corine tutorial

7/23/2019 Corine Tutorial

1/132


2/132


3/132

LEGAL NOTICE

N either the European C om m ission nor any person

acting on behalf of the C om m ission is responsible for the use w hich m ight

be m ade of the follow ing inform ation

EU R 17288 EN

EC SC -EC -EA EC , B russels Luxem bourg, 1997

Printed in Italy


4/132

A prs avoi r lu ce document, il ne semble pas inuti lede rappeler aux auteurs quen matire

doccupation du sol:

parso pi conveniente andare dri tto alla veri teff ettuale della cosa che all' immaginazione di essa

(M achiavelli )

i l est mme possib le d ajouter que ce qui donne leplus penser dans le domaine de loccupation dusol qui donne penser est que nous ne pensons pas

encore loccupation du sol.

Yves Heymann ( )

( ) the father of CO R INE La nd Cover and first project leader

iii


5/132

Th is guide was prepared by

Vanda Perdigo ( ) and Alessandro Annoni ( )

( ) Europea n Commission - Joint R esearch Centre - Space Applicat ions Institute - AI S unit - I spra

Th is guide integrates and develops the results of the previous work s of JRC and RSDE :

Technical guide for CORINE land cover updating. R SD E ,1996 (contra ct n. 9601956).

Provision of prototype demonstration computer system for updating the CORINE Land Cover

data-base. R SD E, 1993 (contract n. 3952 /D JM /D JM).

System definition for updating CORINE Land Cover data-base. R SD E, 1992 (contract n. 4435-91-08

ED ISP I).

and the tests carri ed ou t by di ff erent nati onal teams on the JRC prototype system Co-Pi lot:

Final evaluation of the prototype system for updating the CO R INE La nd Cover data base .

G2

ERE, 1996 (cont ract. 11467-95-11 F1PC ISP I )

Test of the Protot ype system for updating the CO R INE La nd Cover data base: The Netherlands test

site . DLO-Winard Staring Centre, 1994

Test of the Protot ype system for updating the CO R INE La nd Cover data base: The Anda lucia test

site . Agencia de Medio Ambiente - Junta de Andalucia, 1994

Test of the Protot ype system for updating the CO R INE La nd Cover data base: The Luxembourg test

site . G2

ERE , 1994

Test of the Protot ype system for updating the CO R INE La nd Cover data base: The Portugal test site .

CNIG, 1994

Technical assistance in the use of the Prototype computer system for updating the CO R INE L and

Cover d ata -base . R SD E, 1994 (contract n. 10108-94-03 F1E I ISP I)

A ll exampl es i n thi s guide were produced using Co-Pi l ot system devel oped by RSD E under JRC

specifications (contract 3952 / D JM /D JM ). Co-Pil ot is based on Carha for Wi ndows sw.

A dditional inform ation can be found in CORINE L and Cover Technical Gui de ISBN 92-826-2578-8ECSC - EE C - EA EC, Brussels. L uxembourg, 1993.

iv


6/132

v

PREFACE

Within the frame of JR C support to the European Environment Agency (EE A), the AIS U nit of Space

Applications Institute (SAI) o f the Joint R esearch C entre, is a part ner and co-leader of the E uropean Topic

Cent re on La nd C over, responsible for the Task R esearch and D evelopment of new Applications.

This Technical and M ethodological G uide for Updating the C OR INE Land Cover D ata B ase was produced

by AIS, as one of its contribution to this Task regarding the updat ing of CL C d ata base. This is a joint

publication of the SAI and the E EA .

The content o f this G uide is the result of a long experience, first w ith the creation o f the first generation

CO R INE Land C over data ba se (CL C), later on with the specific problems linked to its updating, in a context

of geo-referenced data base and integrat ed geographic information systems in general.

What ma kes the updating of C LC a specific problem, different from the ones G IS experts are used to fa ce is

the combination of the four groups of features:

1. the nature of the CLC data base: land cover classes, which boundaries are often fuzzy, with

different dyna mic rates betw een classes and inside each class;

2. differences in methodology between the creation and updating as a result of the technological

trend;

3. the use of data a nd information derived from different type of sources, in which earth

observation data plays an important role;

4. the European wide dimension and the homogeneity requested.

As a spatial dat abase, CL C is vulnerable to a common misunderstanding about his precision compared to

other G IS processing output. In conventional map ana lysis, precision is usually adapted to accuracy of the

data , while the precision of G IS processing exceeds quite often t he accuracy of the data used. The ability to

change scale and combine data from various sources and scales in a G IS ma y hide that precision is not

always ada pted to the accuracy. Still, there are no adequate means to describe the accuracy of complex

spatial units.Even being the present methodology defined for updating the C LC data base, its core concept is also va lid

for the creation or use of any land cover data bases in which earth observation data a nd G IS are involved.

ACKNOWLEDGEMENT

We wish to acknow ledge the four nationa l CL C teams involved in the test of the proto type computer system

and of the methodology developed for their practical contribute and exchange of ideas.

We are grateful to A. Arozarena, A. Sebastian, L. B ontemps, Neil Hubbard, for the time they invested

improving this G uide with interesting comments and careful reading, in particular C hris Steenmans for his

support and exchange of experience on the CLC database all over Europe.

Thanks to Yves Heymann with who many and fruitful discussions on how to conceive the updating of CLC

were taking in the past and for several year


7/132

vi

This guide had been designed as a working tool for those involved in updating the CORINE land cover

database.

In o rder to justify and clarify t he choices for the upda ting methodology, some chapters discuss the constraints

of the creation methodology.

An o verview o f the possible evolution is given together w ith typical examples of updating.

The guide is organized in the following sections:

=> C hapter 1 shortly introduces the principles of updat ing and the aim of this guide.

=> C hapter 2 provides an overview of the principles for the creation method ology

=> C hapter 3 illustrates the basic principles of the updating methodology.

=> Cha pter 4 further describes the updating methodology.

