data quality control procedures within the common european phenological data platform cost 725
DESCRIPTION
Data Quality Control Procedures within the Common European Phenological Data Platform COST 725. Ana Žust, Andreja Sušnik Environmental Agency of the Republic of Slovenia. 6 th COST 725 MC MEETING. Volos, Greece, September 21, 2006. September 21, 2006. The subject of the presentation: - PowerPoint PPT PresentationTRANSCRIPT
Volos, Greece, September 21, 2006
Data Quality Control Procedures within the Common European Phenological Data Platform
COST 725
6th COST 725 MC MEETING
Ana Žust, Andreja SušnikEnvironmental Agency of the Republic of Slovenia
6th COST 725 MC meeting, Volos, Greece
September 21, 2006
The subject of the presentation:
Results of the QUESTIONNAIRE on national QC procedures
(the decision accepted on 5 th MC meeting in Dublin)
The primary goal of the questionaire:
To assemble information on national QC procedures and to provide the documentation on data quality assurance of the Common European Phenological Database
6th COST 725 MC meeting, Volos, Greece
September 21, 2006
1.Austria 10. France2.Germany 11. Ireland3.Slovakia 12. Finland4. Slovenia 13. Poland5. Romunia (two institutions) 14. Norway6. Litva 15. Spain7. Latvia8. Luxembourg9. UK
The countries which responded to the questionaire:
6th COST 725 MC meeting, Volos, Greece
September 21, 2006
1. Who is the owner of the data?
NHMS, IPG, Research Institutions, Educational and Research Institutions, Private Companies
Government 9
Research 7
Education and research
1
Private 1
Not answered14
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
3. Do the countries perform QC?
A common protocol?For the needs of the common EPDB?
QC -EPDB 3
no QC 2
QC 11
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
4. Use of the software programmes to alleviate the procedures of QC?
7 9
0
5
10
yes no
5. How to keep of data entering errors?
Data entry is inherently prone to errors both simple and complex.
Software programmes with incorporated constraints, syntax check constraints
Simultaneous visual control
Training the observers
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
Steps of QC procedures
6. Visual control - elimination of first sight errors
12
40
5
10
15
visual control no visual control
Correct date formatCorrect phenological phase entry
Completeness of the data.
.
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
7. Logical control - eimination of rough errors
4
3
4
1112
5
no control
comparison of the data
use of climatic models
timing check
corr.-infeasivle dates
corect sequence
interphase duration
allowed period (constraints)
Steps of QC procedures
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
8. Statistical data control- tracing of extreme values – outliers
2
111
11
no statistical control
expected range of phen. dates
data corelation methods
optimal data fitting
outliers detection (3 sigma)
Steps of QC procedures
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
9. How do countries treat the outliers?
7
2313
0
4
8
12
16
20 nothing
contacting the observer
possibility /doublecheck/idividual/bioclimatological analysisflag
exclude - separate the data
Steps of QC procedures
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
10. Spatial control - comparison and correlation of data in space
11
11111
spatial coherence / isophenasmapsoptimal data fitting
comparable stations
latitude bands
GIS tools
no control
Steps of QC procedures
September 21, 2006
6th COST 725 MC meeting, Volos, Greece
Summary Different approaches in national QC procedures (different phenology monitoring
programmes, different spatial coverage (stations, plants and phases) different data periods),
Most frequent used visual and logical control, Less used statistical and spatial control, The methods are frequently combined with each other, Individual data treatment, needs of a very experienced phenologist The importance of exact observation rules and training of phenological observers A few countries are in the procedure of developing seperate steps of QC
procedures
6th COST 725 MC meeting, Volos, Greece
September 21, 2006
11. What is the national estimate of the quality level of the data provided for EPDB?
low1
medium8
high7
Data ownership and custodianship confers responsibilities for its management, quality control and maintenance of the data
6th COST 725 MC meeting, Volos, Greece
September 21, 2006
CONCLUSIONS AND RECOMMENDATIONS
•National phenological databases distributed in the common EDPB should be quality checked in advance by the country – owner of the data,
•Further QC on the level of the common EPDB due to unhomogeneous coverage of different phenological objects is questionable – risk of erasing true extremes, (visualisation of the spatial coverage of phenological objects by points on maps in friendly user manner – ZAMG example)
•In the case that national QC is impossible the data should get a flag,
6th COST 725 MC meeting, Volos, Greece
September 21, 2006
CONCLUSIONS AND RECOMMENDATIONS
•The usage of the data should start,
•To accept the resort to detect the suspicious data or extremes through analyses,
•The usage of the data in different analyses will additionaly enable platform to get a higher quality status.