USING REMOTE SENSING TO ASSESS RUSSIAN FOREST FIRE CARBON EMISSIONS A. S. ISAEV 1 , G. N. KOROVIN 1 , S. A. BARTALEV 2 , D. V. ERSHOV 1 , A. JANETOS 3 , E. S. KASISCHKE 4 , H. H. SHUGART 5 , N. H. F. FRENCH 6 , B. E. ORLICK 7 and T. L. MURPHY 7 1 Center of Problems of Ecology and Productivity of Forests (CEPL), Russian Academy of Science, Novocheriomuskinskaya Str. 69, 117418 Moscow, Russia E-mail: isaev@cepl.rssi.ru 2 Space Application Institute/Joint Research Center, CCR/TP 440, I-21020 Ispra (VA),Italy 3 World Resources Institute, Washington, D.C., U.S.A. 4 Department of Geography, University of Maryland, 2181 Maryland, LeFrak Hall, College Park, MD 20742, U.S.A. 5 Department of Environmental Sciences, University of Virginia, Clark Hall, Charlottesville, VA 22903, U.S.A. 6 Altarum, PO Box 134001, Ann Arbor, Michigan 48113-4001, U.S.A. 7 U.S. National Imagery and Mapping Agency, Washington, D.C., U.S.A. Abstract. Russian boreal forests are subject to frequent wildfires. The resulting combustion of large amounts of biomass not only transforms forest vegetation, but it also creates significant carbon emissions that total, according to some authors, from 35–94 Mt C per year. These carbon emissions from forest fires should be considered an important part of the forest ecosystem carbon balance and a significant influence on atmospheric trace gases. In this paper we discuss a new method to assess forest fire damage. This method is based on using multi-spectral high-resolution satellite im- ages, large-scale aerial photography, and declassified images obtained from the space-borne national security systems. A normalized difference vegetation index (NDVI) difference image was produced from pre- and post-fire satellite images from SPOT/HRVIR and RESURS-O/MSU-E images. A close relationship was found between values of the NDVI difference image and forest damage level. High- resolution satellite data and large-scale aerial-photos were used to calibrate the NDVI-derived forest damage map. The method was used for mapping of forest fire extent and damage and for estimating carbon emissions from burned forest areas. 1. Introduction Forests are an important element of the biosphere and a source of many forest products. Balanced use of forests correlates strongly with type of business activity and careful management of forest ecosystems. New data and non-traditional ap- proaches are needed for monitoring forest condition and dynamics under natural and anthropogenic influences. Data quality can be significantly improved using remote sensing resources and GIS technologies. This results in a more accurate quantitative assessment of dynamic processes in forests, such as large-scale natural events, human activities, and long-term climate change. More accurate assessments Climatic Change 55: 235–249, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.