The MAP COMPARISON KIT:

Comparing maps is an important issue in environmental research. Reasons for comparing maps may be: (i) the different socio-economic scenarios on which they are based, (ii) detection of temporal changes, (iii) calibration/validation of land-use models, (iv) hotspot detection, (v) their use in uncertainty analysis, and (vi) their origin in different methodologies/models. This report addresses the problem of quantifying subsequent map similarities and dissimilarities. Our main focus is on maps denoted as ‘categorical’ or ‘nominal’. A number of the five mapcomparison techniques are described. These techniques differ in mathematical approach (no math, ‘cell by cell’, two types of ‘fuzzy’ and ‘single-map statistics’) and apply to different types of maps (nominal, ordinal, ratio and interval scale). Special attention is given to the comparison of maps through fuzzy-set calculation rules. The rationale is that fuzzy-set map comparison is very close to human judgement, as shown in an Internet experiment. The MAP COMPARISON KIT (MCK) software plays a major role in the report. MCK, a software package for ‘state-of-the-art’ map comparison, contains all the examples used in this report. The software, developed by order of the Netherlands Environmental Assessment Agency, was fully designed by the Research Institute for Knowledge Systems. The software will be made publicly available on the RIVM website early 2004 (www.rivm.nl). RIVM report 550002005 page 3 of 127 Acknowledgements A number of individuals and institutes have contributed to this report. First of all, let me thank the staff at RIKS, the Research Institute for Knowledge Systems. Inge Uljee and Alex Hagen did a marvellous job in designing and programming the MAP COMPARISON KIT software. In good cooperation with RIVM, they improved the various versions of the software needed to realize the final version. Alex Hagen also (co-)authored sections 2.3 and 2.5, and Appendix C on fuzzy-set map comparison versus human judgement. He was the initiator of section 4.1, the analysis of children’s maps. Many thanks to Guy Engelen for both his encouragement and his coordination of the software development at RIKS. A number of RIVM colleagues have made the following contributions: Ton de Nijs (RIM/MNP) as co-author of Chapter 1 and section 5.2, and lead author of section 5.1, on the calibration of the LUMOS Environment Explorer; Raymond de Niet (RIM/MNP) as lead author of section 5.2 on the consistency of land-use maps, and Kees Klein Goldewijk (KMD/MNP) as author of section 4.2 on the analysis of historical agricultural land use. Furthermore, Michel Bakkenes (NLB/MNP) provided the maps on heavy metals in topsoils (section 4.6); Laurens Zwakhals (VTV) the maps on ambulance driving times (section 4.5) and Wideke Boersma (RIM/MNP), a number of maps from the LUMOS Land Use Scanner (sections 4.3 and 4.4). Judith Borsboom-van Beurden (RIM/MNP) was co-author of sections 4.3 and 4.4. Kit Buurman (IMP/MNP) played on important role in the initial phase of the Map Comparison Project. I thank Geert Verspaij and Gert-Jan Stolwijk (both IMP/MNP) for programming and testing a prototype user shell of the MAP COMPARISON MODULE within ArcGis-8. Finally, Arthur Beusen and Arnold Dekkers (both IMP/MNP) designed software tools additional to the MAP COMPARISON KIT software. Peter Verburg (Wageningen UR) was the lead author of Appendix A on single-map statistics. The programming of enrichment factors was performed by Kor de Jong (University of Utrecht). He was also the co-author of Appendix A. The software development was funded by several project leaders at the RIVM, i.e.: Arthur Beusen (IMP/MNP), Harm van den Heiligenberg (NMD/MNP), Peter Janssen (IMP/MNP), and Marianne Kuypers-Linde (RIM/MNP).