Spatial Statistics for Remote Sensing Remote Sensing and Digital Image Processing VOLUME 1 Series Editor: Freek van der Meer, International Institute for Aerospace Survey and Earth Sciences, ITC, Division of Geological Survey, Enschede, The Netherlands and Department ofAppliedEarth Sciences, Delft University ofTechnology, The Netherlands. Editorial Advisory Board: Michael Abrams, NASA Jet Propulsion Laboratoy, Pasadena, CA, U.S.A. Paul Curran, University of Southampton, Department of Geography, Southampton, U.K. Arnold Dekker, CSIRO, Land and Water Division, Canberra, Australia Steven de Jong, Utrecht University, Faculty of Geographical Sciences, Department of Physical Geography, The Netherlands. Michael Schaepman, ETH, Zurich, Switzerland SPATIAL STATISTICS FOR REMOTE SENSING edited by ALFRED STEIN FREEK VAN DER MEER and BEN GORTE International Institute for Aerospace Survey and Earth Sciences, ITC, Enschede, The Netherlands KLUWER ACADEMIC PUBLISHERS NEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW eBookISBN: 0-306-47647-9 Print ISBN: 0-7923-5978-X ©2002 Kluwer Academic Publishers NewYork, Boston, Dordrecht, London, Moscow Print ©1999,2002 Kluwer Academic Publishers Dordrecht All rights reserved No part of this eBook maybe reproducedor transmitted inanyform or byanymeans,electronic, mechanical, recording, or otherwise, without written consent from the Publisher Created in the United States of America Visit Kluwer Online at: http://kluweronline.com and Kluwer's eBookstore at: http://ebooks.kluweronline.com Preface This book is a collection of papers on spatial statistics for remote sensing. The book emerges from a study day that was organized in 1996 at the International Institute for Aerospace Survey and Earth Sciences, ITC, in Enschede, The Netherlands. It was by several means a memorable event. The beautiful new building, according to a design by the famous modern Dutch architect Max van Huet was just opened, and this workshop was the first to take place there. Of course, much went wrong during the workshop, in particular as the newest electronic equipment regularly failed. But the workshop attrackted more than hundred attendants, and was generally well received. The results of the workshop have been published in Stein et al. (1998). The aim of the workshop was to address issues of spatial statistics for remote sensing. The ITC has a long history on collecting and analyzing satellite and other remote sensing data, but its involvement into spatial statistics is of a more recent date. Uncertainties in remote sensing images and the large amounts of data in many spectral bands are now considered to be of such an impact that it requires a separate approach from a statistical point of view. To quote from the justification of the study day, we read: Modern communication means such as remote sensing require an advanced use of collected data. Satellites collect data with different resolution on different spectral bands. These data all have a spatial extension and are often related to each other. In addition, field data are collected to interpret and validate the satellite data, and both are stored and matched, using geographical information systems. Often, statistical inference is necessary, ranging from simple de- scriptive statistics to multivariate geostatistics. Classification, stat- ististical sampling schemes and spatial interpolation are important issues to deal with. Maximum likelihood and fuzzy classification are now used intermixedly. Careful attention must be given as to where and how to sample efficiently. Interpolation from points to areas of land deserves thorough attention to take into account the spatial variability and to match different resolutions at different scales. Finally, the data have to be interpreted for making import- ant environmental decisions. This book reflects the set-up of the study-day. It addresses issues on remote sensing, interpolation, modeling spatial variation, sampling, classification and de- cision support systems, to namejust a few. Several of the authors were also speakers v vi during the workshop, but some topics were at that day not addressed, and hence the set of authors has been enlarged to guarantee a broad coverage of aspects of spatial statistics for remote sensing purposes. The book would not have been possible without contributions from various people. First, we wish to thank the authors, who have all been working very hard