LUCAS Monitoring the European Union territory using a grid approach Maxime KAYADJANIAN Eurostat F-2.

LUCAS

LUCAS Monitoring the European Union territory

using a grid approach

Maxime KAYADJANIANEurostat F-2

LUCAS

100 000 points systematically sampled

LUCAS Grid

Reason: compromise between budget and required statistical representativity at EU level

LUCAS

Why a systematic GRID ?

LUCAS Grid

LUCAS is a multipurpose system for monitoring:

• Land Use / Land cover

• Landscape and Environmental features

Reason: all the territory has to be covered the same way to ensure equiprobabilty of observations

LUCAS

300 m

300

m

Specifications

LUCAS Grid

A two-stage sampling GRID

10.000 PSU 10 SSU

Reason: optimizing trip for visiting points in the field

X

LUCAS

Particular case

LUCAS Grid

Italy Spain

Reason: complying with already existing national system

MAST system

Distance inter PSU =

• 20 km (not 18 km)

AGRIT system

Distance inter SSU =

• 300 m east-west

• 500m north-south

LUCAS

Construction of the GRID

LUCAS

Construction (1/3)

LUCAS Grid

Choice of projection

system:

GRS80 + UTM

One sampling plan per

country to limit ground

distortion

LUCAS

Construction (2/3)

LUCAS Grid

Grid generated in the UTM zone of the country

Origin chosen pseudo-randomly Total coverage of country

LUCAS

Construction (3/3)

LUCAS Grid

SSUs selected according to SABE and SCOLE (© Megrin) boundaries

Selection of SSUs Uncompleted PSUs

LUCAS

Geometrical quality of the sampling plan

LUCAS Grid

LUCAS

Geometrical quality of the sampling plan

LUCAS Grid

Are points regularly spread on the ground?

N

S

18 km18 km

N

S

18 km

18 km ?

EW

N

S

18 km

Once grid generated... …in the UTM projection system...

…how much ground distance is affected ?

LUCAS

LUCAS Grid

Largest line of PSUs in France...

24,05

24,10

24,15

24,20

24,25

24,30

24,35

24,40

24,45

24,50

4 9 14 19 24 29 34 39 44 49 54

n° PSU

1/10

0 d

egre

e

Meridian 30 Meridian 31-3° +3°

PSU col n°

…shows a longitudinal angle reduction apart from central meridian

- 45m

…it corresponds to a reduction of 45m on the ground 500 km apart from central meridian.

Geometrical quality of the sampling plan

LUCAS

LUCAS Grid

How many points are affected by distortion ?

Point out of central median zone 30%

Geometrical quality of the sampling plan

Conclusion: distortions are negligible for statistical estimates

LUCAS

The case of SPAIN (MAST)

No distortion but small gaps

LUCAS Grid

Geometrical quality of the sampling plan

LUCAS

Reaching the point on the ground

LUCAS Grid

A fundamental aspect to ensure a correct monitoring

LUCAS

Reaching the point on the ground

3 basic tools

LUCAS Grid

3/ GPS for navigating along transects

1/ Topographical maps for navigating between PSUs

2/ Aerial photo for reaching precisely the SSU

LUCAS

Reaching the point on the ground

LUCAS Grid

Observation unit

Point location

10 m

0.5 / 2.5 m

1 m

3 m

GPS

DGPS

Image resolution

Orthophoto

5 m Map: 1/25 000

1 m

Positional accuracy

3 m

At the point: observation in a 3 m window

LUCAS

Reaching the point on the ground

LUCAS Grid

Country Resolution /Scale

Year Type Use ofGPS

AT 0,5 m 1990/2000 Orthophoto Yes

BE 1 m 1995/2000 Orthophoto Yes

DE 0,25 to 0,8 m

>1995Orthophoto

& aerial photoyes +DGPS

DK 0,4m 1999 Orthophoto yes

ES 0,75 to 2 m

1985/2000Orthophoto

& aerial photono

FI 1 m 1996 Orthophoto yes

FR 0,5 to 1 m >1995Orthophoto

& aerial photono

GR 1 m 1990/1998 Orthophoto yes

IT 1 m ? Orthophoto yes

LU 1/5000 1997 Topogr. map yes

NL 1m >1995 Orthophoto DGPS

PT 1 m 2000 Orthophoto yes

SE 1 m >1995 Orthophoto Yes

LUCAS

Implementation problem

Projection of UTM x,y into national system

LUCAS Grid

Systematic shift of about 200 m between GPS co-ordinates and location on the photo detected during the ground mission control, in:

• Luxembourg: specific parameters for datum

• Greece & Portugal: bug in the GIS soft

For these countries the original sampling plan has been corrected in return

LUCAS

Conclusion

• Reaching the point: Good quality of field document

• Location of points: only minor distortion in the sampling plan (no impact on estimates)

• Well documented

LUCAS can be considered as a strong system: