Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1
LPIS QA 2016 updates: Changes in the ETS 6.1
Pavel Milenov, Wim Devos, Slavko
Lemajic, Paolo Isoardi
Control and management of agricultural land in IACS workshop
Baveno, 23-25 May 2016
Outline
Observations on the 2015 exercise
Changes to the 2016 LPIS QA
2
Observations on the 2015 exercise
3
Observations on the 2015 exercise
Automated screening
• Better performance with respect to the a-tests (data
completeness, topology)
• 49 failed tests in 2014; 1 failed test in 2015
• Slightly better performance also with respect to the b-tests (data
consistency)
• 396 failed tests in 2014; 326 failed tests in 2015
• Most problematic b-tests
• reporting inspected and skipped parcels - b03 (26/39)
• calculation of area percentage - b13 (28/39)
• Calculation of area difference - b14 (25/39)
4
3
Observations on the 2015 exercise (2)
• EU MS feedback and concerns
� Welcome the “old style” ETS
� Clarifications requested for contamination and classification correctness
� Concerns with respect to the capability to perform the classification
correctness
� Questions on reporting contamination, MMU, use of aerial orthophotos
� Issues with proper interpretation of critical defects
• LPIS QA imagery
� No major issues found
5
2016 Guidance
6
4
Why the need of ETS 6.1?
Triggers for ETS changes
• EU Member state feedback
• External input
• Bilateral communication with AGRI (D3)
• Findings from the GTCAP
• Introduction of the new MTS
Nature of issues
• Methodological
• Reporting (XML/GML/AR)
• Support documentation (WikiCAP)
This presentation deals with the methodological issues only7
Why not to worry?ETS changes DO NOT affect the inspection procedure itself!
• Additions the instructions for the new elements in ETS 6.0
• for better consistency and clarity
• both inspection and reporting
• to reduce the Q&A loop
• to remove further bias in results
• Workarounds in the ETS procedure
• to accommodate methodological limitations
• discrimination of temporal/permanent grassland
• Adaptations required by the new MTS and revised scoping
approach
• general conditions for CD/waivers
• causes for non-conformity
• LUI 8
5
Clarifications to ETS instructionsThe “total absence of eligible land” is reported as critical defect only
• such observation (no eligible area found) is reported only as one single weakness
Re-establish the association between the non-conformities reported and
the RPs affected
• no effect on the procedure, but on the NonConformity.xsd
Rules for CD “Incomplete block” clarified
• applies only if LUI of the RP cannot be defined – RP cannot be measured!
• unaccounted agriculture land should be adjacent to the “missing” boundary of the RP
• perform analysis within the context of the given LPIS Implementation
Attribution of eligible landscape features to the given agriculture land
cover category of the RP
• for the correct conduction of the classification correctness test9
Example
Incomplete block or area non-conformity?
Reference parcel type is PB
LUI obviously extends up to the road
RP can be considered measurable
10
There is an indication for true and
stable boundary of the block
Area non-conformity
6
Skipping code C4
Revision of definition of C4 “LUI interpretation impossible with the given
orthoimage (C4)”, as:
• VHR image quality expected to be close to optimal
• due to the dedicated VHR-profile introduced
• C4 definition inherited from the CwRS methodology
• Often problems of interpretation are due to the particularity of the
land under inspection
• Woody grasslands
• Shrubs and trees obstructing RP boundary
• Should be subject to rapid field visit
• RPs prone to such CAPI problems are more challenging for
monitoring and update
� Therefore more risky. If skipped, a bias in ETS results might be expected.11
Evolution of C4 in time
12
8% of the total skipping reported in all LPIS QA campaigns
7
Example of skipped RP with C4 – LPIS QA 2015
13
More than 80% of
the skipped RPs
coded with C4
14
Skipped RP Measured RP
Example of skipped RP with C4 – LPIS QA 2014
8
Skipping code C4 - definition refined
• Check for occurrence of:
� isolated image processing-related
artefacts that cannot be attributed to
a particular land cover
• technology-inherent issues
OR
� land use phenomenon (smoke from
a chimney or passing airplane).
