Transcript METADATA TO DOCUMENT SURFACE OBSERVATION Michel Leroy, Météo-France METADATA  Metadata is necessary to use efficiently observed data.  Latitude, longitude, altitude, station Id.,

METADATA TO DOCUMENT
SURFACE OBSERVATION
Michel Leroy, Météo-France
METADATA
 Metadata is necessary to use efficiently observed data.
 Latitude, longitude, altitude, station Id., date and time are obvious
metadata.
 A detailed description of the site and the instruments used, their
characteristics, the historic of any instrument and site change, etc. is
highly recommended and wished by climatologists. But the way to
document this information is not yet standardized, this information is
often missing and when available, the information is not easy to use
by automatic means, due to its complexity.
 The site environment is one of the important factor affecting a field
measurement and its representativeness for various applications.
 Though quite well known by the meteorological services, WMO
siting recommendations are not always followed in the real world (or
cannot be followed).
 It is the same thing for the measurement uncertainty when
compared to recommended and achievable measurement
uncertainty stated in WMO doc n°8 (CIMO Guide),
Some “condensed” metadata
 In order to document the site environment and the sustained
characteristics of the measurement system in an easy to handle
way, Météo-France has defined two classifications :
 A siting classification, ranging from 1 to 5, for each basic parameter.
 A “maintained performance” classification, ranging from A to E, for
each basic measurement.
 Reducing the site characteristics and the equipment’s performances
to single numbers or letters hide many interesting details, but a
major advantage is to let the results easy to use. And these single
numbers don’t restrict an additional detailed documentation (such as
photos).
 The definition of these classifications is coming from an initial
analysis of quality factors influencing a measurement
Drawbacks
 These classifications don’t allow any corrections of the data. They
are not developed for that.
 Especially for wind, may be for precipitation, some correction
methods exist and could be applied. These methods need a detailed
knowledge of the site environment and sometimes additional
parameters. There would be a great interest in applying
standardized methods to correct raw measurements using the
available metadata of a site. But the set of metadata needed to
apply corrections is not clearly defined or standardized (except for
wind for the reduction of the measured wind to a “standard” wind at 10
meters with a roughness length of 0.03 m). It would be ideal to have
them, but this approach may be impracticable in the real world.
 The advantage of the proposed classification is its practicability in
the real world, therefore adding a practicable value to the
information.
Quality factors of a measurement
 The intrinsic characteristics of sensors or
measurement methods
 The maintenance and calibration needed
to maintain the system in nominal
conditions.
 The site representativeness
Site representativeness
 Exposure rules from CIMO recommendations.
 But not always followed and not always possible to
follow, depending on the geographical situation.
 In 1997, Météo-France defined a site classification for
some basic surface variables.
– Class 1 is for a site following WMO recommendations
– Class 5 is for a site which should be absolutely avoided for large
scale or meso-scale applications.
– Class 2, 3 and 4 are intermediate
 This classification has been presented during TECO98
in Casablanca.
Classification for wind measurements

Roughness classification : Davenport, see CIMO Guide, WMO Doc
n°8
 Siting classification
 The existence of obstacles nearly always lead to a decrease of the
mean wind speed. Extreme values are generally also decreased,
but not always. Obstacles increase turbulence and may lead to
(random) temporary increase of instantaneous wind speed.
 The following classes are considering a conventional 10 m
measurement.
 Class 1
 The wind tower must be erected at a distance of at least 10
times the height of the nearby obstacles (therefore seen under an
elevation angle below 5.7°)
 An object is considered as an obstacle if it is seen under an
angular width greater than 10°.
 The obstacles must be below 5.5 m within a 150 m distance
around the tower (and if possible be below 7 m within a 300 m
distance).
 The wind sensors must be located at a minimum distance of 15
times the width of thin nearby obstacles (mast, thin tree with angular
width < 10°).
 The surrounding country must not present any relief change
within a 300 m radius. A relief change is a 5 m height change.
 Class 2 (error 10% ?)
 The wind tower must be erected at a distance of at least 10
times the height of the nearby obstacles (elevation angle < 5.7°)
 An object is considered as an obstacle if it is seen under an
angular width greater than 10°.
 A relief change within a 100 m radius is also considered as an
obstacle.
 The wind sensors must be located at a minimum distance of 15
times the width of thin nearby obstacles (mast, thin tree with angular
width < 10°).
 Class 3 (error 20% ?)
 The wind tower must be erected at a distance of at least 5
times the height of the nearby obstacles (elevation angle < 11.3°)
 A relief change within a 50 m radius is also considered as an
obstacle.
 The wind sensors must be located at a minimum distance of 10
times the width of thin nearby obstacles.
 Class 4 (error 30% ?)
 The wind tower must be erected at a distance of at least 2.5
times the height of the nearby obstacles (elevation angle < 21.8°)