=> Cha pter 5 gives some information about validation and quality assurance at the end of updating.

=> C hapter 6 discusses how to define the optimum updating frequency.

=> Cha pter 7 contains a q uick-guide for the updating methodology.

=

> C hapter 8 illustrates possible future improvements of the methodology.

=> C hapter 9 includes some useful references and bibliography.

=> Annex 1 contains several examples (with changes to be updated or not).

=> Annex 2 shows the basic software needs for updating.

=> Annex 3 describes Co-Pilot (the JR C prototype system for updating).

In this guide the following symbols were used:

Name Symbol Description

D efinition This is the complete description of a particular item

Method This is the way to do a particular action

Postulate This is a rule that must be strictly followed


8/132

vii

CONTENTS

1. INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2. OVERVIEW OF CREATION METHODOLOGY PRINCIPLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1 CO R INE Land C over nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.2 Definition of spatial unit.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.3 Scale of wo rk for creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.4 Process of image-interpretation for creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.5 Use of Ima ge P rocessing systems in creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.6 D ata base structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.7 U se of ancillary da ta in creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3. BASIC PRINCIPLES FOR UPDATING. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.1 Definitions of error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.1.1 Errors caused by the specific material used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.1.2 Errors during da ta integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.1.3 Errors caused by the image-interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.1.4 Errors occurred during digitisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.1.5 E rrors during data t ransformat ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3.2 Definition of change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3.3 Possible sources of a ncillary d ata for upda ting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.3.1 Exa mple of implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4. UPDATING METHODOLOGY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.1 G eneral overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.1.1 Specific problems related to the subdivision of the work using regional teams . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.2 Work orga nisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4.2.1 Work Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294.2.1.1 Staff selection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.2.1.2 Train ing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.2.1.3 Cri teri a for work subd ivi sion f or several staff. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.2.2 D ata collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4.2.3 D ata standard isation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4.2.4 Da ta correction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.2.5 Da ta organisation (sheets and zo nes) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.2.6 D ata cata loguing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.2.6.1 Data subdi vision ( sheets and zones). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.3 Preliminary opera tions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.3.1 C hecking the geometric qua lity of the o riginal da ta (data base and ima ges). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.3.2 G eometric correction of the original dat aba se . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.4 Cha nges detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4.4.1 Image interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4.4.2 Scales of work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

4.4.3 How to detect changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

4.4.3.1 Manual interpr etation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

4.4.3.2 Contr ibu tion of computer procedur es for automatic change detection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

4.4.4 How to update geometric changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.4.4.1 Polygon O ri ented M ethodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.4.4.1.1 CRE ATE A NEW OB JE CT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4.4.4.1.1.1 Example of Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.4.4.1.2 MOD IFYING B OU NDAR IE S OF AN OB JE CT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56


4.4.4.1.3 D E LE TING AN OB JE CT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614.4.4.1.3.1 Example of Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62


9/132

viii

4.4.4.1.4 SPLITTING AN OB JE CT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65


4.4.4.1.5 ME R G E A SE T OF O B JE CTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66


4.4.4.2 Spaghetti O ri ented Appr oach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

4.4.5 How to update thematic changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

4.4.5.1 Examp le of Code Checking implementation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

724.4.6 Special B uttons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

4.4.6.1 Example of Buttons implementation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

4.4.7 Polygon's Marking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

4.5 Work management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

4.5.1 U se of Ma rks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

4.5.1.1 Example of M ark status moni toring implementation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

4.5.2 Sto ring informa tion a bout shee ts/zones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

4.5.3 Storing informat ion ab out t he project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

4.5.4 Integration and co-ordination of work of different teams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

4.6 D ata base management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

4.6.1 Maintenance of several databases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

4.6.2 Maintenance of one database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

4.6.3 Da tab ase structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

83

5. VALIDATION AND QUALITY ASSURANCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

6. UPDATING FREQUENCY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

6.1 Upda ting cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

7. QUICK GUIDE FOR UPDATING. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

7.1 Standards to be used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

8. IMPROVING THE METHODOLOGY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

9. REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

10. ANNEX 1: EXAMPLES OF UPDATING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

11. ANNEX 2: SOFTWARE NEEDS FOR UPDATING. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

12. ANNEX 3: CO-PILOT SYSTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121


10/132

ix

List of Figures

Figure 1 - Ima ge-interpretation schema for crea tion ............................................................................................................................................................................................... 6

Figure 2 - Polygons o n sheet bo undary .............................................................................................................................................................................................................................................. 9

Figure 3 - Small units located on sheet bound ary .......................................................................................................................................................................................................... 9

Figure 4 - B ad geo metry of the ma terial ..................................................................................................................................................................................................................................... 12

Figure 5 - Minimum cartographic unit ............................................................................................................................................................................................................................................ 13

Figure 6 - E rrors in polygon locat ion ................................................................................................................................................................................................................................................ 14

Figure 7 - E rrors on geometry caused by t he image-interpreter ..................................................................................................................................................... 15

Figure 8 - Errors during data transformation .................................................................................................................................................................................................................... 16

Figure 9 - Small units ...................................................................................................................................................................................................................................................................................................... 17

Figure 10 - C hanges provo king object creation .............................................................................................................................................................................................................. 17

Figure 11 - B ounda ry changes ........................................................................................................................................................................................................................................................................ 17

Figure 12 - Linear changes ................................................................................................................................................................................................................................................................................... 18

Figure 13 - Sum of sma ll changes to be considered .............................................................................................................................................................................................. 19

Figure 14 - Sum of small changes not to b e considered .................................................................................................................................................................................. 19

Figure 15 - Land Cover data overlapped to 1985 and 1991 TM images (original video scale 1:100,000) ............. 20

Figure 16 - Land Cover data overlapped to 1985 and 1991 TM images (in red a segment of 150 m) ......................... 20

Figure 17 - R aster ma p and satellite image in a multiwindows environment .......................................................................................................... 21

