Photo-interpretation in LUCAS alters temporal trends of woody and wetland grazing extents

Abstract

The Land Use/Cover Area frame Statistical survey (LUCAS) is a large-scale harmonised data collection exercise with wide-spread use cases, particularly those related to training artificial intelligence models. The survey provides valuable information regarding landcover, land use, and environmental parameters across European Union countries on a three-year basis since 2006. Recently, the survey has been used to monitor grazing in semi-natural habitats such as livestock agroforestry. However, in recent years LUCAS has become more reliant on photo-interpretation in the second phase of the survey. We identified an underreporting (p < 0.001) of grazing in points surveyed using photo-interpretation and statistically corrected for this systematic bias. We analysed the extents of grazing under woody and wetland landcovers in the United Kingdom over time, comparing extents estimated from the raw LUCAS data, statistically corrected data, and data from only field surveyed points. To the best of our knowledge, this paper provides the first estimation of the extent of grazing under wetland habitats in the United Kingdom, which we estimated to be approximately 72 000 hectares in 2018. This paper updated the extent of livestock agroforestry in the United Kingdom, estimating that over 300 000 hectares of woody landcovers were grazed in 2018, which would have been underestimated by almost 100 000 hectares if statistical corrections had not been applied. Use of the LUCAS dataset can provide valuable insight into the effectiveness of environmental policies, but the impact of photo-interpretation on temporal trends needs to be accounted for.