts at 5 March at 00:00, the 100% 1-month period is until 5 April at 00:00; (2) the sample starts at 30 January at 00:00, the 100% 1-month period would be until “virtual” 30 February, which is actually 2 March at 00:00 (no leap year).

The only statistics calculated for these averaging times are:

  • annual mean
  • 50 percentile
  • 95 percentile
  • 98 percentile
  • maximum

Note: n-hour values are aggregated into daily values. The statistics are based on these daily values.

All statistics calculations are done in analogy to the hourly/daily statistics calculations. The only exception is the data coverage calculation and the annual mean calculation. The data coverage is calculated as follows:

Data Coverage = ΣiNvalid,i / Nyear · 100%

where Nvalid,i is the number of hours in the valid sample period i and Nyear is the number of hours in the year.

The annual means are calculated according to the formula:

Annual mean = ΣiNiCi / ΣiNi

where Ci is the concentration in the valid sample period i and Ni is the number of hours in sample period i. The summation is over all valid periods in the year.

Remark: if a period is partially outside the year, only hourly values are taken into account between 1 January and 31 December of the year.

3. Calculation of NOx values

To obtain a better coverage of NOx measurements in AirBase, there are in AirBase NOx values available which have been derived from reported NO and NO2 measurements following the formula:

CNOx = CNO2 + ((MNO2 / MNO) * CNO)

where

CNOx = NOx concentration in µg NO2/m3
CNO2 = NO2 concentration in µg/m3
CNO = NO concentration in µg/m3
MNO = Molecular mass of NO = 30
MNO2 = Molecular mass of NO2 = 46

For defining the measurement configuration of the derived NOx measurements, the information is used of the measurement configuration of NO. In case NO, NO2 and NOx are all reported, the reported NOx values will have priority over the derived NOx values.

top of page

Valid XHTML 1.0 Transitional