Figure 18 - Aeria l photos referenced b y mea ns of their flight pla n ........................................................................................................................................... 22

Figure 19 - C onsultation o f a photo collected o n the ground ............................................................................................................................................................... 24

Figure 20 - C onsultation o f a n aeria l photo .......................................................................................................................................................................................................................... 25

Figure 21 - Merge two regions ....................................................................................................................................................................................................................................................................... 28

Figure 22 - Work P lann ing .................................................................................................................................................................................................................................................................................... 30

Figure 23 - Splitting the work in zones........................................................................................................................................................................................................................................... 32

Figure 24 - Z ones and Sheets for w ork subdivision.............................................................................................................................................................................................. 34

Figure 25 - How to verify the geometric quality of old data .................................................................................................................................................................. 35

Figure 26 - Example of points to be selected to verify the geometric quality of the old database ................................... 36

Figure 27 - H ow to recognise similar points (based o n polygons intersections)................................................................................................

38Figure 28 - H ow to recognise similar points (based o n part icular shapes) ................................................................................................................... 38

Figure 29 - G eometric correction of the old d at aba se ...................................................................................................................................................................................... 39

Figure 30 - Old a nd new images and t he dat aba se displayed in tw o different w indows a t 1:100,000 scale....................... 40

Figure 31 - Old and new ima ges and D B displayed at 1:50,000 scale..................................................................................................................................... 41

Figure 32 - C OR INE unit with not homogeno us spectral a spect (311) ............................................................................................................................. 42

Figure 33 - Multiscale multiwindows analysis.................................................................................................................................................................................................................... 43

Figure 34 - D ifferent t ypes of stret ch ................................................................................................................................................................................................................................................ 44

Figure 35 - Contra st stretching...................................................................................................................................................................................................................................................................... 45

Figure 36 - La nd cover change with a raster classificat ion ........................................................................................................................................................................ 48

Figure 37 - Post classification of a raster cla ssification ..................................................................................................................................................................................... 49

Figure 38 - C hange t he bounda ries of a polygo n ......................................................................................................................................................................................................... 50

Figure 39 - Polygon Oriented methodology work flow.................................................................................................................................................................................... 51

Figure 40 - C reation of a new o bject.................................................................................................................................................................................................................................................. 52

Figure 41 - Mod ification of ob jects intersected ............................................................................................................................................................................................................ 52

Figure 42 - Small polygons genera ted b y the new object creation .............................................................................................................................................. 53

Figure 43 - Polygons (left) or Spaghet ti (right) oriented a pproach? ........................................................................................................................................ 68

Figure 44 - Arc-Nodes versus Spaghetti approach .................................................................................................................................................................................................... 69

Figure 45 - D ifferent phases of Spaghet ti oriented a pproach ............................................................................................................................................................... 69

Figure 46 - Spaghet ti O riented a pproach (lines updating) ........................................................................................................................................................................ 70

Figure 47 - Spaghet ti O riented a pproach (lines re-digitisation) ...................................................................................................................................................... 71

Figure 48 - Management of teams for updating ............................................................................................................................................................................................................ 79

Figure 49 - D ata base management: suggested approach for the original data base .................................................................................... 80

Figure 50 - D at aba se to be mainta ined for evolution ana lysis............................................................................................................................................................ 81Figure 51 - Maintenance of various databases ................................................................................................................................................................................................................. 81


11/132

x

Figure 52 - D ata base maintena nce, sheets oriented ............................................................................................................................................................................................... 82

Figure 53 - D ata base to be maintained for C OR INE project.............................................................................................................................................................. 82

Figure 54 - D ata base a spect for multiple partia l updating processes ..................................................................................................................................... 83

Figure 55 - Validation plan ................................................................................................................................................................................................................................................................................... 85

Figure 56 - Frequency of upda ting and observable changes ................................................................................................................................................................... 88

Figure 57 - D ifficulty of interpreta tion related to U pdat ing period ......................................................................................................................................... 88

Figure 58 - Co st depends on the upda ting freq uency .......................................................................................................................................................................................... 90

List of Definitions

D efinition 1 - Spatia l unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

D efinition 2 - Scale of w ork for creat ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

D efinition 3 - Ancillary da ta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

D efinition 4 - Systemat ic error. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

D efinition 5 - E rror on geometric location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

D efinition 6 - Cha nge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

D efinition 7 - Land Cover C hange Indicator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88D efinition 8 - R elevant changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

List of Postulates

Postulate 1 - Smaller units ..................................................................................................................................................................................................................................................................................... 16

Postulate 2 - Creation of a polygon ...................................................................................................................................................................................................................................................... 17

Postulate 3 - B ounda ry change definition ................................................................................................................................................................................................................................. 17

Postulate 4 - Set o f linear changes less than 100 m with a t ota l area great er than 25 ha .................................................................... 18

Postulate 5 - U se of a ncillary da ta in non-numeric form .............................................................................................................................................................................. 22

Postulate 6 - Coding for fragmented territory..................................................................................................................................................................................................................

28Postulate 7 - Training for fragment ed territory ............................................................................................................................................................................................................... 28

Postulate 8 - Material necessary for updating ................................................................................................................................................................................................................... 29

Postulate 9 - Old satellite images.............................................................................................................................................................................................................................................................. 32

Postulate 10 - Format for dat a import ........................................................................................................................................................................................................................................... 32

Postulate 11 - G eometric correction of old images .................................................................................................................................................................................................. 35

Postulate 12 - G eometric correction of the old data base ............................................................................................................................................................................. 36

Postulate 13 - Use of new and old image for updating during image-interpretation ................................................................................. 42

Postulate 14 - Scales of w ork ............................................................................................................................................................................................................................................................................ 46

Postulate 15 - Use of computer procedures for automatic change detection .......................................................................................................... 50

Postulate 16 - Accidenta l generation of o bjects smaller tha n 25 ha .......................................................................................................................................... 53

Postulate 17 - Data Entry of CORINE codes ................................................................................................................................................................................................................... 72

