CEN/TS 18275:2026

• Abstract:

  This document defines and describes the methodology to calculate Modelling Quality Indicators (MQI) and determine fulfilment of the Modelling Quality Objectives (MQO). MQO are provided to help ensure that modelling-based assessments of air quality in the context of the ambient air quality directive [1] are objective and comparable, and of sufficient quality to obtain reliable information about concentrations of air pollutants in ambient air. The method uses measurement uncertainty and a level of stringency as a benchmark for the acceptable level of difference between modelled and measured values.; This document concerns the performance of an entire modelling system therefore the term “modelling quality objectives” is used rather than “model quality objectives”. This document concerns the use of modelling results for assessment as specified in [2]. Such modelling systems aim to capture both the spatial and temporal variability of the environmental indicator under assessment in the modelling domain. This document establishes a method to determine if the results of a modelling system fulfil the MQO and therefore reach an adequate data quality level within the modelling domain defined for assessment.; The procedures described in this document are limited in scope as they concern only statistical performance indicators. A full evaluation of a modelling system considers additional elements of quality assurance, but such procedures are outside the scope of this document.; This document only addresses modelling applications where measurements of pollutant concentrations are available that meet the data requirements for the validation defined in this document. This document specifies MQO that are applicable to all concentration ranges that may occur in ambient air. In the context of this document, MQI and MQO are specified for:; - daily and annual averaged concentrations of particulate matter with aerodynamic diameter less or equal to 2,5 µm (PM2.5);; - daily and annual averaged concentrations of particulate matter with aerodynamic diameter less or equal to 10 µm (PM10);; - hourly, daily and annual averaged concentrations of nitrogen dioxide (NO2);; - maximum daily 8-hour mean and seasonal averaged concentrations of ozone (O3);; - hourly, daily and annual averaged concentrations of sulphur dioxide (SO2);; - maximum daily 8-hour mean and daily averaged concentrations for carbon monoxide (CO);; - annual averaged concentrations for benzene (C6H6);; - annual averaged concentrations for lead (Pb);; - annual averaged concentrations for arsenic (As);; - annual averaged concentrations for cadmium (Cd);; - annual averaged concentrations for nickel (Ni);; - annual averaged concentrations for benzo[a]pyrene (BaP).; This document addresses competent authorities, research institutions, consultants, or other bodies responsible for the performance of air quality modelling when applied for assessment purposes.; NOTE Fulfilment of MQO is either normative or informative, depending on the quality of information used to determine the uncertainty parameters and stringency factors set out in this document.

• Status: Published
• Publication Date: 07-Jul-2026
• ICS: 13.040.20 - Ambient atmospheres
• Technical Committee: CEN/TC 264 - Air quality
• Drafting Committee: CEN/TC 264/WG 43 - Model quality objectives
• Current Stage: 6060 - Definitive text made available (DAV) - Publishing
• Start Date: 08-Jul-2026
• Due Date: 15-Apr-2026
• Completion Date: 08-Jul-2026
• Directive: Not Harmonized2008/50/EC - Directive 2008/50/EC of the European Parliament and of the Council of 21 May 2008 on ambient air quality and cleaner air for Europe
• Mandate: M/612 - Ambient Air - Modelling quality objectives
• Ref Project: SIST-TS CEN/TS 18275:2026 - Ambient air - Definition and use of modelling quality objectives for air quality assessment

SLOVENSKI STANDARD
01-oktober-2026
Zunanji zrak - Določitev in uporaba ciljev kakovosti modeliranja za oceno
kakovosti zraka
Ambient air - Definition and use of modelling quality objectives for air quality assessment
Außenluft - Definition und Verwendung von Modellierungsqualitätszielen für die
Beurteilung der Luftqualität
Air ambiant - Définition et utilisation des objectifs de la qualité des modélisations pour
l'évaluation de la qualité de l'air
Ta slovenski standard je istoveten z: CEN/TS 18275:2026
ICS:
13.040.20 Kakovost okoljskega zraka Ambient atmospheres
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

CEN/TS 18275
TECHNICAL SPECIFICATION
SPÉCIFICATION TECHNIQUE
July 2026
TECHNISCHE SPEZIFIKATION
ICS 13.040.20
English Version
Ambient air - Definition and use of modelling quality
objectives for air quality assessment
Air ambiant - Définition et utilisation des objectifs de la Außenluft - Definition und Verwendung von
qualité des modélisations pour l'évaluation de la Modellierungsqualitätszielen für die Beurteilung der
qualité de l'air Luftqualität
This Technical Specification (CEN/TS) was approved by CEN on 24 May 2026 for provisional application.

The period of validity of this CEN/TS is limited initially to three years. After two years the members of CEN will be requested to
submit their comments, particularly on the question whether the CEN/TS can be converted into a European Standard.

