Global Land Cover Map Validation Experiences: Toward the Characterization of Quantitative Uncertainty

Validation, as described in the CEOS-LVP report, is a whole process with various components, including quality control, qualitative assessment, cross-comparisons, confidence maps, and accuracy assessment. Recognizing that these components are truly a part of the validation process and are needed to get the whole picture, this chapter focuses particularly on the characterization of quantitative uncertainty. Whereas the overall validation process aims at reporting only on the fitness for use of a given product, and the accuracy assessment only on thematic accuracy, the characterization of uncertainty aims at documenting quantitatively the intrinsic product quality. On the basis of current experiences, this chapter seeks to contribute to further standardization of the characterization of uncertainty. As proposed in the Guide to the Expression of Uncertainty in Measurement (GUM) (JCGM, 2008), a readily implemented, easily understood, and generally accepted procedure is necessary for characterizing the quality of a result, that is, for evaluating and expressing its uncertainty. First, a set of eight uncertainty components for a land-cover map is proposed and described briefly. The first seven components are also illustrated and discussed in the light of the four validation exercises effectively completed. The eighth component is addressed in a separate section dedicated to the accuracy assessment process. The most recent experience-corresponding to the GlobCover accuracy assessment-is further detailed as a case study to report on the recent findings and raise current issues.

DEFOURNY Pierre;  MAYAUX Philippe;  HEROLD Martin;  BONTEMPS Sophie; 

2013-01-08

Taylor and Francis

JRC73563

978-1-4200-7074-3,   

https://publications.jrc.ec.europa.eu/repository/handle/JRC73563,