Postulate 18 - Polygon Ma rking .................................................................................................................................................................................................................................................................. 76

Postulate 19 - Maintain information about data ........................................................................................................................................................................................................... 78

List of Methods

Method 1 - How to subdivide the work........................................................................................................................................................................................................................................ 31

Method 2 - How to verify the geometric quality of the old images ........................................................................................................................................... 36

Method 3 - Ho w to verify the geometric quality of the old da taba se.................................................................................................................................... 36

Method 4 - Ho w to correct the geometry of the old data base............................................................................................................................................................ 37

Method 5 - H ow to use image processing functions .............................................................................................................................................................................................. 44

Method 6 - How to use dynamic contrast stretch ....................................................................................................................................................................................................... 46

Method 7 - How to detect geometric changes .................................................................................................................................................................................................................. 47Method 8 - Check the status of the work at polygons level ..................................................................................................................................................................... 76


12/132

xi

List of Tables

Table 1 - CO R INE Land Cover Nomenclature................................................................................................................................................................................................................. 5

Table 2 - Ancillary da ta for upda ting ................................................................................................................................................................................................................................................ 23

Table 3 - Distribution of the da ta ............................................................................................................................................................................................................................................................. 34

Tab le 4 - Ma trix of possible event s........................................................................................................................................................................................................................................................ 47

Tab le 5 - Ma trix of po ssible transitions ......................................................................................................................................................................................................................................... 47

Tab le 6 - Ma rking Fla g................................................................................................................................................................................................................................................................................................. 76

Table 7 - Current attributes of G ISC O C OR INE La nd Cover data base........................................................................................................................ 83

Table 8 - Attributes to be used in the CO R INE Land Cover da taba se, during updating ................................................................. 84


13/132

CO R INE La nd Co ver Upda ting - Technical and Methodological G uide

1. INTRODUCTION

This Technical a nd Methodological G uide for updating the C OR INE Land Cover (C LC ) dat aba se is based

upon the results of a methodological study concerning the updating of a rea l Land C over dat a set. The study

was performed as the J R Cs technical support to the C OR INE Land Cover project under the responsibility

of the Task Force for the E uropean E nvironmental Agency (D G XI ).

To implement the defined methodology, a protot ype computer system wa s developed (annex 3) and tested

by four national land cover teams. For each of the four teams, the tests consisted of both a n updating, and a lso

a retrospective detection of changes in land cover (known a s do wndating ). The tests used a set of the land

cover databa se corresponding to a quarter of TM scene, and followed the proposed methodology using the

dedicated software too l. The final evaluation of these tests was d one within the framework of t he R esearch

and D evelopment Task of the E uropean Topic Centre on La nd Co ver, under the responsibility of the AI S

unit. The suggestions and improvements coming from the general evaluation w ere alread y ta ken into account

in producing the guide.

The updat ing of the la nd cover da ta base must follow three ba sic principles:

1. It must take into account the methodology and data sources of the creation phase,

2. It should continue to lead to an harmonised data base at national and E uropean level,

3. The process of updating should be faster and more cost-effective than t he process of the original creation

of the database.

These principles have as a consequence the need for defining a methodology for updating that can be

consistently repeat ed in time in all countries, and by peo ple that ma y not ha ve been involved in the creation

phase. The procedure should be established in all its steps, with defined sta nda rds and t he subjectivity related

to interpretation procedure should be reduced.

With this scope, the present methodology was developed, including the use of a n Integrated G eographic

Informa tion System in all phases of wo rk, obliging the interpreter to fo llow the defined rules whilst providing

support to respect standa rds. Several conversion steps linked to different pha ses of wo rk are now red uced to

a single activity, with the consequent reduction of error. In order to fulfil the requirements of the CLC

methodo logy as close as possible, and to keep the homogeneity amo ng the Europea n countries, a customised

system with tailored functions is favoured. Such a system has been realised by the JR C, called C o-Pilot

(CO R INE Photo-Interpretation La nd cover Oriented Tool) (see Annex 3). D ifferent systems can also be

used but the availability of specific tools can speed up the updating process and facilitate the harmonisation.

The best technical solution when upda ting is a compromise between cost efficiency a nd the need of precision

for a specific product. The fact tha t this data base is primarily based on sat ellite da ta should be kept in mind.

This limits the precision that can be obta ined, and o ne temptation to avoid is that of over-updating. The

consequence can be increasing costs to o bta in wrong results or a slight improvement which may even be lost

when the data is passing through the generalisation procedure for the European database.

The updating can be the occasion to introduce improvements necessary for national applications of the

data base. Ho wever it should be kept in mind that the final product will be also the European da taba se.

D ue to this continental scale coverage, the geometry is a specific and complex aspect of the C LC data base. In

fact, it is a geo-referenced data base which uses for its creation and updating data provided by different

sources, each of them associated to a different geometric accuracy and which are often time dependent. This

is one of the ma in reasons to give in this G uide an emphasis to the geometric aspect of the da tab ase.

A correct identification of a cha nge in land cover a nd consequent upda ting is dea ling not only with a cost but

also with homogeneous and comparable statistics of changes at European level.

Some changes in land cover that can be easily detected with satellite data may not correspond to changes in

CL C classes. For example, if oil seeds replace winter cereals, then this is still the same a griculture class. Thismakes it difficult to use automa tic procedures for upda ting, and justifies the proposed methodology a s the

1


14/132

CO R INE Land Cover U pdating - Technical and M ethodological G uide

only one operational at present. Eventual technological improvements (hard and softwa re) may be

considered in the future, when they ha ve proved to be operational.

La nd cover mapping units are defined in terms of spatial, thematic and t emporal components; the associated

error has the three corresponding dimensions. Error is used here in its widest sense to include not only

mistakes but also the statistical concept of error meaning bias.