CEN members are required to announce the existence of this CEN/TS in the same way as for an EN and to make the CEN/TS
available promptly at national level in an appropriate form. It is permissible to keep conflicting national standards in force (in
parallel to the CEN/TS) until the final decision about the possible conversion of the CEN/TS into an EN is reached.

CEN members are the national standards bodies of Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia,
Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway,
Poland, Portugal, Republic of North Macedonia, Romania, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Türkiye and
United Kingdom.
EUROPEAN COMMITTEE FOR STANDARDIZATION
COMITÉ EUROPÉEN DE NORMALISATION

EUROPÄISCHES KOMITEE FÜR NORMUNG

CEN-CENELEC Management Centre: Rue de la Science 23, B-1040 Brussels
© 2026 CEN All rights of exploitation in any form and by any means reserved Ref. No. CEN/TS 18275:2026 E
worldwide for CEN national Members.

Contents Page
European foreword . 4
1 Scope . 5
2 Normative references . 6
3 Terms and definitions . 6
4 Symbols and abbreviations . 8
5 Modelling Quality Indicator (MQI) and Modelling Quality Objective (MQO) . 9
5.1 General . 9
5.2 MQI for short-term indicators .11
5.3 MQI for long-term indicators .12
5.4 The 90 % principle - additional condition for available sampling points .12
5.5 Modelling Quality Objective (MQO) .13
6 Measurement uncertainty .13
7 Data requirements in the modelling system validation.15
7.1 General .15
7.2 Measurement data quality .15
7.3 Minimum number of sampling points .15
7.4 Independence of the measurement data from the modelling system results .15
Annex A (normative) Measurement uncertainty parameters .16
A.1 General .16
A.2 Short-term averages.16
A.3 Long-term averages .17
Annex B (normative) Stringency of the MQO: the β parameter .21
Annex C (informative) Effect of the number of available sampling points (N ) on calculating
s
the MQO .22
Annex D (informative) Options to cope with few sampling points .25
Annex E (informative) Statistical performance indicators .26
Annex F (informative) Generalization to other pollutants.28
Annex G (informative) The MQO in practice .30
G.1 General .30
G.2 Long-term MQO .30
G.3 Short-term MQO .34
Annex H (informative) Complementary performance indicators .39
H.1 General .39
H.2 Temporal performance indicators .40
H.3 Spatial performance indicators (SI) .41
Annex I (informative) Visualization . 42
I.1 Long-term MQI/MQO . 42
I.2 Short-term MQI/MQO . 43
Bibliography . 45
European foreword
This document (CEN/TS 18275:2026) has been prepared by Technical Committee CEN/TC 264 “Air
quality”, the secretariat of which is held by DIN.
Attention is drawn to the possibility that some of the elements of this document may be the subject of
patent rights. CEN shall not be held responsible for identifying any or all such patent rights.
This document has been prepared under a standardization request addressed to CEN by the European
Commission. The Standing Committee of the EFTA States subsequently approves these requests for its
Member States.
Any feedback and questions on this document should be directed to the users’ national standards body.
A complete listing of these bodies can be found on the CEN website.
According to the CEN/CENELEC Internal Regulations, the national standards organisations of the
following countries are bound to announce this Technical Specification: Austria, Belgium, Bulgaria,
Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland,
Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Republic of
North Macedonia, Romania, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Türkiye and the
United Kingdom.
1 Scope
This document defines and describes the methodology to calculate Modelling Quality Indicators (MQI)
and determine fulfilment of the Modelling Quality Objectives (MQO). MQO are provided to help ensure
that modelling-based assessments of air quality in the context of the ambient air quality directive [1] are
objective and comparable, and of sufficient quality to obtain reliable information about concentrations of
air pollutants in ambient air. The method uses measurement uncertainty and a level of stringency as a
benchmark for the acceptable level of difference between modelled and measured values.
This document concerns the performance of an entire modelling system therefore the term “modelling
quality objectives” is used rather than “model quality objectives”. This document concerns the use of
modelling results for assessment as specified in [2]. Such modelling systems aim to capture both the
spatial and temporal variability of the environmental indicator under assessment in the modelling
domain. This document establishes a method to determine if the results of a modelling system fulfil the
MQO and therefore reach an adequate data quality level within the modelling domain defined for
assessment.
The procedures described in this document are limited in scope as they concern only statistical
performance indicators. A full evaluation of a modelling system considers additional elements of quality
assurance, but such procedures are outside the scope of this document.
This document only addresses modelling applications where measurements of pollutant concentrations
are available that meet the data requirements for the validation defined in this document. This document
specifies MQO that are applicable to all concentration ranges that may occur in ambient air. In the context
of this document, MQI and MQO are specified for:
— daily and annual averaged concentrations of particulate matter with aerodynamic diameter less or
equal to 2,5 µm (PM );
2.5
— daily and annual averaged concentrations of particulate matter with aerodynamic diameter less or
equal to 10 µm (PM );
— hourly, daily and annual averaged concentrations of nitrogen dioxide (NO );
— maximum daily 8-hour mean and seasonal averaged concentrations of ozone (O );
— hourly, daily and annual averaged concentrations of sulphur dioxide (SO );
— maximum daily 8-hour mean and daily averaged concentrations for carbon monoxide (CO);
— annual averaged concentrations for benzene (C H );
6 6
— annual averaged concentrations for lead (Pb);
— annual averaged concentrations for arsenic (As);
— annual averaged concentrations for cadmium (Cd);
— annual averaged concentrations for nickel (Ni);
— annual averaged concentrations for benzo[a]pyrene (BaP).
This document addresses competent authorities, research institutions, consultants, or other bodies
responsible for the performance of air quality modelling when applied for assessment purposes.
NOTE Fulfilment of MQO is either normative or informative, depending on the quality of information used to
determine the uncertainty parameters and stringency factors set out in this document.
2 Normative references
There are no normative references in this document.
3 Terms and definitions
For the purposes of this document, the following terms and definitions apply.
ISO and IEC maintain terminology databases for use in standardization at the following addresses:
— ISO Online browsing platform: available at https://www.iso.org/obp/
— IEC Electropedia: available at https://www.electropedia.org/
3.1
assessment
application of a modelling system (3.10) to calculate levels of ambient air quality retrospectively over a
calendar year
Note 1 to entry: When modelling-based assessments of air quality are carried out in the context of [1], then
assessment is defined as in [1], except that [1] has a wider scope and includes any method used to measure,
calculate, predict or estimate levels. Within this document, the scope is limited to modelling.
3.2
competent authority
authority responsible for the air quality assessment
Note 1 to entry: When modelling-based assessments of air quality are carried out in the context of [1], then the
competent authority is the national authority who has been designated responsible for air quality assessment in the
Member State.
3.3
data coverage
proportion of the calendar year for which valid measurement data are available, expressed as percentage
Note 1 to entry: When modelling-based assessments of air quality are carried out in the context of [1], then the
requirements for data coverage are those found in [1].
3.4
fixed measurements
measurements taken at sampling points (3.15), either continuously or by random sampling, at constant
locations for at least 1 calendar year to determine the levels of ambient air quality
Note 1 to entry: When modelling-based assessments of air quality are carried out in the context of [1], then the
requirements in terms of data quality objectives for fixed measurements are those found in [1].
3.5
indicative measurements
measurements, taken either at regular intervals during a calendar year or by random sampling, to
determine the levels of ambient air quality in accordance with quality criteria that are less strict than
those required for fixed measurements (3.4)
Note 1 to entry: When modelling-based assessments of air quality are carried out in the context of [1], then the
requirements in terms of data quality objectives for indicative measurements are those found in [1].
3.6
modelling application
modelling results obtained with a modelling system (3.10) over a given modelling domain (3.7), over a
given time period
3.7
modelling domain
geographic area covered by a modelling system (3.10) for assessment (3.1)
3.8
modelling quality indicator
MQI
statistical indicator for modelling validation (3.11) calculated on the basis of measurement and modelling
results
Note 1 to entry: The MQI describes the difference between measurements and modelling results, normalized by
measurement uncertainty and a stringency factor.
3.9
modelling quality objective
MQO
criterion for modelling validation (3.11) that values of the MQI should fulfil
3.10
modelling system
chain of models and sub-models, including all necessary input data, and any post-processing
3.11
modelling validation
comparison of modelling results against measured data, using MQI indicators
Note 1 to entry: When validating air quality modelling applications (3.6), this typically implies a wider range of
statistical indicators. However, in this document, validation of a modelling application (3.6) is limited to the use of
MQI indicators and fulfilment of the MQO.
3.12
maximum measurement uncertainty
maximum accepted scatter of the measured values around the true value that is attributed to the
measurement of a pollutant
Note 1 to entry: In this document, we use expanded uncertainties expressed at a 95 % confidence level. This
corresponds to the maximum measurement uncertainty (3.12) as specified in [1].
3.13
maximum modelling uncertainty
maximum accepted scatter of the modelled values around the true value of a pollutant
Note 1 to entry: The maximum modelling uncertainty is calculated as the maximum measurement uncertainty
(3.12) multiplied by a stringency factor.
3.14
reference value
RV
constant value, used to define the relationship between measurement uncertainty and concentration
level, for each pollutant and time average
Note 1 to entry: RV is expressed in the same units as the concentration level of the pollutant.
3.15
sampling point
location where sampling of a given pollutant in ambient air takes place
Note 1 to entry: Each sampling point is location-pollutant-methodology specific.
4 Symbols and abbreviations
AQUILA European network of national Air QUalIty reference LAboratories
As Arsenic (in PM )
AR Acceptability Range
Fraction of the maximum measurement uncertainty that is not proportional to the
α , α
short long
concentration, ranging between 0 and 1, for short- and long-term averages. It is
pollutant and measurement-type specific.
Ratio between the maximum modelling and measurement uncertainties for short-
β , β
short long
and long-term averages. It is pollutant and measurement-type specific, and it
determines the stringency of the MQO, as . The stringency factor
1+β β
corresponds to the maximum ratio of uncertainty of modelling applications as
specified in [1].
BaP Benzo[a]pyrene (in PM )
BIAS Difference between modelling results and measurements
Cd Cadmium (in PM )
CO Carbon monoxide
CRMSE Centred Root Mean Square Error
C H Benzene
6 6
D Non-integer value (distance) used to calculate the value of the 90th percentile MQI
FAIRMODE Forum for AIR quality MODelling in Europe
GDE Guide to the Demonstration of Equivalence
M(t) Value of modelling result at time t
Time-average of modelling results M(t)

