NASA and Russian Scientists Observe Land-Cover and Land-Use ...

University of Michigan Environmental Spatial Analysis Laboratory

NASA and Russian Scientists Observe Land-Cover and Land-Use Change and Carbon in Russian Forests

ABSTRACT

Kathleen M. Bergen, Susan G. Conard, R.A. Houghton, Eric S. Kasischke, Vyacheslav I. Kharuk, Olga N. Krankina, K. Jon Ranson, Herman H. Shugart, Anatoly I. Sukhinin, and Rudolf F. Treyfeld In 1997, several project teams of the NASA Land-Cover Land-Use Change Program began working with Russian organizations to try to quantify and understand the past, present, and future land-cover and land-use trends in Russian boreal forests. Selected results of completed and ongoing research projects are discussed in four categories: forest dynamics, fire and fire behavior, carbon budgets, and new remote sensing analysis methods. This research has helped pave the way for collaborations with international organizations and other networks, and collaborations at several scales are now making it possible for Russian and US scientists to work together to further our knowledge on the influence of land-cover and land-use change throughout the world. Keywords: international forestry; remote sensing; Russia

T

he Earth Science Enterprise (ESE) is a program of the National Aeronautics and Space Administration (NASA) studying the terrestrial, oceanic, and atmospheric

34

Journal of Forestry

•

June 2003

dynamics of our planet. Under the umbrella of the ESE, the NASA LandCover Land-Use Change Program (LCLUC) focuses on the human and natural drivers and consequences of

land-cover and land-use change (http:// lcluc.gsfc.nasa.gov). The LCLUC program is particularly focused on regions of significant change involving the interaction of human and natural disturbances, especially those interactions affecting the terrestrial component of the global carbon cycle. Because forests hold the largest reserves of terrestrial carbon, a number of LCLUC scientific studies around the globe—from the United States and North America to Central and South America, Africa, Figure 1. Ecosystem types within the Russian Federation. Sources: Data from ESRI and World Wildlife Fund.

Europe, and Asia—have focused on the dynamics of forest cover: forest conversion, fire, logging, insects, pollution, and post-disturbance succession. Most LCLUC studies use satellite remote sensing data, field and inventory data, socioeconomic data, and modeling methods to characterize and predict forested land-cover and landuse change trends in the geographic region of study. In the past decade, new opportunities have opened up for American scientists sponsored by NASA to collaborate with Russian researchers from the Russian Academy of Sciences, Russian State Forest Inventories, Russian Federal Forest Service, and other organizations, to study the dynamics of the world’s largest forest land base—the forests of the Russian Federation (henceforth referred to as Russia) (fig. 1). Russia covers 8 percent of the world’s terrestrial land mass and supports over 20 percent of the world’s forests (Kukuev et al. 1997). This forested landscape is dynamic. Fire is the major annual disturbance throughout the Russian forest, especially in large portions of the more remote areas, where fire suppression is less intensive or impossible. The timber resources of Russia may be increasingly vital to the health of a restructuring economy, and at the same time they are part of a globally important ecosystem. The interactions between the impacts of humans, logging and fire, and climate change in the Russian boreal forest are complex, and their consequences for carbon dynamics are not well understood. In 1997, several NASA LCLUC project teams began working with Russian organizations to try to quantify and understand the past, present, and future land-cover and land-use trends of this important, largely forested region. Selected results of completed and ongoing research projects are grouped under four LCLUC categories: forest dynamics, fire and fire behavior, carbon budgets, and new remote sensing analysis methods.

Figure 2. Trends in forest sector statistics for the Russian Federation and for each of the three territories of the LCLUC case study sites: Total wood removal (106m3). Sources: Pre-1991 data from Forest Complex of USSR, 1991, All Union Scientific Research Institute of Economics; post-1991 data from Industry of Russia, Goskomstat of Russia.

Forest Dynamics

Forest conversion in Russia has occurred over the past several centuries to make way for human settlements and limited agriculture. Russia currently has a population of approximately 150 million people, and its 13 largest cities have over 1 million inhabitants each. Agriculture is practiced on approximately 8 percent of arable land, including land converted from forests. As noted, Russian forests are subject to further change from several continuing natural and anthropogenic factors: fire, changing climate, logging, pollution, mineral exploration, insects, and agricultural abandonment. These dynamics are the focus of a pair of LCLUC projects in Central Siberia: “Effects of the Development of the Baikal-Amur Mainline Railroad on Patterns of Carbon Flux in Southern Siberia” and “Modeling Siberian Forest Land-Cover Change and Carbon under Changing Economic Paradigms.” Researchers at the University of Michigan, Sukachev Institute of Forest Research, Novosibirsk Institute for Economics, Irkutsk Ministry of Natural Resources, University of Virginia, University of Maryland, ERIMInternational, and ESSA Technologies have used time-series remote sensing, Russian government statistical data, and models to investigate the interaction between natural and anthropogenic drivers and consequences of land-cover change. Case study sites have been established in Tomsk Oblast, Krasnoyarsk Krai, and Irkutsk Oblast—three of the four largest provinces in Siberia in terms of timber resources (Blam et al. 2000). Each of

