11. Land Cover Mapping and Accuracy Assessment

Unit 11 focuses on the subsequent processing of land cover data after a classification has been developed. It begins by looking at how classification accuracy can be measured through comparison with reference data. In this context, reference data are geospatial data concerning land cover that have been derived from a source that is independent of the imagery used to develop a land cover classification. Often, this comparison is made using a table known as a confusion matrix, and in 11.3, we look in detail at confusion matrices and the accuracy measures that can be derived from them.  In 11.3 we also look at how supervised classification using Envi through a practical exercise.  Note that this practical exercise is assessed and contributes towards your overall grade for this module – see BlackBoard for further details.

In 11.4 we turn our attention to the different land cover classifications that are available internationally as data products, comparing some of their key characteristics. We then look in 11.5 at the ways that we can measure land cover change – in other words, variation in land cover between two dates. There are of course other ways of analysing land cover data that we do not consider here. For example, the growing discipline of landscape ecology looks at how different land cover polygons may form a mosaic of habitat patches when thinking about the ecology of a particular species. A landscape ecologist might think about the size and shape of habitat patches, their inter-connectivity and the degree of land cover variability across the landscape.

Reading Activity

In preparation for these materials, we suggest you look at Chapters 13, 14 and 20 of Campbell and Wynne

J.P. Campbell and Wynne R.H (2011): Introduction to Remote Sensing. 5th edition. The Guilford Press, New York.