Spatial monitoring of grassland management using multi-temporal satellite imagery. (June 2020)

• Record Type: Journal Article
• Authors: Stumpf, Felix; Schneider, Manuel K.; Keller, Armin; Mayr, Andreas; Rentschler, Tobias; Meuli, Reto G.; Schaepman, Michael; Liebisch, Frank
• Is Part Of: Ecological indicators. Volume 113(2020)
• Publication Date: 2020-06
• Subjects: Grassland management -- Spatial monitoring -- Spectral time series -- Management practice -- Mowing -- Grazing -- Use intensity; Environmental monitoring -- Periodicals; Environmental management -- Periodicals; Environmental impact analysis -- Periodicals; Environmental risk assessment -- Periodicals; Sustainable development -- Periodicals; 333.71405
• DOI: 10.1016/j.ecolind.2020.106201
• Languages: English
• ISSNs: 1470-160X

Abstract

Highlights: Remote sensing imagery allows mapping of grassland management at large scales. Landsat phenology metrics capture complex spatio-temporal management patterns. Mapped management classes describe use intensities of mowing and grazing practices. Management maps contribute to balance intensification and ecosystem functioning. Abstract: Spatial monitoring of grassland management is crucial for ecosystem assessment and the establishment of sustainable agriculture. Switzerland is covered by large areas of small structured grassland parcels differing in management practices and use intensities, making the mapping of grassland management challenging. We present a monitoring tool to map grassland management, distinguishing between mowing- and grazing practice, and between different use intensities for Swiss agroecosystems. By analyzing pixelwise spectral time series of 2015, derived from satellite imagery of the Landsat archive, we estimated the number of management events and biomass productivity. Both estimates were used to map classes of dominant management practices and use intensities following a stepwise clustering approach. The grassland management (GM) classes were evaluated relative to established spectral and topographical patterns of grassland use intensity, and in terms of spatial conformity with available regional land use data. The GM classes were also analyzed with respect to management related vegetation plot data on species diversity, as well as on indicator values for nutrient supply and management tolerance. The stepwise clustering gave three use intensity classes for each dominant management practice of grazing (pasture) and mowing (meadow). Use intensity was higher for meadows than pastures with a distinct intensity gradient for each grassland practice. The GM classes reproduced established spectral and topographical patterns of grassland use intensity, indicated by increased standard deviations (SD) of spectral time series profiles (e.g. mean SD of 0.048 for pastures and 0.054 for meadows) and lower slopes (e.g. mean slopes of 10° for pastures and 7° for meadows). The averaged spatial conformity of the GM classes with a cantonal land use map was 82% for meadows and 97% for pastures. The GM classes spatially matched with land use patterns of three subregions, e.g. with an areal proportion of 73% pasture classes for a subregion dominated by grazing. Moreover, the GM classes reproduced established vegetation patterns of grassland use intensity along the GM intensity gradient, showing a mean decrease in species richness (33%), as well as a mean increase in indicator values for nutrient supply (5%), grazing tolerance (4%), and mowing tolerance (6%).