1. General remarks

1.1 Centres participating in the exchange of monitoring results will implement standard procedures and use agreed formats for communicating the information both to other centres and to the data providers. The following list is incomplete and requires further development in the light of practical experience. Guidance will be given through the initiative of the lead centres in their corresponding fields of responsibility.

1.2 Lead centres who are informed of remedial actions being taken should provide this information to all participating centres. The WMO Secretariat shall forward, every six months, the information it receives to the relevant lead centres. All lead centres shall produce for the WMO Secretariat a yearly summary of information made available to them and/or of those actions taken within their area of responsibility.

2.1 Monthly exchange for upper-air observations should include lists of stations/ships with the following information.

2.1.1 List 1: GEOPOTENTIAL HEIGHT

Month/year
Monitoring centre
Standard of comparison (first-guess/background field)

Selection criteria: FOR 0000 AND 1200 UTC SEPARATELY, AT LEAST THREE LEVELS WITH 10 OBSERVATIONS DURING THE MONTH AND 100 M WEIGHTED RMS DEPARTURE FROM THE FIELD USED FOR COMPARISON BETWEEN 1000 hPa AND 30 hPa.

The gross error limits to be used for observed minus reference field are as follows:

Level             Geop

1000 hPa  100 m
925  hPa   100 m
850 hPa    100 m
700 hPa    100 m
500 hPa    150 m
400 hPa    175 m
300 hPa    200 m
250 hPa    225 m
200 hPa    250 m
150 hPa    275 m
100 hPa    300 m
70 hPa      375 m
50 hPa      400 m
30 hPa      450 m

Weights to be used at each level are as follows:

Level             Weight

1000 hPa  3.70
925 hPa    3.55
850 hPa    3.40
700 hPa    2.90
500 hPa    2.20
400 hPa    1.90
300 hPa    1.60
250 hPa    1.50
200 hPa    1.37
150 hPa    1.19
100 hPa    1.00
70 hPa       0.87
50 hPa       0.80
30 hPa       0.64

Data to be listed for each station/ship should include:

        -    WMO identifier
        -    Observation time
        -    Latitude/longitude (for land stations)
        -    Pressure of the level with largest weighted RMS departure
        -    Number of observations received (including gross errors)
        -    Number of gross errors
        -    Percentage of observations rejected by the data assimilation
        -    Mean departure from reference field
        -    RMS departure from reference field (unweighted)
        -    Gross errors should be excluded from the calculation of the mean and RMS departures.
            They should not be taken into account in the percentage of rejected data (neither the numerator nor denominator).

2.1.2 List 2: TEMPERATURE

Besides the geopotential height, temperature monitoring should be included at standard levels. As an initial criteria the gross error thresholds to be considered could be:

15 degrees for p>700 hPa
10 degrees for 700>=p>50 hPa
15 degrees for p <= 50 hPa

2.1.3 List 3: WIND

Month/year
Monitoring centre
Standard of comparison (first-guess/background field)

Selection criteria: FOR 0000 AND 1200 UTC SEPARATELY, AT LEAST ONE LEVEL WITH 10 OBSERVATIONS DURING THE MONTH AND 15 m s-1 RMS VECTOR DEPARTURE FROM THE FIELD USED FOR COMPARISON, BETWEEN 1000 hPa AND 100 hPa.

The gross error limits to be used are as follows:

Level             Wind

1000 hPa  35 m s-1
925 hPa    35 m s-1
850 hPa    35 m s-1
700 hPa    40 m s-1
500 hPa    45 m s-1
400 hPa    50 m s-1
300 hPa    60 m s-1
250 hPa    60 m s-1
200 hPa    50 m s-1
150 hPa    50 m s-1
100 hPa    45 m s-1

Data to be listed for each selected station/ship should include:

        -    WMO identifier
        -    Observation time
        -    Latitude/longitude (for land stations)
        -    Pressure of the level with largest RMS departure
        -    Number of observations received (including gross errors)
        -    Number of gross errors
        -    Percentage of observations rejected by the data assimilation
        -    Mean departure from reference field for u-component
        -    Mean departure from reference field for v-component
        -    RMS vector departure from reference field
        -    Gross errors should be handled in the same way as for List 1.