As a land cover databa se, the thematic accuracy could be the most important component. It is linked to the

process of computer aided image interpretation which is complemented by the use of exogenous data and to

the definition of the ma pped classes. It mea ns that, for classifying a mapping unit, the interpreter can use all

available sources of da ta; providing they are reliable, the interpreter skilled a nd with a good reference level,

those errors must be minimal a nd the thema tic accuracy different according to the nomenclature class. The

available validation at Co untry level confirms this statement. Important sources of errors will be: (i)

differences in difficulty of interpretation for the same class according to the characteristics of each Country,

(ii) when a ssembling national d ata bases, to guarantee tha t the mapping units along the frontiers are getting

the same code in all countries. Themat ic accuracy is also not spatially uniform (e.g. relief condition, land cover

complexity, rad iometric effects of the images); it is time dependent (e.g. recent cut forest at the satellite image

date and all classes dynamic enough in time to change category) and it depends to a certain extent on the

spatial a ccuracy.

Spatial accuracy has proved to be the most critical component of error in the CL C data base. It ha s two main

aspects: (i) locational accuracy related to the geometric quality of the source da ta (satellite images and

topographic maps used to correct them), (ii) positional accuracy of polygon boundaries, related with the

delineation of cartographic units.

Temporal accuracy is not eq ually relevant for all mapping classes. For the image da ta , a land cover class or

some cartogra phic units belonging to a certain class may not b e identifiable. Ano ther type of tempora l error

is due to the tempora l dynamic of a land cover class. A ma pping unit correctly classified can be alread y

occupied by a d ifferent land cover class when the validation is carried out. In ot her words, what is the optimal

period for updat ing land cover classes in a specific geographic region ? Temporal a ccuracy also interacts w ith

spatial and thematic accuracy.

The first updat ing is the most critical phase beca use in most countries it will be in the interfa ce betw een the

data base obta ined by a different method and the one proposed for the updating. This will be not the case for

successive updating events. Nevertheless, updat ing must not be conf used with validat ion. The validat ion must

be performed at the end of each updating. Also, for the first updating validation must be performed also

before upda ting begins. This is because of the changing methodo logies (from creation t o updat ing) as well as

the several transformation steps between original mapping and European d ata base compilation, each step

producing and propa gat ing errors. The methodo logy described in this guide is independent from scale.

Ho wever, the ado pted tolerances were established for the updating of a corrected da taba se set up at the scale

of 1: 100 000.

The updating of the C LC data base can be linked to national applications and consequent updating of other

data bases. Therefore, the cost a nd frequency of the updating at the European level can vary a ccording to

national programmes and the possibilities of sharing da ta, staff, etc. Specific European projects like Lacoast(assessment of land cover changes in the European coastal zone) can contribute to the updating of certain

zones. It leads to an a pproach of spatial updating.A n alternative approach is the thematic updating,w hen only

selected classes must be considered for updating.This can be relevant w hen, for example, an E uropean U rban

project is launched, or when a ny land cover class at a certain moment is relat ed to a relevant po licy issue.

Anyhow, it is most probable that an European CL C da taba se will always have discontinuity in time. Ho w to

mana ge the time in such a da ta base is still a to pic for research. It is difficult to define the optimal freq uency

and precise cost of upda ting because these are strongly dependent from the land cover evolution rat e

(determined by the nature a nd mans a ctivities). A real cost evaluation should be done before starting the

updat ing, based on the relevant characteristics of each site.

This Methodo logical and Technical G uide has been published when the Technical G uide Vol. 2 and the

technical guide on qua lity assurance and contro l are in progress. As a consequence, some chapters or items

like validation, quality control, the new CLC generalisation rules, metadat a are not defined in detail. Furtherinformat ion will be found in these publications.

2


15/132

CO R INE Land Cover U pdating - Technical and M ethodological G uide 3

2. OVERVIEW OF CREATION METHODOLOGY

PRINCIPLES

The aim of this chapter is to shortly illustrate the principles of the creation methodology which can be

considered important for their consequences on the successive updating.

The term creation methodology includes all the activities and data related to the creation of the data base.

The following topics of the creation methodology will be illustrated:

no me ncla t ure,

def init ion of spatial unit,

sca le of wo rk,

process of image-interpretation,

heterogeneity of the satelli te data used (MSS, TM, SPOT, other ),

use of Image Processing systems in creation,

use o f ancilla r y da ta .

It must be underlined that some principles for the creation of the C OR INE La nd Co ver datab ase have been

now revised. This chapter is mainly related to the definitions included in the first technical guide and

applicable to the first countries involved in the project.

For the countries in which the new a pproach was a dopted (use of integrated G IS system directly in creation

phase) only a pa rt of the following considerations must be considered relevant.

2.1 CORINE Land Cover nom enclat ure

The CORINE Land Cover nomenclature is a physical and physiognomic land cover nomenclaturecomprising three levels (illustrated in table 1).

The nomenclature is strongly related to the process of image interpretation, the w orking and publishing scale

and the smallest carto graphic unit.

In fa ct, it is easy to find, in the 3rd level of the nomenclature, a group of classes for which a deductive analysis

is required.

For example, the classes 1.4.2(Spor t and leisure facil iti es), 1.4.1(Green ur ban areas) and 1.2.3(Sea por ts) arenot just categories of different land cover, but are classifiable according to their different land use.

This implies that the image-interpreter chooses the class to be updat ed, taking into a ccount the context in

which the unit is placed (i.e. a green spa ce in a tow n must proba bly be class 1.4.1 or class 1.4.2) and the

add itional information ava ilable (photos, cartography, ...).

An o ther example has to do w ith non homogeneous classes, like:

2.1.1 (N on ir rigated arable land)2.4.2 (Complex cultivation patterns)2.4.3 (L and pri ncipally occupied by agricultur e, with signifi cant areas of natural vegetation)

...........

In this case the aggregat ion of primitive objects may be a subjective process based o n specific pattern.

In ot her terms the first consideration to be made regarding the nomenclature a nd the use of satellite da ta isthe following one: classes in CO R INE are not related only to an homogeneous spectral response !