 Class 5 (error > 40% ?)
 - Obstacles are existing at a distance less than 2.5 times their
height.
 - Obstacles with a height greater than 8 m, at a distance less than
25 m.
St-Sulpice
Nord
Est
St-Sulpice
Sud
Ouest
St-Sulpice. Relevé de masques
 Class 4 for wind.
 New Radome AWS
settled at a distance of
60 m, away from the
woods  class 3
Saint Sulpice, DIRCE
Ratio of mean wind speed (10 min.) between Patac et Xaria
South winds
North winds
Classification of stations
 Between 2000 and 2006, 400 AWS have been installed
for the Radome network.
 The objective was class 1 for each parameter (Temp,
RH, wind, precip., solar radiation).
 But class 2 or class 3 were accepted when class 1 not
possible.
 Météo-France is now classifying al the surface observing
stations, including the climatological cooperative
network: ~4300 sites, before the end of 2008.
 Update at least every 5 years.
Where are we ?
Other quality factors
 Intrinsic performances
 Maintenance and calibration
 Within a homogeneous network, these factors are known
and generally the same. But Météo-France is using data
from various networks:
– Radome (554)
– Non-proprietary AWS (~800)
– Climatological cooperative network (> 3000)
 The intrinsic performances, maintenance and calibration
procedures are not the same.
Several reasons
 The objectives may be different.
 But some uncertainty objectives are sometimes (often)
unknown !
– To get cheap measurements ?
 The maintenance and/or the calibration are not always
organized !
 Within the ISO 9001-2000 certification process, MétéoFrance was forced to increase his knowledge of the
various networks’ characteristics.
Another classification !
 After site classification (1 to 5), definition of an additional
classification, to cover the two quality factors :
– Intrinsic performances
– Maintenance and calibration
 5 levels were defined :
– Class A : WMO/CIMO recommendations (Annex 1B of CIMO guide)
– Class B : Lower specs, but more realistic or affordable : “good”
performances and “good” maintenance and calibration. RADOME
specs.
– Class C: Lower performances and maintenance, but
maintenance/calibration organized.
– Class D : No maintenance/calibration organized.
– Class E : Unknown performances and/or maintenance
 This classification is called : Maintained performance classification
Air temperature
 Class A: Overall uncertainty of 0.1°C. Therefore, the uncertainty of
the temperature probe lower than 0.1°C and use of a “perfect”
artificially ventilated screen. Achievable measurement uncertainty is
0.2°C.
 Class B: Pt100 (or Pt1000) temperature probe of class A ( 0.25°C).
Acquisition uncertainty < 0.15°C. Radiation screen with known
characteristics and over-estimation of Tx (daily max. temperature) <
0.15°C in 95% of cases. Laboratory calibration of the temperature
probe every 5 years.
 Class C: Temperature probe with uncertainty < 0.4°C. Acquisition
uncertainty < 0.3°C. Radiation screen with known characteristics
and over-estimation of Tx < 0.3°C in 95% of cases.
 Class D: Temperature probe and/or acquisition system uncertainty
lower than for class C. Radiation screen or with “unacceptable”
characteristics (for example, over-estimation of Tx > 0.7°C in 5% of
cases).
Relative humidity
 Class A: Overall uncertainty of 1%! Achievable 2%.
 Class B: Sensor specified for  6%, over a temperature
range of –20°C to +40°C. Acquisition uncertainty < 1%.
Calibration every year, in an accredited laboratory.
 Class C: Sensor specified for  10%, over a temperature
range of –20°C to +40°C. Acquisition uncertainty < 1%.
Calibration every two years in an accredited laboratory,
or calibration every year in a non-accredited laboratory.
 Class D: Sensor with specifications worst than  10%
over the common temperature conditions. Calibration not
organized.
Global solar radiation
 Class A: Pyranometer of ISO class 1. Uncertainty of 5%
for daily total. Ventilated sensor. Calibration every two
years. Regular cleaning of the sensor (at least weekly).
 Class B: Pyranometer of ISO class 1. No ventilation.
Calibration every two years. No regular cleaning of the
sensor.
 Class C: Pyranometer of ISO class 2. No ventilation.
Calibration every five years. No regular cleaning of the
sensor.
 Class D: Sensor not using a thermopile. Calibration not
organized.
Other parameters