2.1.4 List 4: WIND DIRECTION

The method used for computing the wind direction bias should be included in the reports (clockwise or anticlockwise)

Month/year
Monitoring centre
Standard of comparison (first-guess/background field)

Selection criteria: FOR 0000 AND 1200 UTC SEPARATELY, AT LEAST FIVE OBSERVATIONS AT EACH STANDARD LEVEL FROM 500 hPa TO 150 hPa, FOR THE AVERAGE OVER THAT LAYER, MEAN DEPARTURE FROM REFERENCE FIELD AT LEAST +/- 10 DEGRESS, STANDARD DEVIATION LESS THAN 30 DEGREES, MAXIMUM VERTICAL SPREAD LESS THAN 10 DEGREES.

Same limits for gross errors as above. Data for which the wind speed is less than 5 m s-1, either observed or calculated, should also be excluded from the statistics.

Data to be listed for each selected station/ship should include:

        -    WMO identifier
        -    Observation time
        -    Latitude/longitude (for land stations)
        -    Minimum number of observations at each level from 500 hPa to 150 hPa (excluding gross errors and data                   with low wind speed)
        -    Mean departure from reference field for wind direction, averaged over the layer
        -    Maximum spread of the mean departure at each level around the average
        -    Standard deviation of the departure from reference field, averaged over the layer
            (to be completed with information from other lead centres)

NOTES:

(1) The responsibility for updating this attachment rests with the lead centres.

(2) Urgent changes to this attachment recommended by the lead centres shall be approved, on behalf of the Commission for Basic Systems, by the president of the Commission.

2.1.5 The Profilers should be monitored (suspect platforms) using the same criteria as for the radiosondes.

3.1 The criteria for the production of monthly list of suspect stations are as follows:

3.1.1 List 1: MSL PRESSURE

Element: MSL pressure, surface synoptic observations at 0000, 0600, 1200 or 1800 UTC compared to the first guess field of a data assimilation model (usually a six-hour forecast).

Number of observations: at least twenty for at least four observation times, without distinguishing between observation times.

One or more of the following:

Absolute value of the mean bias > 4 hPa
Standard deviation > 6 hPa
Percentage gross error > 25% (gross error limit: 15 hPa).

3.1.2 List 2: STATION LEVEL PRESSURE

The criteria for station level pressure monitoring is the same as for MSLP above.

3.1.3 List 2: GEOPOTENTIAL HEIGHT

Element: Geopotential height, from surface synoptic observations or derived from station-level pressure, temperature and published station elevations at 0000, 0600, 1200 or 1800 UTC compared to the first-guess field of a data assimilation model (usually a six-hour forecast).

Number of observations: at least twenty for at least four observation times, without distinguishing between observation times.

One or more of the following:

Absolute value of the mean bias 30 m
Standard deviation 40 m
Percentage gross error 25 per cent (gross error limit: 100 m).

3.1.4 PRECIPITATION

General guidance reflecting Global Precipitation Climatology Centre (GPCC) procedures for precipitation quality monitoring is given in section 6.3.3.1 of the Guide on the Global Data-processing System.

NOTES:

(1) All monitoring centres are asked to conform to the above specified criteria. These monthly lists should be prepared for at least the regional association of the lead centres and, if possible, for other regional associations. Consolidated lists of suspect stations should be produced every six months by the lead centres (January–June and July–December) and forwarded to the WMO Secretariat for further action.

(2) The stations on these consolidated lists should be those appearing on all six-monthly lists of the lead centres. Other stations could be added to the consolidated list if the lead centres judges that there is sufficient evidence for their inclusion. Each centre should send its proposed consolidated list to all participating monitoring centres for comment. The final list would then be forwarded to the WMO Secretariat.

4.1 Monthly exchange for surface marine observations should include lists of ‘suspect’ ships/buoys/platforms with the following additional information:

Month/year
Monitoring centre
Standard of comparison: first-guess/background field of a global data assimilation model –
often a 6-hour forecast, but the background values may be valid at the observation time for non-main hour         data using 4-D VAR or time-interpolation of T+3, T+6, T+9 forecasts, say; for SST the first-guess/background      field may be from a previous analysis.

All surface marine data may be included, not just observations at the main hours of 00, 06, 12 and 18 UTC.

4.2 The elements to be monitored should include

mean sea level pressure
wind speed
wind direction
and, where possible:
air temperature
relative humidity
sea surface temperature

4.3 Data to be listed for each ship/buoy/platform and each element should include:

        -    WMO identifier
        -    Observation times (if not all times)
        -    Latitude/longitude (for buoys and platforms)
        -    Number of observations received (including gross errors)
        -    Number of gross errors
        -    Percentage of observations rejected by the data assimilation quality control
        -    Mean departure from reference field (bias)
        -    RMS departure from reference field
        -    Gross errors should be excluded from the calculation of the mean and RMS departures.
                They should not be taken into account in the percentage of rejected data (neither the numerator nor denominator).

4.4 The criteria for the production of the monthly list of suspect stations are as follows:

4.4.1 List 1: Mean Sea Level Pressure

Number of observations: at least 20.

One or more of the following:

Absolute value of the mean bias > 4 hPa
Standard deviation > 6 hPa
Percentage gross error > 25% (gross error limit: 15 hPa).

4.4.2 List 2: Wind Speed

Number of observations: at least 20.

One or more of the following:
Absolute value of the mean bias > 5 ms^-1
Percentage gross error > 25% (25 ms^-1 vector wind).

4.4.3 List 3: Wind Direction

Data for which the wind speed is less than 5 ms^-1, either observed or calculated, should be excluded from the statistics.

Number of observations: at least 20.

One or more of the following:

Absolute value of the mean bias > 30^o
Standard deviation > 80^o
Percentage gross error > 25% (gross error limit: 25 ms^-1 vector wind)

4.4.4 List 4: Air Temperature

Number of observations: at least 20.

One or more of the following:

Absolute value of the mean bias > 4^oC
Standard deviation > 6^oC
Percentage gross error > 25% (gross error limit: 15^oC).

4.4.5 List 5: Relative humidity

Number of observations: at least 20.

One or more of the following:

Absolute value of the mean bias > 30%
Standard deviation > 40%
Percentage gross error > 25% (gross error limit: 80%).

4.4.6 List 6: Sea Surface Temperature

Number of observations: at least 20.

One or more of the following:

Absolute value of the mean bias > 3^o C
Standard deviation > 5^oC
Percentage gross error > 25% (gross error limit: 10^oC).

5.1 The criteria for the production of the monthly list of suspect Aircraft temperatures and winds observations are as follows:

5.1.1 Automated aircraft observations, both AMDAR and ACARS, will separately be listed as suspect for temperatures and winds in three pressure categories if the data statistics exceed criteria defined in Table below. The three pressure categories are: Low surface to 701 hPa; Mid 700 to 301 hPa; and High 300 hPa and above. To be considered suspect, the number of observations must meet minimal counts and the data statistics versus the guess must exceed at least one criterion or the gross rejection rate must exceed 2%. Thus if the magnitude of the temperature or speed bias exceed the criterion or the RMS differences to the guess exceed the limit for the pressure category, the aircraft is listed as suspect for that pressure category. Observations differing from the guess by amounts larger than gross check limits will be considered gross and not used in computing bias and RMS differences. If the number of gross observations (NG) for a pressure category exceeds 2% of the total number of checked observations, then the aircraft will be listed as suspect. After data thinning for assimilation, the remaining number of observations is NT. The number of rejected observations excluding thinning (NR) is an optional statistic for information, and for which operational practice should be documented

List: Temperature and wind

Month/year
Monitoring centre
Standard of comparison (first guess/background field)

Each aircraft that is suspect will be listed as follows in one line:

        -    aircraft ID
        -    pressure category
        -    total number of available observations (NA)
        -    NG
        -    NT
        -    NR
        -    bias
        -    RMS difference to the guess
        -    for wind reports, the number of exactly calm winds (NC).

5.1.3 AIREP

Monthly exchange for AIREP observations should include lists of airlines with the following information:
Month/year
Monitoring centre
Standard of comparison (first guess/background field)
Selection criteria
            Number of observations>=20
Levels monitored
            300 hPa and above
Elements monitored
            Wind and temperature

Data to be listed for each airline

        -    Airline ID
        -    Number of observations
        -    Number of rejected observations
        -    Number of gross errors
        -    Number of calm winds (<5 m/s)
        -    RMS excluding gross errors
        -    Bias excluding gross errors (wind speed and temperature)
        -    Gross error limits are:
                    Wind 40 m/s
                    Temperature 10 degrees