16/132

CO R INE La nd Co ver Upda ting - Technical and Methodological G uide4

As a consequence of this assertion, automa tic updating based on supervised o r unsupervised classification,

neural networks, fuzzy sets, ... of remotely sensed data cannot be considered exhaustive for all possible

classes.

As any land cover mapping CORINE Land Cover creation methodology strongly depends on image-

interpreter capability.

D ifferent image-interpreters can sometimes delinea te different complex carto graphic units when unit is a

collection of elementary land cover entities. In t his case the w ay to collect these objects can va ry a ccording

to the styles and experience of the various image-interpreters.

Some land cover boundaries are fuzzy on the ground. D ifferences between land cover classes may o ccur

through a gradua l transition.

In particular, the classes H eterogeneous agricul tur al areas: 2.4.1 / 2.4.2 / 2.4.3 / 2.4.4 and the classesScrub and/or herbaceous vegetation associati ons: 3.2.1 / 3.2.2 / 3.2.3 / 3.2.4 are easily subjected todifferent assignments.

A synoptic view is indispensable to identify complex units.

2.2 Def ini t ion of spat ial unit

D efi nition 1 - Spatial unit

The spatial unit corresponds both to a n area of homogeneous cover (wat er, forest,...) and to a n

aggregation of small homogeneous areas that represent a land cover structure.

The following constraints must be respected: it represents a significant surface in relation to the work scale,

it is well distinguishable from other surrounding units,

it is suff icient ly stab le in t ime.

The minimum cartographic unit for CORINE Land Cover at 1:100,000 scale is 25 ha.

2.3 Scale of work for creat ion

D efi nition 2 - Scale of work for creation

In the o riginal method ology t he image-interpretatio n wa s done on 1:100,000 image printout s in

which a transparent film was overlapped to the images and digitised at the end of the image-

interpretation process.

In the revised methodology (ad opting a G IS a pproach) the scale (and as consequence the

precision) can be different .


17/132

CO R INE Land Cover U pdating - Technical and M ethodological G uide 5

Table 1 - CORI NE L and Cover N omenclature

Level 1 Level 2 Level 3

1. Artificial surfaces 1.1. Urban fabric 1.1.1. Continuous urban fabric

1.1.2. D iscontinuous urban fa bric

1.2. Industrial, commercial 1.2.1. Industrial or commercial unitsand tra nsport units 1.2.2. Ro ad and ra il networks and associated land

1.2.3. Sea ports

1.2.4.A irport

1.3. Mines, dumps and construction 1.3.1. Mineral extraction sitesites 1.3.2. D ump

1.3.3. Construction site

1.4.A rtificial non-agricultural 1.4.1. G reen urban areasvegetated a reas 1.4.2. Sport and leisure facilities

2. A g ricu lt ur al a re as 2. 1. A r a ble la nd 2. 1. 1. N on ir rig at ed a ra b le la n d

2.1.2. Permanent ly irrigated land

2.1.3. R ice fields

2.2. Perma nent crops 2.2.1.Vineyards

2.2.2. Fruit t rees and berries plantations

2.2.3. Olives groves

2.3. Pa stures 2.3.1. Pa stures

2.4.H eterogeneous agricultural areas 2.4.1.A nnual crops associated with permanent crops

2.4.2. Complex cultivation pat terns

2.4.3. La nd principally occupied by a griculture, withsignificant areas of na tural vegetation

2.4.4.A gro-forestries area s

3. Forest and semi 3.1. Forests 3.1.1. B road leaved-forestnatural areas 3.1.2. Coniferous forest

3.1.3. Mixed forest

3.2. Scrub and/or herbaceous 3.2.1. Natural grasslandvegetation a ssociations 3.2.2. Moors and heathlands

3.2.3. Sclerophylous vegetation

3.2.4.Transitio nal wo odla nd-scrub3.3. Open spaces with little 3.3.1. B eaches, dunes, sandor no vegetation 3.3.2. B are rocks

3.3.3. Sparsely vegetated areas

3.3.4. B urnt areas

3.3.5. G lacier and permanent snow -fields

4. Wetlands 4.1. I nland wetlands 4.1.1. I nland marshes

4.1.2. Pea t bogs

4.2. Coastal wetlands 4.2.1. Sa lt marshes

4.2.2. Salines

4.2.3. Intertidal flats

5. Wa ter bodies 5.1. C o ntinenta l w aters 5.1.1. S trea m courses

5.1.2.Water bod ies

5.2. Marine wa ters 5.2.1. Coasta l lagoons

5.2.2. Estuaries

5.2.3. Sea and ocean


18/132


2.4 Process of im age- int erpretat ion for creation

After collection of the necessary data, CO R INE Land Cover creation methodology suggests to ad opt the

scheme illustrated in figure 1 ( fr om CORINE L and Cover Technical G uide - page 50).

Figure 1 - Image-interpretation schema for creation

Stratification of false colourimage 1:250,000

First delineation/identificationof false colour images 1:100,000

Study of aerial photographsstereoscopic pairs

Seconddelineation/identification of

false colour images 1:100,000

Additional processing of

satellite data

Third delineation/identificationof false colour images 1:100,000

Further study on aerialphotographs

Preparing and carrying outfield surveys

Quasi-definitive delineation/identification before checking

Study of ancillarydocumentation

Selection of aerialphotographs

Evaluation/location ofinterpretation problems

Evaluation/location of remaininginterpretation problems


19/132

CO R INE Land Cover U pdating - Technical and Method ological G uide 7

Add itional important stat ements reported from the CO R INE La nd Cover Technical guide (page 53)are:

Addi tional pr ocessing o f satell ite data at an interactive processing station can serve to:

compl ete the interpr etation,

corr obor ate the results obtained..

