int. j. remote sensing, 1998, vol. 19, no. 16 , 3141±3168 Global land cover classi®cations at 8 km spatial resolution: the use of training data derived from Landsat imagery in decision tree classi®ers R. S. DE FRIES, M. HANSEN, J. R. G. TOWNSHEND and R. SOHLBERG Laboratory for Global Remote Sensing Studies, Department of Geography, University of Maryland, College Park, Maryland, 20742, USA ( Received 10 June 1997; in ®nal form 13 March 1998 ) Abstract. This paper reports a study which aims to (i) develop methodologies for global land cover classi®cations that are objective, reproducible and feasible to implement as new satellite data become available in the future and (ii) provide a global land cover classi®cation product based on the National Aeronautics and Space Administration /National Oceanic and Atmospheric Administration Path®nder Land (PAL) data that can be used in global change research. The spatial resolution for the land cover classi®cation is 8km, intermediate between our previously published coarse one degree by one degree spatial resolution and the 1km global land cover product being developed under the auspices of the International Geosphere Biosphere Program. We ®rst derive a global network of training sites from Landsat imagery, using 156 Landsat scenes mostly from the Multispectral Scanner System, to identify over 9000 pixels in the PAL data where we have high con®dence that the labelled cover type occurs. We then use the training data to test a number of metrics that describe the temporal dynamics of vegetation over an annual cycle for potential use as input variables to a global land cover classi®cation. The tested metrics are based on: (i) the ratio between surface temperature and Normalized Dierence Vegetation Index (NDVI); (ii) seasonal metrics derived from the NDVI temporal pro®le, such as length of growing season; (iii) a rule-based approach that determines cover type through a series of hierarchical trees based on surface temperature and NDVI values; and (iv) annual mean, maximum, minimum and amplitude values for all optical and thermal channels in the Advanced Very High Resolution Radiometer (AVHRR) (PAL) data. Highest mean class accuracies from a decision tree classi®er were obtained using the annual mean, maximum, minimum, and amplitude values for all AVHRR bands. Finally, we apply these metrics to 1984 PAL data at 8km resolution to derive a global land cover classi®cation product using a decision tree classi®er. The classi®cation has an overall accuracy between 81.4 and 90.3%. The Landsat images used for deriving the training data and the methodology for classi®cation of AVHRR data at 8km resolution can also be applied to 1km AVHRR data and, in the future, Moderate Resolution Imaging Spectroradiometer (MODIS) data at 250 and 500m resolution. Digital versions of the land cover dataset and detailed documentation can be found on the World Wide Web at http:// www.geog.umd.edu / landcover/8km-map.html. 1. Introduction The geographic distribution of vegetation on the Earth’s land surface plays a central role in many Earth system processes. Terrestrial vegetation modulates the Earth’s climate by taking up carbon dioxide from the atmosphere for photosynthesis, losing water vapour through the leaves’ stomates, and strongly in¯uencing albedo (Sellers et al . 1996). Furthermore, the distribution of vegetation over the land surface 0143± 1161/98 $12.00 Ñ 1998 Taylor & Francis Ltd Downloaded By: [Duke University] At: 23:33 16 June 2010