to give the book the level as it has just now. Second, we like to thank the ITC man- agement for making book and workshop possible. We thank Dr. Elisabeth Kosters, Mr. Bert Riekerk and Mrs. Ceciel Wolters for their contributions at various stages. Finally, we like to thank Mrs. Petra van Steenbergen at Wolters Kluwer Academic Publishers, for her patience and her continuing assistance of giving the book the appearance it has by now. Alfred Stein Freek van der Meer Ben Gorte Enschede, April 1999 Contributors and editors Peter Atkinson Dept of Geography, Univ of Southampton, Highfield, Southampton SO17 1BJ, United Kingdom [email protected] Sytze de Bruin Wageningen University, Dept of GIRS, PO Box 33, 6700 AH Wageningen, The Netherlands [email protected] Paul Curran Dept of Geography, Univ of Southampton, Highfield, Southampton SO17 1BJ, United Kingdom [email protected] Jennifer Dungan Johnson Controls World Services Inc, NASA Ames Research Center, MS242-2, Moffet Field, CA 94035-1000. USA [email protected] Ben Gorte ITC, Division of Geoinformatics, PO Box 6, 7500 AA Enschede, The Neth- erlands [email protected] Jaap de Gruijter SC-DLO, PO Box 125, 6700 AC Wageningen, The Netherlands [email protected] Cees van Kemenade CWI, PO Box 04079, 1090 GB Amsterdam, The Netherlands [email protected] Freek van der Meer ITC, Division of Geological Survey, PO Box 6, 7500 AA Enschede, The Netherlands [email protected] Robert J Mokken University of Amsterdam, Sarphatistraat 143, 1018 GD Amsterdam, The Netherlands [email protected] vii viii Martien Molenaar ITC, Division of Geoinformatics, PO Box 6, 7500 AA Enschede, The Neth- erlands [email protected] Andreas Papritz ETH Zurich, Inst. Terrestrial Ecology, Soil Physics, Grabenstraße 11a, CH- 8952 Schlieren, Switzerland [email protected] Han laPoutré CWI, PO Box 04079, 1090 GB Amsterdam, The Netherlands [email protected] Ali Sharifi ITC, Division of Social Science, PO Box 6, 7500 AA Enschede, The Neth- erlands [email protected] Andrew Skidmore ITC, Division of Agriculture, Conservation and Environment, PO Box 6, 7500 AA Enschede, The Netherlands [email protected] Alfred Stein ITC, Division of Geoinformatics, PO Box 6, 7500 AA Enschede, The Neth- erlands [email protected] Contents Preface v Contributors and editors vii Introduction xv I 1 1 Description of the data used in this book — Ben Gorte 3 1.1 The Landsat Program 4 1.2 Imageradiometry 4 1.3 Image geometry 6 1.4 Study area 7 2 Some basic elements of statistics — Alfred Stein 9 2.1 Population vs. sample 10 2.2 Covarianceandcorrelation 15 2.2.1 Significance of correlation 17 2.2.2 Example 18 2.3 Likelihood 18 2.4 Regression and prediction 21 2.5 Estimation and Prediction 22 3 Physical principles of optical remote sensing — Freek van der Meer 27 3.1 Electromagnetic Radiation 27 3.2 Physics of Radiation and Interaction with Materials 28 3.3 Surface Scattering 29 3.4 Reflectance properties ofEarth surface materials 30 3.4.1 Minerals andRocks 30 3.4.2 Vegetation 34 3.4.3 Soils 35 3.4.4 Man-made and other materials 36 3.4.5 Spectral re-sampling 36 ix x 3.5 Atmospheric attenuation 37 3.6 Calibration of LANDSAT TM to at-sensor spectral radiance 38 4 Remote sensing and geographical information systems — Sytze de Bruin and Martien Molenaar 41 4.1 Data modeling 42 4.1.1 Geographic data models of real world phenomena 42 4.1.2 GIS data structures 42 4.2 Integrating GIS and remote sensing data 45 4.2.1 Geometric transformation 46 4.2.2 GIS data used in image processing and interpretation 47 4.2.3 GeographicinformationextractionfromRS imagery 49 4.3 Propagation of uncertainty through GIS operations 53 4.4 Software implementation 54 II 55 5 Spatial Statistics —Peter M. Atkinson 57 5.1 Spatial variation 57 5.1.1 Remote sensing 57 5.1.2 Spatial data and sampling 58 5.1.3 Remotely sensed data 59 5.1.4 Spatial variation and error 60 5.1.5 Classicalstatistics and geostatistics 61 5.2 Geostatistical background 61 5.2.1 Structure functions 62 5.2.2 The Random Function model 64 5.2.3 The sample variogram 66 5.2.4 Variogrammodels 68 5.3 Examples 71 5.3.1 Remotely sensed imagery 71 5.3.2 Sample variograms 74 5.3.3 Indicator variograms 77 5.3.4 Cross-variograms 79 5.4 Acknowledgments 81 6 Spatial prediction by linear kriging — Andreas Papritz and Alfred Stein 83 6.1 Linear Model 87 6.2 Single Variable Linear Prediction 90 6.2.1 Precision criteria 90 6.2.2 Simple kriging 90 6.2.3 Universalkriging 95 6.2.4 Ordinary Kriging 99