M
MQI Modelling Quality Indicator
MQO Modelling Quality Objective
Ni Nickel (in PM )
NO Nitrogen dioxide
N Number of time steps (t)
t
N Number of available sampling points (s)
s
O Ozone
Pb Lead (in PM )
PM Particulate matter which passes through a size-selective inlet as defined in the
2.5
reference method for the sampling and measurement of PM , EN 12341 [3], with a
2.5
50 % efficiency cut-off at 2,5 μm aerodynamic diameter
PM Particulate matter which passes through a size-selective inlet as defined in the
reference method for the sampling and measurement of PM , EN 12341 [3], with a
50 % efficiency cut-off at 10 μm aerodynamic diameter
O(t) Value of measurement (O) result, i.e. Observation (fixed or indicative) at time t
Ō Time-average of measurement values O(t)
R Correlation coefficient
RMSE Root Mean Square Error
RMSU Root Mean Square Uncertainty of measurement (O)
O
RSS Residual Sum of Squares
RV , RV Reference Values for short- and long-term averages
short long
S Sampling point
SI Spatial performance Indicator
SO Sulphur dioxide
sp Spatial average over sampling points
Standard deviation of measurement (O) and modelling (M) values
σσ,
OM
TI Temporal performance Indicator
U (O(t)) and Maximum uncertainty of measurement (O) and modelling (M) at measured
O
U (O(t)) concentration levels at time t
M
U Maximum relative uncertainty of measurements
Or
5 Modelling Quality Indicator (MQI) and Modelling Quality Objective (MQO)
5.1 General
The modelling quality indicator (MQI) calculation is based on a comparison of measurements O t and
( )
modelling values Mt for a time series (or an average of measurement values O and modelling values
( )
M over a time period), divided by the maximum measurement uncertainty (or an averaged
U O t
( )
( )
O
UO ) scaled by an associated level of stringency.
( )
O
The main elements of the MQI, the maximum measurement uncertainty and the associated stringency
level are defined and discussed in subclauses 5.2, 5.3 and 6. The minimum level of modelling quality that
shall be achieved for assessment, i.e. the modelling quality objective (MQO) calculated with the MQI, is
described in subclauses 5.4 and 5.5. Clause 7 reviews the requirements regarding the reliability,
availability and independence of sampling points input data. Additional complementary performance
validation indicators are defined in Annex H (informative), these support the visualizations of the
MQI/MQO which are provided in Annex I (informative).
The MQI and associated MQO are proposed to provide stakeholders with a consistent and unbiased
estimate of the quality of results of modelling systems when applied for assessment. The MQI and MQO
are defined for both short (e.g. hourly, daily: MQI ) and longer (e.g. annual, seasonal: MQI ) time
short
long
averages of pollutant concentrations corresponding to the time period required for assessment of a
particular environmental indicator under [1]. In this document, “short” refers to hourly time averages for
NO and SO , 8 h daily maximum time averages for O , 8-hour averages for CO and daily time averages for
2 2 3
PM and PM whereas “long” refers to annual time averages per calendar year for NO , PM , PM , SO ,
10 2.5 2 2.5 10 2
