joint research centre - european commissioncc threshold is

Post on 22-Jul-2020

1 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

The European Commission’s

science and knowledge service

Joint Research Centre

A modern JRC

in a modern Commission

Pär J Åstrand & al.

Title: Closing the 2016 controls

campaign, and preparing for 2017

Lisbon, PT, 24-25 November, 2016

Outline

1. Closing 2016, and Campaign issues 2016 -> 2017

postIR, Campaign Results statistics, Image acquisition statistics

2. Preparing 2017

preIR, status and ongoing iterations

3. Sentinel2 usage in the CwRS

Sentinel2 (S2) compared to the HR profile

EC, MS, share responsibility for an effective and efficient control

• Since 1995 through 5 CAP Reforms with its relative Regulations

• Call for collaboration – a ‘shared management’ , and ‘smooth

operation” to implement the EAGF and to operate the Funds

Preamble:

postIR, Campaign Results statistics; 2016

PreIRs gives:

• generic information, methods, profiles, justifications

Campaign Results statistics = our earlier so-called “summary stats”, gives:

• Average farm size and parcel size, OTSC results (traditional, CwRS, at

application, and at parcel level, coding etc.), ‘Other’ information (e.g. use of

which DEM, if use of S1/S2 etc.)

• Essential for MS to complete these. Only with them, and couple these with

ongoing QC, can JRC give MS ‘best practices’ and technical guidance.

• All MS completed preIR, and postIR, but only partially the Campaign Result

statistics; completion should be done in early 2017 when all Campaign results

are available.

1. Closing 2016, and Campaign issues 2016 -> 2017

… information given allows JRC to follow MS through Campaigns

e.g. do we experience an increase in greening control by CwRS ?

Changes from preIR, postIR, and in stats …

???

1. Closing 2016, and Campaign issues 2016 -> 2017

Image acquisition statistics; 2016

What was acquired in 2016?

Some lessons learnt for Campaign 2017

More details will be given in:

1. Challenges and solutions for more efficient HR/HHR image acquisition (Airbus)

2. VHR: trade-off between capacity and quality (EUSI); presentation and poster

1. Closing 2016, and Campaign issues 2016 -> 2017

Control zones

- VHR only 277 zones

- VHR+HR 360 zones

- Aerial+HR 38 zones

∑ VHR 637 zones

Av. images/zone 1.2

∑ HR 398 zones

Av. images/zone 2.5

- Sentinel2 (S2) used by

13 MS/MS regions

CwRS Methods (2016 final)

1. Closing 2016, and Campaign issues 2016 -> 2017

Image Statistics - VHR evolution (1000’s km2)

1. Closing 2016, and Campaign issues 2016 -> 2017

In yellow - forecast

Image Statistics - VHR details

CwRS

• Success rate: 99.9%

• Main sensor: WV2 67 %

• Average cloud cover: 2.09%

• Average Elevation Angle: 62,37⁰

• Mean acquisition time: 17 days

LPIS QA

• Success rate: 100%

• Average CC: 0.20% (maximum value 3%)

• Average Elevation Angle: 83.15⁰ (minimum value 78.50⁰)

1. Closing 2016, and Campaign issues 2016 -> 2017

WV4 Launched Nov. 11, 2016

Image Statistics - HR evolution (# images)

1. Closing 2016, and Campaign issues 2016 -> 2017

In yellow - forecast

Image Statistics - HR details

• Success rate: 95.5%

• Mean acquisition time: 16 days

• Average CC 0.6%

• Split of profiles

• HR MSP Profile ≤ 25m Sentinel2 (S2)

• Challenges:

• Weather conditions spring/summer (FR), late summer (IE)

1. Closing 2016, and Campaign issues 2016 -> 2017

Acquisition scenario is complex … (1)2016 -> 2016issues; 2016 ->2017

1. Closing 2016, and Campaign issues 2016 -> 2017

• Possible discontinuation of HR in favour of S2 (13 MS/MS Regions used

S2 in 2016)

• Use HHR where S2 not sufficient

• Can HHR substitute VHR2?

• Use of right profile (e.g. A6)

• Windows - “earliest” “latest” start dates, dead period, length (6 weeks)

• Feasibility HHR (making use of above)

• Feasibility VHR

• Method chosen must fit an effective CAP check approach

Acquisition scenario is complex … (2)

UK DMC-22

HR -> S2 ->

HHR

SENTINEL-2A

HR -> S2 ->

HHR

SPOT 7

HR -> S2 ->

HHR

preIR, status and ongoing iterations; 2017

JRC role in the ‘shared management’ and ‘smooth operation’

• Fine-tuning of image requests

• Analysis of preIRs inserted in G4CAP (all MS met deadline 01/11/2016)

• Optimize methods, optimize costs

• VHR2 -> HHR where possible

• HR -> S2 where possible

• applying Indicators/budget [9.13 M euro] [present request 0.5M too

high]

• Cost of requested satellite images per total OTSC area (EUR/ha)