Pressure
Amount of precipitation
Wind
Visibility
Temperature above ground
Soil temperature
Status of the RADOME network
 Air temperature : Class B
 RH : Class B
 Amount of precipitation : Class B or Class C, depending
on the rain gauge used.
 Wind : Class A
 Global solar radiation : Class A for manned station, class
B for isolated sites.
 Ground temperatures : Class B
 Pressure : Class B
 Visibility (automatic) : Class B
Status of the cooperative network
 Air temperature (liquid in glass thermometers) : Class C
 Amount of precipitation : Class B
Status of non-Météo-France additional
networks







Air temperature : Class B to D
RH : Class B to D
Amount of precipitation : Class B to C
Wind : Class B to D
Global solar radiation : Class B to D
Ground temperature : Class B to C
Pressure : Class B to D
Metadata
 These classification for each site are meta data, part of
the climatological database.
 Site classification is on going.
 Maintained performance classification has been defined
this year and is being applied : is it possible to “easily”
classify the additional networks.
 With these two classifications, a measurement on a site
can be given a short description.
– Example : C3 for global solar radiation is for a class 2
pyranometer without ventilation, calibrated every 2 years,
installed on a site with direct obstructions, but below 7°.
An image of a network
E
There is still hope
D
Is it really usable ?
C
Still useful
B
Good
A
Dream ?
1
2
3
4
5
An image of the RADOME network
Conclusion
 These classifications are intended to describe the real
world of measuring networks, which is sometimes far
form the WMO/CIMO recommendations.
 WMO (CIMO, CBS) has decided to develop a site
classification, on the example of this classification. Such
a standard would be further recognized by ISO.
 This topic has been recently discussed by the CIMO
Expert Team on Surface Technology and Measurement
Techniques.
 Any suggestions or comments are welcomed. To be
addressed to Michel Leroy
Proposed change for precipitation
 Change class 1 for having in class 1 a well protected site :
homogeneous obstacles around the rain gauge which can reduce
the wind speed at the gauge level.
 Class 2 unchanged : no obstacles closer than 2 times their height.
Proposed change for temperature/humidity
 To use the climatology of wind for temperature classification.
– % of low wind speed ( < 1.5 or 2 m/s) ?
Trappes
St Denis, La Réunion
Proposed change for temperature/humidity
 The perturbation from artificial surface is greatly reduce with wind.
With a 1 m/s wind, the air moves by 60 m in one minute. The
frequency of mean wind speed (at 10 m) below 1.5 m/s could be
used to reduce the influence of artificial surface in the classification.
 The shading conditions currently used are a big constraint. It could
be partly replaced by the global angle of view of obstacles :
–
–
–
–
–
No obstacles, angle of view is 0
Obstacles everywhere : angle of view is 2P (100%).
Screen along a wall : angle of view is P (50%).
Angle of view thresholds could be 5%, 10%, 20%
But more difficult to evaluate.