benzene, lead, arsenic, cadmium, nickel and BaP and seasonal time averages for O . For assessment,
additional short-term time averages are requested in [1] for NO and SO ; the quality of modelling results
2 2
for these pollutants can be estimated using MQI for these short-term parameters (see Table 1).
NOTE The MQI and MQO corresponding to a particular environmental indicator are either normative or
informative depending on the quality of the information used to determine the uncertainty parameters and
stringency factors set out in this document.
Table 1 — Overview of status of MQI/MQO by pollutants and time averages and associated
annexes for the uncertainty parameters and stringency level
Time period Status of the
Time Status of the
covered by MQI/MQO for
Time resolution of MQI/MQO for fixed
Pollutant modelling indicative
category the data used measurements
validation measurements
in the MQI (Annexes)
data (Annexes)
Long Annual Year Normative (A, B) Normative (A, B)
PM
Short Daily Year Normative (A, B) Normative (A, B)
Long Annual Year Normative (A, B) Normative (A, B)
PM
2.5
Short Daily Year Normative (A, B) Normative (A, B)
Long Annual Year Normative (A, B) Normative (A, B)
NO Short Daily Year Informative (F) Informative (F)
Short Hourly Year Normative (A, B) Normative (A, B)
Long Hourly Season Normative (A, B) Normative (A, B)
O
Short Hourly Season Normative (A, B) Normative (A, B)
Long Annual Year Informative (F) Informative (F)
SO Short Daily Year Informative (F) Informative (F)
Short Hourly Year Informative (F) Informative (F)
Short Daily Year Informative (F) Informative (F)
CO
Short 8h daily max Year Informative (F) Informative (F)
Benzene Long Annual Year Informative (F) Informative (F)
Lead Long Annual Year Informative (F)
Arsenic Long Annual Year Informative (F)
Cadmium Long Annual Year Informative (F)
Nickel Long Annual Year Informative (F)
BaP Long Annual Year Informative (F)
5.2 MQI for short-term indicators
For one point in time (t) at one sampling point, the MQI is defined as the ratio between the absolute
short
difference between modelling results and measurements (absolute bias) and the square root of the sum
of the squares of the maximum measurement uncertainties U O t and maximum modelling
( )
( )
( )
O
uncertainties U O t , estimated at the measured concentration level ( O t ):
( ) ( )
( ))
(
M
O t − M t
( ) ( )
MQI t = (1)
( )
short
U O t + U O t
( ) ( )
( ) ( )
OM
where O t and Mt are the measured and modelled values at time t , respectively. The maximum
( ) ( )
modelling uncertainty is assumed to be proportional to the maximum measurement uncertainty:
U O t =β U O t (2)
( ( )) ( ( ))
M short O
β is a proportionality coefficient that is pollutant and measurement specific (see Annex B and
short
Annex F). The value of β determines the maximum modelling uncertainty and therefore the
short
stringency of the validation process. The values of β are provided in Annex B (normative) for PM ,
short
PM , NO and O and in Annex F (informative) for other pollutants. Formula (1) can then be simplified
2.5 2 3
as follows:
O t −−M t O t M t
( ) ( ) ( ) ( )
MQI t (3)
( )
short
2 2 2 2
U O t ++ββU O t 1 U O t
( ) ( ) ( )
( ) ( ) ( )
O short O short O
For a time series at one sampling point, we generalize Formula (3) to:
N
t

