Validation of the Global High-Resolution Globeland30 Land Cover Map in Europe Using Land Cover Field Survey Database - Lucas

Abstract

<p><strong>Abstract.</strong> High-resolution land cover maps are one of the technological innovations driving improvements in many fields influenced by Geographic Information Systems (GIS) and Remote Sensing. In particular, the GlobeLand30 (GL30), global LC map with spatial resolution of 30<span class="thinspace"></span>m, is thought to be one of the highest quality high-resolution products. However, these LC maps require validation to determine their suitability for a particular purpose. One of the best ways to provide useful validation reference data is to do a high-level accuracy field survey, but this is time consuming and expensive. Another option is to exploit already available datasets. This study assesses thematic accuracy of GL30 in Europe using LUCAS as a validation reference, because it is a free and open field survey database. The results were generally not good, and very bad for some classes. Analysis was then restricted to a small region (Lombardy, Italy) where LC data of higher resolution than those of GL30 were available. LUCAS was also found to be incoherent with this product. Further comparisons of LUCAS with other independent sources confirmed that the LC attributes of LUCAS are inconsistent with expectations. Although these findings may not be generalized to other regions, the results warn against the suitability of LUCAS as ground truth for LC validation. The paper discusses the process of thematic accuracy assessment of the GL30 and the applicability of LUCAS for high-resolution global LC validation.</p>

Deforestation Mapping Using Sentinel-1 and Object-Based Random Forest Classification on Google Earth Engine

“…Classified thematic maps need some validation to be evaluated their accuracy and suitability for future use (Bratic et al, 2018b). As in most of the times it is impossible to carry out field survey and to validate the output, it is widely adopted to do a validation using external data source as a reference.…”

Open educational resources for validation of global high-resolution land cover maps

“…Hereafter, two case studies included in the educational resources, which make use of these two types of reference datasets are described in detail. More information can be found in the work of Bratic et al (2018a) and Bratic et al (2018b). The educational resources show how to use the tools available in QGIS 2.18 to aid in preprocessing and how to use the software developed to perform the validation.…”