I n order to keep additi onal pr ocessing time to a minimum , the image-intepreter must prepare the wor k

careful ly at an interactive wor k station. This preparation entails:

identif ying on the false colour image those areas where additi onal p rocessing i s requi red,

mar k ing the areas to be checked and the areas where the interp retation has to be completed.

gathering the requi site ancil lar y documentation concern ing these areas.

planni ng the sequence processing stages accord ing to the prob lems to be solved.

The interpreting team must be very cautious about the time they spend at the interactive processing work

station. Not all problems remain ing at the end of the image-interpr etation p rocess (using the basic data) can

be solved thr ough the interactive processing of either basic or additi onal data.D epending on the type of sensor

selected for the data, the interpretation team wil l have to:

identif y some prob lems (topi cs) whi ch can be solved by interactive processing,

depending on area compl exity, establi sh the maximum time to be spent.

Nevertheless the image-interpreter can use the image processing system. In this case the guide states:

Addi tional processing can be carri ed out on two types of data:

the ori ginal data used to produce false-colour images,

additi onal satell ite data acquir ed where justifi ed.

The foll owing fi ve types of mul tispectral data processing:

adjustment of dynami cs,

vegetatio n index,

automati c classif ication,

pr incipal component analysis,

two-d imensional spatial fi lter,are recommended fo r an interactive wor k station because:

experience has shown them to be the best adapted to the pro ject obj ectives,

they are generally available on commercial interactive processing work stations.

R esuming the concepts expressed in the C OR INE La nd C over Technical guide: image processing must be

considered a useful tool to eliminate eventual doubts; the image-interpreter can work only on small areas

to detail its first identification.

The final geometric delineation was done using the original image maps or hardcopies of the screen (at

1:100,000 scale).

These choices were related to the status of the market of the image processing systems in 1985 when the

methodology wa s defined. As a consequence, the creation methodology wa s oriented to minimise the use ofvery expensive systems, favo uring the manual w ork on pa per supports.

Now d irect video digitisat ion capabilities offer better performa nce and time saving. The cost for purchase of

hardware a nd software ha s greatly decreased.

As a consequence although the old creation methodology principles maintain their validity from a logical

point of view, their original suggested implementat ion cannot currently be considered the best way t o make

economy and to speed up the process without loss of quality of the work.

In the past some constraints of the creation methodology were often ignored or not respected.

For example the following constraints (fr om CORI NE L and Cover Technical Guide - page 43):

T he transparencies are impor tant documents and must be produced with great care. Theaccuracy of the resulti ng database wil l depend also on the quality of the transparencies,


20/132


the geometry of the transparency is the same as that of the corr espondi ng 1:100,000 standard

topograph ic map,

to obtain a good r epresentati on of thi s geometry some simple ru les are suggested. To

guarantee the qual ity the image-in terpr eter must check :

* that map documents are properly aligned,

* the compl ex categories interpr etation,

* that adjacent maps link up,* that only one code has been used f or each area,

* that the areas that have been compl eted are marked off.

These rules are extremely important t o guara ntee a goo d q uality of the fina l data base. Ignoring these simple

suggestions may prod uce bad classifications or delineations of some units.

Because of the error propagation it may be difficult to know if all these operations have been correctly

performed. Normally the valida tion process can focus some errors and consequently can show a n incorrect

application o f the methodo logy taking into account tha t some errors are intrinsic of the methodology itself.

U pdating is so sensitive to the original dat aba se qua lity because a correct qua ntification of changes can be

performed only after errors removal.

2.5 Use of Image Processing systems in creation

The use of Ima ge P rocessing systems wa s often very low or completely a bsent in the creat ion phase.

The Image Processing systems were mainly used:

to help in interpretation of particular areas,

to produce additional materials to be used for interpretation,

to check the interpretation results by over-laying the geocoded sat ellite da ta with the la nd

cover vector data .

The first case (mainly used for w etlands), proved very useful to d etect differences in some other classes (i.e.broa d leaved forest and coniferous forest areas). This kind of solution that can be defined interactive

support to the interpretation by means of Image Processing system could be higher if Multitemporal

images were available or modi fying the methodologyif no image prints were produced (directly performing

the video d igitisation). These cases describe for example the approach fo r the creat ion in some regions of

Ita ly (Veneto, Toscana, ...).

In the second case the image-interpreter used the I mage P rocessing system for prod uction of slides or prints

(obtained with particular procedures like vegetation index, principal component analysis, linear

combination,...) and used t hese slides during the image-interpretat ion session. This situation typically

describes a image-interpreter who doesn't have his own Image Processing system and must collect all

suita ble materials before sta rting its interpretat ion session. This type of solution can be defined: bat ch use

of Ima ge Processing system .

The third case proved to b e very efficient for final verification of t he dat aba se.


21/132

2.6 Database struc ture

The present structure of C OR INE D B was influenced by the choice of the sheet oriented approach and

of the topological structure. In fact, all COR INE ob jects are polygons, but their storage is organised by

coded arcs and labelled points (one for each polygon).

This structure guarant ees the topology coherence because the overlapping of t wo polygons is impossible.

One aspect that must be considered (for its influence on Updating) is the problem of polygons situated on

the map limits. The CORINE Techni cal Gu ide establi shed (page 79): the features situated at the edge of the

map sheet must be matched up exactly i n the fi nal f il e.

This sta tement mea ns that a unique polygon is created merging the t wo original ones ( Figure 2 ).

Figure 2 - Polygons on sheet boundary

The figure 3 illustrates the presence of units < 25ha in a sheet oriented dat a structure, which would be

merged with neighbouring polygons resulting in land cover units >25ha.

Figure 3 - Smal l un its located on sheet boundary

Left Map Right Map Left Map Right Map Left Map Right Map

two adjacent polygons two adjacent polygons one polygon obtained bywithout geometric error with geometric error merging the 2 adjacent ones


Before Removing After Removing

Top M ap Top M ap

Bottom M ap Bottom M ap

25ha Units


22/132


2.7 Use of anci l lary dat a in creation

Anot her important a spect is concerned with the use of exogenous dat a (photos, maps, ..) that can influence

the qua lity of the dat aba se (for example according to the presence or absence of photos in the same area s).