O t − M t
( ( ) ( ))
∑ 
t=1
N
t RMSE
MQI  (4)
short
N 2
t
22 1+β RMSU
U O t + U O t short O
( ( )) ( ( ))
OM
∑ 
t=1
N
t
where N is the number of data points in the time series. With this MQI formulation, the Root Mean
t
Square Error ( RMSE ) between measured and modelled values (numerator) is compared to the product
of the Root Mean Square of the maximum measurement uncertainty ( RMSU ) and a stringency factor
O
1+β , a value representative of the maximum modelling uncertainty (denominator). The maximum
short
measurement uncertainty parameters for the statistical indicators used in this section and following are
described in Annex A (normative) and Annex F (informative).
==
==
5.3 MQI for long-term indicators
To evaluate air quality modelling of long-term (e.g. annual or seasonal) averaged pollutant
concentrations, the MQI is modified so that the mean absolute bias between the long-term average
short
measurement and modelling values is normalized by the square root of the sum of squares of the
maximum uncertainties of the averaged measurement and modelling values:
O− M
BIAS BIAS
MQI (5)
long
2 2 2 2 2 2
UO + U O UO ++ββUO 1 UO
( ) ( ) ( ) ( ) ( )
O M O long O long O
where O is the measurement averaged value, M is the modelled averaged value and β is the
long
proportionality coefficient that is pollutant and measurement specific (see Annex B (normative) and
Annex F (informative) for details). The value of β determines the maximum modelling uncertainty
long
and therefore the stringency of the validation process. The maximum modelling uncertainty for long-
term UO is defined proportionally to the maximum measurement uncertainty for long-term averages
( )
M
UO .
( )
O
U O =β UO (6)
( ) ( )
M long O
NOTE The formulations of the maximum measurement uncertainties UO and UO are provided in
( )
( )
O O
Clause 6.
5.4 The 90 % principle - additional condition for available sampling points
Related to the requirement of [1], the MQO should be verified at least at 90 % of the available sampling
points. The MQI associated with each sampling point is therefore calculated, and ranked in ascending
order, from which the 90th percentile value is inferred by linear interpolation:
MQI MQI N+ MQI N+−1 MQI N D (7)
( ) ( ( ) ( ))
90th 90th 90th 90th
N
where is the calculated 90th percentile of the number of sampling points defined as the integer
90th
part of a given real number (e.g. floor(3,9) = 3):
N = floor 0, 9N (8)
( )
90th s
with N the total number of available sampling points.
s
D is the non-integer distance defined as
D 09, N− N (9)
s 90th
=
=
===
5.5 Modelling Quality Objective (MQO)
The MQO sets the minimum level of modelling quality that shall be achieved for assessment. This requires
that the MQI and MQI are less or equal to unity, i.e.:
short,90th long,90th
For the long term MQO :, MQI ≤ 1 00 (10)
long,90th
For the short term MQO :,MQI ≤ 1 00 (11)
short, 90 th
where the 90th subscript indicates that the MQI or MQI shall be less than 1,00 for at least 90 %
short long
of the available sampling points (see subclause 5.4 for a practical derivation). The values of MQI and
long
MQI shall be calculated with at least two decimals. When the number of sampling points (i.e. the
short
number of MQI) is less than the minimum number specified in Clause 7.3, the approach described in
subclauses 5.4 and 5.5 does not apply (see Clause 7.3).
For the purpose of assessment of long-term air quality standards as per [1], the MQO shall be fulfilled,
long
regardless of the temporal resolution of the modelling values (hourly, daily or annual). For assessment
of short-term air quality standards as per [1] based on average hourly or daily values, modelling system
results shall fulfil both the MQO and the MQO to ensure that results are consistent time wise.
long short
The calculation of both the long- and short-term MQO is illustrated in Annex G with an example.
6 Measurement uncertainty
In practice, measurement values include additional sources of uncertainty on top of the instrument
uncertainty, such as factors related to spatial representativeness, the influence of the microscale location
of the sampling point or the inlet situation (e.g. height, distance from source, etc.) that lead to differences
with modelling results. These additional uncertainties are accounted for by the proportionality
coefficient (or stringency factor) β that determines the stringency of the MQO (values of β are provided
in Annex B (normative) and Annex F (informative)).
A simplified and general expression for the maximum measurement uncertainty U O t is derived
O
( ( ))
here. O t are the measured values at each point in time t , where t is a number between 1 and N .
( )
t
A general expression for the maximum measurement uncertainty is derived by considering that the
uncertainty U O t of a measurement at concentration level O t , can be decomposed into a
( ( )) ( )
O
component that is proportional to the concentration level, and a non-proportional
U O t
( )
( )
Op
component, U O t [4, 5]:
( ( ))
Onp
22 2
(12)
U O t U O t+ U O t
( ) ( ) ( )
( ) ( ) ( )
O Op Onp
The non-proportional component U O t is by definition independent of the concentration and is
( ( ))
Onp
defined as a fraction α (ranging between 0 and 1) of the uncertainty at the reference value:
short
2 2 2
U O t = α U RV (13)
( ( )) ( )
Onp short O short
=
Assuming that U O t behaves sufficiently linearly with concentration level O t , Formula (12) can
( ( )) ( )
O
be approximated by the equation of a straight line, y mx+ c , with y= U O t , x= O t and the
( )
( ( ))
O
intercept c being the non-proportional component U O t .
( )
( )
Onp
The proportional component U O t can then be derived from Formula (12) by solving for the
( ( ))
Op
gradient term m where O t = RV and defining the maximum relative uncertainty of measurements
( )
short
around the reference value as U RV = U RV / RV This leads to
( ) ( )
Or short O short short
m=1−⋅α U RV and hence:
( )
