Mapping of grassland using seasonal statistics derived from multi-temporal satellite images

Abstract

Grasslands cover about 40 % of the earth’s surface. Due to its great expanse and diversity, low-cost tools for inventory, management and monitoring are needed. Remote sensing is a useful technique for providing accurate and reliable information for land use planning and to support large scale grassland management. In the context of “GIO land” (Copernicus initial operations land), which is currently implemented by the European Environment Agency (EEA), the permanent grasslands of 39 countries in Europe has to be mapped with an overall classification accuracy of more than 80 %. Since grassland canopy density, chlorophyll status and ground cover is highly dynamic throughout the growing season, no unique spectral signature can be used to map grasslands. Therefore, it is necessary to use time series to characterize the phenological dynamics of grasslands throughout the year to be able to discriminate among them and other vegetation which shows similar spectral response such as crops. The article outlines the adopted classification method using multi-temporal, multi-scale and multi-source remotely sensed data. The approach is based on the supervised decision Tree (DT) classifier C5 in combination with previous image segmentation and seasonal statistics of bio-physical parameters. In this paper the results of entire Hungary are presented. The accuracy assessment of the grassland classification was carried out using 340 sample points mainly derived from a ground-based European field survey program. The multi-temporal grassland classification of Hungary reached an overall accuracy of 92.2 %.