15
• When aerial othophoto is used in combination with VHR,
� Reference parcel is skipped for image-related issues when
skipping conditions are present on at least one of the image
pairs
RP_CLS: Handling Permanent and Temporary Grassland
• ETS requires differentiation between Permanent Grassland and
Arable land
• Often it is not possible from a single orthoimage
• Permanent Pasture, Temporary Grassland, and Fallow land can be
misclassified
• Classification correctness test requires a comparison between
AL/PG/PC observed with AL/PG/PC recorded in LPIS
• Data recorded in LPIS can be accurate, since it relies on multiple data
sources and field knowledge
• ETS mapping might not be accurate enough to allow comparison
16
9
• Detect, map and report only what you can “see”
• limit yourself up to the level of detail allowed by your reference
data and method used
• New temporal code is introduced in MML of the eligibility profile for
report the presence of herbaceous vegetation (HV)
• when it is impossible to identify the permanent (code G) or
temporary (code A) nature from the imagery
• the observation and interpretation can be limited to the
delineation of this observed herbaceous vegetation
RP_CLS: Proposed workaround
18
In reality Recorded in LPIS
Mapped in ETS ETS polygon assigned for the RP_CLS as
Fallow land AL HV Depending on the individual choice of ETS operator for each mapped HV polygon within RP
Natural grassland PG PG PG
Arable land AL AL AL
Temporal grassland
PG HV Depending on the individual choice of ETS operator for each mapped HV polygon within RP
RP_CLS: Proposed workaround (2)
Natural grassland Arable
landTemporal grasslandFallow land
10
ETS Adaptation Impact of IXIT and MTS
• Check of critical defects
• Rules made explicitly dependent on LPIS implementation decisions and
choices as reported in the TG IXIT (Qualifier A)
• General conditions (IXIT) clearly separated from the local conditions –
Detailed instruction 1
• Abnormal occurrence of CDs might indicated issues with some of
the other implementation options (Qualifiers B, C, D)
• CD rules are not changed!!
19
20
Example of check for critical defect
Is it CD “Invalid common RP boundaries”?
NO, if IXIT Qualifier A -> AP
YES, if IXIT Qualifier A -> PB
11
ETS Adaptation Impact of IXIT and MTS (2)
• Use of waivers
• General conditions explicitly linked with IXIT and MTS-log outcomes –
Detailed Instruction 2
• Categorization of non-conformities
• Instructions has been revised to accommodate the inputs from the
MTS-log - Detailed Instruction 3
• List of actions per given cause is considered indicative and not
exhaustive
21
Refined definition of LUI from the new scoping approach
• LUI extended to accommodate the temporal adjudication of
eligible landscape features
• associated with the reference parcel
• attributed to particular agriculture LC (AL/PG/PC)
• see TG population (ETS Annex IX)
22
LUI 1 LUI 2
12
Mapping and reporting of small non-agriculture features
• Reporting of the artificial sealed features in ETS is regardless
their size
• for accounting all potential triggers for contamination
• following the LPIS guidance of AGRI
• stating that “man-made constructions …. should be excluded from
the RP by delineation”
• However, ETS do not require a delineation of all non-agriculture
features
• only those larger than or equal to 0.03 ha
• or 0.01 ha depending on the orthoimage and nature of land feature
• other smaller features are reported as points only23
Reporting area-based non-conformities
24
Test Reported items Scope
Area conformance Report all RPs with area difference above the specific threshold
All measured RPs with comparable area
Contamination Report all RPs with un-waivered contamination
All measures RPs with correct area (within threshold)
Area correctness Report all RPs with area difference par LC category above the specific threshold
All measured RPs with comparable area, exceptthose with:• with only one type of
agriculture land cover category recorded in the LPIS AND
• same category found on the LUI during the ETS
13
Conclusion
ETS 6.1 aims to
•Facilitate and further clarify the inspection processes
• Mainly with respect to the new elements introduced in ETS 6.0
•Provide some workarounds in the ETS procedure
• To enable correct reporting
• Reduce the risk of false alarms in the ETS scoreboard
•Ensure and strengthen the link between the ETS observations
made and their meaning
• In the context of the local LPIS implementation (MTS)
• To support the analysis of the causes for non-conformity
• To enable effective remedial actions taken
25
Conclusion
Impact of ETS 6.1
• No changes in methodology
• ETS annexes will be preserved in 2016
• Some WikiCAP pages need update
• Changes in some XML/GML schema expected
• Triggered mostly by the MTS and new scoping approach
26
14
Questions?
27