( )
short Or short
U O t=1−⋅α U RV O t (14)
( ( )) ( ( ) ( ))
( )
Op short Or short
Combining Formulae (12) to (14), U can be expressed as:
22 2 2
U O t U RV 1−ααO t+ RV (15)
( ( )) ( ) ( )
( )
O Or short short short short
From Formula (15) for the maximum measurement uncertainty of a single time step, the following
expression for the root mean square of the maximum measurement uncertainty ( RMSU ) is derived for
O
a measurement time series:
N
t
U O t
( ( ))
O
∑
t=1 2 22 2 2
RMSU U RV1−α O+ σα+ RV (16)
( )
) )
O Or short ( short ( o short short
N
t
in which and σ are the averaged measured value and the standard deviation of the measured time
O
O
series, respectively.
For annual or seasonal averaged measurements, the following expression for the maximum
measurement uncertainty is derived in a similar manner:
2 2 2 2
U O U RV 1−+α O α RV (17)
( ) ( ) )
(
O Or long long long long
The parameters U RV , , α ,, α RV and RV for O (maximum 8h daily
U RV 3
( )
( )
Or short Or long short long short long
mean and seasonal), NO (hourly and annual), PM (daily and annual) and PM (daily and annual) are
2 10 2.5
provided in Annex A (normative). Parameters for other pollutants and/or time averages are provided in
U RV and , are set to
Annex F (informative). Uncertainties at the reference value, U RV
( )
( )
Or short Or long
those in [1].
=
==
=
=
7 Data requirements in the modelling system validation
7.1 General
In the context of assessment under [1], a modelling system should be able to capture both the spatial and
temporal variability of the environmental indicator in the modelling domain. The resolution of the
modelling system should be such that all measurements can be reproduced, regardless of the
classification of the monitoring locations. When using a modelling system, the main principle for
validation is therefore that the modelling results should be validated against all measurement data that
meet the requirements defined in subclauses 7.2 to 7.4.
In exceptional circumstances, validation against all measurement data meeting these requirements might
not be appropriate. In such cases, the exemptions shall be duly justified and the reasoning shall be
reported as part of the assessment to the competent authorities.
7.2 Measurement data quality
The primary measurement data to be used for modelling validation shall comply with requirements of
measurements defined as fixed and indicative measurements.
NOTE 1 When modelling-based assessments of air quality are carried out in the context of [1], then the
requirements for fixed and indicative measurements are those found in [1].
NOTE 2 Measurements with lower data coverage than required for fixed measurements fall into the indicative
measurement category (provided that the minimum time coverage required for indicative measurements is met)
and hence are available for validation with the method described in this document.
7.3 Minimum number of sampling points
The reliability of the statistical validation of modelling performance against measurements depends
directly on the amount of data available. The dependence of the statistical strength of the evaluation on
the number of available sampling points is discussed in Annex C. For a sufficiently robust evaluation of
the MQO, a minimum number of 10 sampling points (fixed and/or indicative measurements) is required.
The reason of this number is given and discussed in Annex C. This minimum number of sampling points,
fixed to 10, is a compromise to balance the limited number of sampling points available for validation in
practice, with the risk of an incorrect evaluation of the modelling performances with the MQO, as noted
in Annex C. Below this minimum number, the MQIs for all available sampling points shall be provided to
and considered by the competent authority to decide on the acceptance of the modelling application.
Some options to cope with the situations where fewer sampling points are available are proposed in
Annex D.
When multiple measurements are available for a given pollutant at one sampling point, only the
measurement with the lowest uncertainty for the given time scale should be retained for the evaluation
process.
7.4 Independence of the measurement data from the modelling system results
For a robust modelling system validation, independent measurement data shall be used.
The combination of measurement data with modelling results is designated as data fusion or data
assimilation. So when incorporating measurement data in the modelling approach, independent
measurements shall be used when validating the modelling system. This independent set of sampling
points shall meet the data quality criteria set in 7.2 and 7.3 and the procedure to select it shall be accepted
by the competent authorities.
Annex A
(normative)
Measurement uncertainty parameters
A.1 General
Table A.1 presents the values for the measurement uncertainty parameters in Formulae (15) and (16)
and Table A.2 for the parameters in Formula (17). Parameters are set for NO (hourly and annual), O
2 3
(maximum 8h daily mean and seasonal average), PM10 (daily and annual) and PM2.5 (daily and annual)
both for fixed and indicative measurements.
A.2 Short-term averages
A.2.1 Fixed measurements
The value of the maximum relative measurement uncertainty U RV around the selected reference
( )
Or
value is as set in [1]. The remaining parameter α , is obtained through fitting of uncertainty data for
short
fixed measurements provided by AQUILA [6].
A.2.2 Indicative measurements
The maximum measurement uncertainty at the reference value equals the data quality objectives set in
[1]. The method for estimating uncertainties follows the guide for demonstration of equivalence (GDE)
[7]. Based on GDE Formula 7.30, these measurement uncertainties are computed as the square root of
the quadratic sum of the contribution from the random component (sum of square residuals, RSS) and
the bias out of a linear regression line.

