Mapping grassland use with Landsat and Sentinel-2 time series
Topics: Remote Sensing
, Land Use
, Agricultural Geography
Keywords: remote sensing, Landsat, Sentinle-2, Grassland, Mowing, Land use, Land cover
Session Type: Virtual Paper
Day: Thursday
Session Start / End Time: 4/8/2021 03:05 PM (Pacific Time (US & Canada)) - 4/8/2021 04:20 PM (Pacific Time (US & Canada))
Room: Virtual 13
Authors:
He Yin, Department of Geography, Kent State University, USA
Patrick Griffiths, European Space Agency (ESA), Directorate of EO Programmes, Science Applications & Climate Department, Italy
Patrick Hostert, Geography Department, Humboldt-Universität zu Berlin, Germany
Volker C. Radeloff, SILVIS Lab, Department of Forest and Wildlife Ecology, University of Wisconsin-Madison, USA
,
,
,
,
,
,
Abstract
Grassland, among the largest ecosystems in the world, provide valuable ecosystem services, and harbor unique biodiversity . One important question for both land use management and agricultural policy is whether grasslands are used, and especially mowed, or not. However, mapping grassland use is challenging because it is highly dynamic over space and time, requiring dense and long time series of satellite imagery. With both Landsat and Sentinel-2 data being available, it is now possible to analyze dense time series of satellite imagery for grassland mapping. However, doing so requires novel methodologies. We developed and compared two approaches for detecting mowing events in grassland using a combined time series of Landsat and Sentinel-2 imagery. The first approach was a residual analysis using an iterated harmonic model fitting. The second was a temporal moving window analysis. We tested our approaches in France, a major agricultural producer in Europe with diverse grasslands that cover 45% of its agricultural land area. Our results suggest the reliability of both methods with their own merits. We found more than half the grasslands in France were mowed in 2019. Our results demonstrate the opportunities offered by multi-sensor satellite data for monitoring land use dynamics.