Copernicus Land services to improve EU land cover statistics.

This report analyses the potential use of the Copernicus program as a tool to provide harmonized statistical data on several environmental topics. Some of the Copernicus land services are specifically considered, but the report aims mainly at stating some good practices that can be applied to different topics whenever remote sensing is able to identify the relevant land cover types with a good accuracy. The main message that comes out is that straightforward estimation of land cover area is seldom acceptable. The so-called pixel counting approach should not be used without a previous estimation of the bias coming from the unbalance of commission and omission errors. General statistical rules for the estimation and correction of the bias are reviewed. The LUCAS survey (Land Use/Cover Area-frame Survey) is run by Eurostat every 3 years on a two-phase sample of points in the European Union (EU). Initially conceived to provide independent agricultural estimates, it has switched to agro-environmental purposes aiming at providing comparable information for the different countries of the EU. In the last LUCAS survey (2015), a sample of around 274,000 has been selected for field visits, with a complementary sample of nearly 60,000 photo-interpreted points to cover areas of difficult access. In this report we focus on the observation of linear landscape features by counting the number of intersections with 250 m transects associated to each of the visited points. The survey gives valuable observations for geographic comparison and spatial analysis, but still needs some polishing to provide credible estimates of the changes between two reference dates. Simultaneous photo-interpretation of images of different years seems to be a key element to limit the number of fake changes identified. Automatic identification of linear elements on Very High Resolution images should become an efficient covariate for the estimates (both spatial and temporal), but the attempts carried out in the framework of the Copernicus program still need some improvement.JRC.D.5-Food Securit