RSS
 
U C=k× − u + ab+ −×1 C (A.1)
( )  ( )
indicative fixed
 
n− 2



Bias
Random
It is assumed in this document that at the limit value, the contribution from the random component is
60 % for both PM and PM whereas the bias includes an intercept (denoted as “ a ”) and a slope
2.5 10
components (denoted as “ b ”). k is the coverage factor and is set to 2. The intercept is assumed to be 10 %
for PM and 40 % for PM . See the GDE for more details on the process to use these numbers to estimate
2.5 10
the parameters of Table A.1.
In this document, the parameter α is then obtained through a least-square fit with the GDE based
uncertainty estimates at different concentration levels ( C ).
The short-term uncertainty parameters are listed in Table A.1.
Table A.1 — List of the parameters used to calculate short-term maximum relative uncertainties
for fixed and indicative measurements
Fixed measurements Indicative measurements
RV U RV α U RV α
Short-term
( ) ( )
short Or short short Or short short
NO hourly 200 μg/m 15 % 0,20 25 % 0,90
O Maximum daily
120 μg/m 15 % 0,40 25 % 0,80
8h mean
PM daily 45 μg/m 25 % 0,35 50 % 0,85
PM daily 25 μg/m 25 % 0,60 35 % 0,83
2.5
A.3 Long-term averages
The process applied for short-term averages in Clause A.2 is followed here for long-term maximum
measurement uncertainties UO , excluding particular exceptions:
( )
O
— For indicative measurements, all parameters in Formula (A.1) are equal to those in Clause A.2, with
the exception of the random component which are set to 60 % for both PM and PM whereas the
2.5 10
intercept component (denoted as “ a ”) is 20 % for PM and 25 %, for PM . The long-term
2.5 10
uncertainty parameters are listed in Table A.2.
— For O long term measurements, the value of the maximum relative measurement uncertainty
U RV around the selected reference value (60 μg/m ) are set to 15 % for fixed measurements
( )
Or
and to 25 % for indicative measurements. The remaining parameter is obtained through fitting
α
long
of uncertainty data for fixed and indicative measurements provided by AQUILA [6].
Table A.2 — List of the parameters used to calculate long-term maximum uncertainties for fixed
and indicative measurements
Fixed measurements Indicative measurements
RV α α
U RV U RV
Long term
( ) ( )
long long long
Or long Or long
NO annual 20 μg/m 30 % 0,975 40 % 0,975
O seasonal 60 μg/m 15 % 0,40 25 % 0,80
PM annual 20 μg/m 20 % 0,60 30 % 0,80
PM annual 10 μg/m 30 % 0,80 40 % 0,90
2.5
NOTE For O long-term fixed and indicative measurements, values are proposed in this document, although
not provided in [1].
The values reported above were used to produce maximum measurement uncertainty curves for each
pollutant (see Figure A.1 to Figure A.4 below).
Key
X concentration (μg/m )
Y maximum measurement uncertainty (μg/m )
short-term maximum measurement uncertainty for indicative measurements

short-term maximum measurement uncertainty for fixed measurements

long-term maximum measurement uncertainty for indicative measurements

long-term maximum measurement uncertainty for fixed measurements

Figure A.1 — Absolute maximum measurement uncertainties for PM fixed and indicative
2.5
measurements as a function of concentration

Key
X concentration (μg/m )
Y maximum measurement uncertainty (μg/m )
short-term maximum measurement uncertainty for indicative measurements

short-term maximum measurement uncertainty for fixed measurements

long-term maximum measurement uncertainty for indicative measurements

long-term maximum measurement uncertainty for fixed measurements

Figure A.2 — Absolute maximum measurement uncertainties for PM fixed and indicative
measurements as a function of concentration
Key
X concentration (μg/m )
Y maximum measurement uncertainty (μg/m )
short-term maximum measurement uncertainty for indicative measurements

short-term maximum measurement uncertainty for fixed measurements

long-term maximum measurement uncertainty for indicative measurements

long-term maximum measurement uncertainty for fixed measurements

Figure A.3 — Absolute maximum measurement uncertainties for NO fixed and indicative
instruments as a function of concentration
Key
X concentration (μg/m )
Y maximum measurement uncertainty (μg/m )
short-term maximum measurement uncertainty for indicative measurements

short-term maximum measurement uncertainty for fixed measurements

long-term maximum measurement uncertainty for indicative measurements

long-term maximum measurement uncertainty for fixed measurements

Figure A.4 — Absolute maximum measurement uncertainties for O fixed and indicative
instruments as a function of concentration
Annex B
(normative)
Stringency of the MQO: the β parameter
The values of given in Table B.1 correspond to the maximum ratio of uncertainty of modelling
β
applications as specified in [1]. In principle, the modelling uncertainty is independent of the
measurement uncertainty. In order to derive values of β for indicative measurements, they are assumed
to be equal to the values of β of fixed measurements multiplied by the ratio of the fixed and indicative
uncertainties around the reference value (RV).
NOTE The values of α for indicative
...