The original creation metho dology req uired to store the history of each image-interpretat ion step (ancillary

data used, processing, ...) but sometimes there is not a precise track of the ancillary data used forinterpretation on the different areas of each country and different teams. Some of these materia ls were

made available for that purpose and could not be archived.

According to the C OR INE Land Cover Technical G uide a fact never to be lost sight of is that in the L and

Cover methodo logy, the satell ite data may constitute the fundamental database but the requir ed pr oject

inventor y output cannot be derived solely fr om them. The methodology is a data integration methodo logy,

wherein the coll ection and analysis of ancill ary data in conj unction wi th the satell ite spectral data provide a

convergence of evidence that leads to a reli able identifi cation of the particular land cover class occurr ing on a

given land uni t.

Definition 3 - A ncillary data

The term ancillary da ta refers to a ny documentary, cartogra phic or photographic information

concerning land cover which does not come directly from the sat ellite da ta base.

As a consequence of this definition a spatia l unit is att ributed to a class not only on the b asis of the satellite

imagery, but also through the additiona l data ava ilable for the image-interpreter. This means that in these

cases the satellite image is not enough for a satisfactory identification o f the class.

Such data essentially comprise: topographic maps, thematic maps related to land cover, statistical

information, aerial photographs.

The following list concerns some stat ements regarding type and use of the main importa nt da ta in order toevaluate their possible contribution (from CO RINE L and Cover Technical G uide - page 40-42).

Standard topographic maps are essential to the L and Cover pr oject.T hey are used at various stages:

to prepare the transparency overlays for the interpretation work , and thus establishing the

geometry of the interpr etation,

for geometric correction of satelli te data,

they constitute the reference document for controlli ng the geometry of the digitisation o f the

in terpr etation sheets,

they are a very important source of inf ormation on land cover.

Statistical in fo rmation provide a general perspective fo r the CORINE l and cover project:

they provi de a comprehensive picture of land cover,

they are a means of verify ing the L and Cover r esults.

The aeri al photographs play a major rol e in the L and Cover project.T hey are used:

to identify (as a nomenclature class) units delineated on the false-colour images which might be

incor rectly classif ied,

to determ ine the exact boundari es of un its which are not resolved clearly on the satell ite image,

to verif y and validate the results of the land cover mapping.

The consequences of these statements f or t he updating process are evident as discussed af ter.


23/132


3. BASIC PRINCIPLES FOR UPDATING

3.1 Definit ion s of error

We can distinguish 5 sources of error in the existing database:

errors caused by the material used,

errors during data integration,

errors caused by the image-interpretat ion,

errors occurred during digitisation,

errors during data transformation.

3.1.1 Errors caused by the specific material used

The error caused by the used material is common to all objects (polygons) of each sheet and it can so be

considered a s a systematic error.

D efi niti on 4 - Systematic err or

We define as systematic error every error affecting the database or the Images that can be

considered as common to the whole data set and can so be modeled and corrected with some

techniques of global correction. This type of error must be identified and corrected b efore a ny

eventual local error correction.

As described in the chapter geometric correction of the original data base it is possible sometimes to

proceed to a global correction of each Sheet data if the error can be classified of systematic type and if the

operator provided ancillary information necessary to its evaluation (for example co-ordinates of control

points acquired on the geographic reference and on the database to be corrected).

Figure 4 shows an example of ba d document (satellite image on pa per) used for image-interpretation a nd its

consequence on the interpreted polygons.

3.1.2 Errors during data integration

As often mentioned, ancillary data play a considerable role in the CO R INE land cover creation and

updating. It is highly recommended to use all complementary information during the interpretat ion.

As any kind of information, also ancillary data a re affected by random or systematic errors. As a

consequence, the ancillary data can provoke a diffusion of their own errors over the CO R INE data base.

In particular, the two t ypes of errors that can be originated by ba d ancillary data are on geometry and in

class identification.

For example, the use of distorted topogra phic maps has a s a consequence an incorrect geometric precision

or different levels of geometric precision according to the different scale of ava ilable cartogra phic maps.

Instead, the use of ba d thematic ancillary data can suggest to the image-interpreter a particular

interpretation that is not the correct one.


24/132


Correct image geometry

Bad image geometry

Figur e 4 - Bad geometry o f the materi al

3.1.3 Errors caused by the image-interpretation

The errors caused by the image-interpret er can be classified as:

bad code attribution for a class,

bad boundarys delimitation for a cartographic unit.

The error in the class identification, when recognised, is easily correctable.The G IS system used for updating

must conta in some special functions to auto matically check the errors caused by no code a ttribution (NU LL

or 0 code) and errors caused by the attribution of a not existing code (using appropriate Codes Look up

tables).

The errors of bad code attribution as a consequence of a bad interpretation or a data entry error (for

example code 311 instead of 312) cannot be a utomat ically detected !


25/132


The errors of geometry due to bad boundary delimitation can be caused by various factors and can be of

different relevance. For some land cover classes the boundary is fuzzy on the ground. It also happens that

a sharp boundary on the ground can appear fuzzy on the satellite data.

A q uantification of the error is necessary: overall it must be considered that in the CO R INE Land Cover

Method ology t he smallest unit mapped is defined as a surface of 25 ha tha t represents at 1:100,000 scale a 5

x 5 mm square or a circle with 2.8 mm radius (see figure 5).

Figure 5 - M inimum cartographic unit

The main a spect concerning geometric precision are given in the C OR INE La nd C over Technical G uide

(page 76):

on ly units wi th a minimum surf ace area are captured in the land cover project. The small est surf ace area

mapped is 25 ha.The only li near features covered are those exceeding 100 m i n width.

Not a ll the elements that can influence the geometric precision were o riginally defined in a standa rd wa y.

Two d ifferen t aspects must be considered con cerning geometric precision:

location,

surface.

The two aspects ar

corine tutorial

Documents