Data Quality

Browse

General

Data Availability

Data Quality

The information presented on data quality is based on differences between observations and the values of the most recent ECMWF forecast ("first guess") of the same parameter. Depending on the time of the observation, the forecast range is between 9 and 15 hours. The ability of a modern data assimilation system to provide the diagnostic facilities to monitor the performance of the observational network is demonstrated by A. Hollingsworth et. al., Monthly Weather Review, Vol 114, No. 5, May 1986.

It should be noted that:

(i)	all results are based on software that may undergo further development;

(ii)	although the quality of the ECMWF first-guess fields is of a generally
	high standard this is only true to a limited extent in the tropics,
	where small-scale processes such as convection are of much greater
	importance than in mid-latitudes, and the observations will sometimes
	not be representative of the scales of motion given by the first-guess;

(iii)	the first-guess fields themselves will vary in accuracy depending on the
	density and quality of data, particularly in the upstream regions and
	over Antarctica and the southern hemisphere mid-latitudes. Direct
	comparisons between stations (or airlines) should preferably be restricted
	to observations in a reasonably homogeneous climatic region.

Tables 1-9 contain lists of SHIPs (including fixed marine platforms), DRIFTERs, TEMPs and TEMPs/PILOTs believed to have supplied suspect reports of surface pressure, geopotential height or wind during the month. The format of the tables is according to Recommendation 3 CBS-Ext(85) and the criteria for stations or data platforms to be classified as suspect are given at the top of each table. For tables 7 and 8 data for the worst standard pressure level are shown. Units of RMS, standard deviation and bias are hPa in tables 1 and 4, m in table 7 and ms^-1 in tables 2, 5 and 8. In tables 7 and 8 the station position is indicated; in the case of TEMPSHIPs and PILOTSHIPs this position is obtained from the first report of the month. The gross error limits for first-guess deviations of geopotential in table 7 are as follows:


Level	Geop


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

The corresponding limits for wind (table 8) are:


Level	Wind


1000	35ms^-1
925	35ms^-1
850	35ms^-1
700	40ms^-1
500	45ms^-1
400	50ms^-1
300	60ms^-1
250	60ms^-1
200	50ms^-1
150	50ms^-1
100	45ms^-1

In table 7 the weighted RMS values at standard levels are calculated using the following weights:


Level	Weight


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

Tables 10 and 11 provide geopotential and wind quality statistics (100 hPa level) for TEMPSHIPs and PILOTSHIPs received during the month. Units and display format are identical to those in tables 7 and 8 respectively. Tables 13, 14 (50 hPa), 15 and 16 (100 hPa), 17 and 18 (500hPa), 19 and 20 (850hPa) provide similar radiosonde statistics for the EUCOS area.

Tables 21-23 are similar to tables 4-6 with data coverage restricted to the EUCOS area.

Figures 14-18 show global charts of SATOB and aircraft wind quality, where the statistics have been averaged over latitude/longitude boxes of 5 degrees square, and the mean observed minus first-guess (or 'bias') wind vectors have been plotted. All observations in the specified layers have been used. For comparison the mean observed wind (from the SATOB reports only) for each layer is shown in figures 14 and 15. A reference value of wind speed is plotted in the top right corner of each figure. An arrow is only plotted if 10 or more observations have been received in that 5 degree square.

Table 12 provides quality statistics of aircraft wind observations in the layer 300-150 hPa stratified by airline carrier. The format and specifications of the table have been defined by NMC Washington, the lead centre for the monitoring of aircraft and satellite data.

Summary of Revisions (in reverse order)


Revision 23 (Dec 00) -  Coverage charts for Noaa_14 MSU replaced by
                        ATOVS AMSU-A for Noaa_16

Revision 22 (Aug 99) -  Coverage charts for TOVS thickness 300-100
                        hPa replaced by (A)TOVS AMSU-A and MSU
                        (Noaa_15 and Noaa_14)

Revision 21 (May 99) -  Monitoring statistics ceased for Noaa_11 as
                        satellite is no more available.

Revision 20 (Sep 98) -  Changes to tables and annex to remove all
                        mention about data usage. Two more levels
                        (50 and 850 hPa) added to the COSNA statistics
                        for Sondes.

Revision 19 (Jul 98) -  From June 29th, 1998 ECMWF model assimilates
                        temperature data instead of geopotential from
                        radiosondes. As a consequence the number of
                        used geopotential data drops to zero in tables
                        7, 10, 13 and 15.

Revision 18 (Apr 98) -  Changes to tables and annex to introduce the
                        usage of accepted numbers and observations
                        instead of precentage of rejection.

METOPS-ECMWF

15-05-2012