• % of AOI area used for control

• developing other indicators

• Farm reference parcel (RP) spread, and target coverage of a holding -> efficient AOIs

• % of eligible area in AOI

• … other indicators from postIR and Campaign Results statistics

2. Preparing 2017

Cost of requested satellite images per total OTSC area (EUR/ha)

Iterations ongoing …

Preparing 2017

% of AOI used in control

2. Preparing 2017

MS average 28%

Campaign 2017 image costs distribution

8 MS use approx. 77% of the budget, but also count for 72% of the EU farm applications

2. Preparing 2017

Next steps for 2017

Continue contacts with ‘outliers’ MS here in Lisbon

• Finalize image requests and workprogramme before end of 2016

Autumn zones

• Bilateral contacts with JRC

• Start S2alert, or other mechanism, for S2 as soon as window opens

Next deadline for MS Administrations

• Insert shapefiles in G4CAP, prepare zone shapefiles/IRs/windows in

G4CAP before January 15th

• Complete Campaign Result statistics asap (early 2017) in G4CAP

2. Preparing 2017

Comparison S2 vs. HR/HHR in Campaign 2016

• Copernicus [S2] acquisition efficiency vs. [SPOT6/7, and DMC22]

• Visit poster … (S2alert statistics, and Feedback from MS/contractors survey)

3. Sentinel2 usage in the CwRS

Note.

Sample is 65% (630/966 of total HR windows of Campaign 2016)

Original windows requested by MS at campaign start

CC threshold is <5.3% (AB had 2 retained (5,1% and 5,2% CC) same threshold was set for S2

No partials included

Analysis - whole campaign, by window, by no. of S2 acquisitions - 65% sample

At least 1 complete acquisition / window

3. Sentinel2 usage in the CwRS

Note.

Number of acquisitions and respective CC%

Analysis – Cloud Cover (CC) over AOI on images acquired

3. Sentinel2 usage in the CwRS

HR+1 satellite

zones 352/966;

the most used

window

231 (66%) acquired by AB

235 (66%) acquired by S2

Trends: Both Airbus fleet and S2 acquire well in the South and

East, randomly in central EU, and worse in NW

Geographic distribution of HR+1 window

3. Sentinel2 usage in the CwRS

Figures: ESA hubs some figures …

• approx. 200 API calls are needed for one single Sentinel Request (SR)

from G4CAP (a typical 4w window)

• Each call (search in catalogue, metadata DL or QL DL) is made to the

COPHUB and takes 1-3 seconds (APIHUB is up to 4 times slower …)

• 1 SR 200x3 seconds = 10min

• Start your SR as soon as your window opens (avoid backlog)

• DL of imagery from APIHUB

• APIHUB is very unstable and DL speeds vary greatly

• Possibility to use other hubs (Amazon, Google …)

3. Sentinel2 usage in the CwRS

Figures: ESA hubs some figures …

Fig. shows typical DL speed of the APIHUB during Sept/Oct. It checks for latest ingested S2 image

every 10 minutes and then does a DL of the B8 band. It is especially unstable after 27 Sept (one-

granule-per-scene delivery), because it cannot handle the large volume of requests, or due to

changes in access protocols etc. Prior to 27 Sept, overall DL speed was much better, but of course,

scene size also much larger.

3. Sentinel2 usage in the CwRS

Observations from this analysis:

1.) Comparing S2 with Airbus (SPOT 6/7, and DMC-2) acquisition

efficiency for whole campaign, and for each zone window

• Overall slightly less success rates for S2 compared to Airbus fleet at CC<5% over the

control zones AOIs -> 66% / 75%

• Same success rate for largest window (HR+1) –> 66%

• Airbus fleet appears better in autumn/winter and in shorter windows

2.) Where (geographically) does one perform better than the other?

• Both Airbus and S2 perform well in the South and East where S2 could be used to

substitute the F0. MSP profile

• Both Airbus and S2 perform random in the centre of EU, and overall worse in the NW,

where S2 will be coupled with the HHR profile.

3.) ESA hubs

• COPHUB calls are up to 4 times faster than APIHUB to retrieve metadata. The APIHUB is

currently very unstable and image DL speeds vary greatly.

3. Sentinel2 usage in the CwRS

Overall Conclusions

1. Closing 2016, and Campaign issues 2016 -> 2017

• Good image acquisition success rates 99.9% VHR, 95.5% HR; difficult weather

conditions in summer (FR and IE)

• Need more time to analyse campaign results “summary stats” – MS late in input.

Essential for MS to give input to JRC, and to couple with ongoing QC in order to give

best technical guidance.

2. Preparing 2017

• Iterations ongoing with selected MS to fit budgetary ceiling; reductions within “effective

controls”

3. Sentinel2 usage in the CwRS

• Quite good success rates for S2. Equivalence with HR especially in the S and E. Couple

with HHR otherwise. MS should continue trials with S2A (and S2B will soon come …)

top related