the case study sites is the footprint of a Landsat scene (185 km × 185 km) and lies between 55° and 60° north latitude. The forest in the case study sites is typical of the Siberian taiga: a mosaic of several communities at various age and disturbance stages. The “light-needle” coniferous forest includes Scots pine (Pinus silvestris) and larch (Larix sibirica, Larix gmelinii) and is found in a range of terrain from xeric to mesic. In the study area, the “dark needle” forests of spruce (Picea obovata), fir (Abies sibirica), and Siberian pine (Pinus sibirica) grow mainly on slopes of northern exposure, or as accompanying species in light-needle-dominated communities (Bondarev 1990). Further north these types are more broadly distributed across the landscape. Deciduous birch-aspen (Betula pendula, Populus tremula) forests and mixed pine-deciduous-larch forests are present mainly on postdisturbance sites. Analyses of statistical data from Russian agencies at national and territorial levels document a steady high rate of logging from 1965 to 1990, a subsequent sharp decrease in logging activity starting in 1990 (coinciding with the dissolution of the Soviet Union and its state-controlled and subsidized economy), and a very slight increase in the past several years (fig. 2). Statistics derived from remote sensing show a similar pattern. A high-resolution Corona image from the Krasnoyarsk case study site from 1965 showed that a high percentage of the forested landscape had been logged or burned by that date. This is also evident from the amount of forest in early secondary June 2003

•

Journal of Forestry

35

Figure 3. SPOT VEGETATION 10-day composite image collected over Sakhalin Island in September 1998. The 1998 fires are the dark red areas. Large fires also occurred in 1989, with the scars from these fires still visible in the 1998 satellite imagery as areas of pink.

succession in the 1974 Landsat MSS image. Over the 25-year Landsat analysis of the Krasnoyarsk case study site, the recently logged class declined from 10 percent to 4 percent in the first 16 years (1974–90) and subsequently remained constant. The area of young forest regrowth in the case study site increased from 17 percent to 24 percent during the first 16 years (1974–90) and decreased to 9 percent afterward (1990–2000) due to decreased logging. Total forest cover increased by only 1 percent (from 57 to 58 percent) between 1974 and 1990, and then by 10 percent between 1990 and 2000 (to 68 percent) (Bergen et al. 2003). Other disturbances are found in the case study sites. Between 1930 and 1957 the Siberian silkmoth (Dendrolimus superans sibiricus) damaged and killed about 7 million ha of forest. The most recent outbreak occurred in 1994–96 in Krasnoyarsk Krai where 0.7 million ha of forest were affected (Kharuk et al. 2001; Kharuk et al. in press). Insect infestation leads to damage or death of forest stands and increased susceptibility to fire. Results from the LCLUC remote sensing analysis for the Krasnoyarsk site show that 8 percent of total land cover was affected by insect infestation in 1994–96. In the case of fire, remote sensing analysis of the Krasnoyarsk site 36

Journal of Forestry

•

June 2003

shows that fire may have accounted for less than 2 percent of total area of change over the period 1974–2000. However, this analysis is not entirely conclusive as some low-intensity ground fires and fires that occurred between the image dates may have been missed, and the actual percent of area disturbed by fire over Krasnoyarsk Krai is significantly higher. Remote sensing assessment of the Krasnoyarsk site shows that agricultural use of the landscape may have been reduced from 14 percent to 8 percent gradually over the study period 1974–2000 largely through abandonment of collective farming areas (Bergen and Zhao 2003). Most of this recently abandoned agricultural land is reverting to young deciduous forest. These remote sensing analyses also show that the representation of specific forest communities on the landscape is changing. At the Krasnoyarsk site, the proportion of coniferous forest has been reduced continuously over the period 1975–2000, whereas deciduous and mixed forests have increased. These changes are due to logging prior to 1975, agricultural abandonment, and regrowth after insect damage and fire. Using models, it is possible to project what the forest in these regions will look like in future decades and centuries. Forest gap models calibrated for the Siberian boreal forest show that the composition of the forest changes significantly after logging and fire and may not return to its original composition for centuries even without additional disturbance (Shugart et al. 1992). Remote sensing land-cover classifications, remote sensing land-cover change results, and forest gap model output are being incorporated into a landscape simulation model. VDDT/ TELSA (Vegetation Disturbance Dynamics Tool/Tool for Exploratory Landscape Scenarios Analyses) is a GIS-based semi-Markovian modeling software tool (Beukema and Kurz 1998). This modeling framework will allow development of spatially explicit forested land-cover change and carbon projections for these important timber resource regions in the Central Siberian boreal forest.

Fire and Fire Behavior

A major natural disturbance process in boreal systems, fire affects about 12 to 15 million ha of closed boreal forest annually, most of it in Eurasia. In Russia as a whole, this disturbance exceeds the annual area harvested or disturbed by other natural agents, such as insects. Data on the extent and impacts of fire in boreal forests are scarce and often contradictory. Several recent papers indicate that the impact on terrestrial carbon storage of fires in boreal forest regions has been vastly underestimated (Conard and Ivanova 1997; Kasischke et al. 1999; Conard et al. 2002). Furthermore, changes in land management, land-use practices, and regional climate, coupled with fire suppression capability, will affect fire risk and ecosystem damage from fires in ways that are poorly understood at present. Improved understanding of the landscape extent and severity of fires; factors affecting fire behavior and intensity; and effects of fire on carbon storage, air chemistry, vegetation dynamics and structure, and forest health and productivity is needed before such considerations can be adequately addressed in regional planning. Monitoring effects on a landscape scale and providing inputs into global and regional models of carbon cycling and atmospheric chemistry requires development of validated remote sensing–based approaches for measurement of burned areas and fire severity. Extent of fires in the boreal forest is the focus of the project “Determining the Contribution of Emissions from Boreal Forest Fires to Inter-annual Variations in Atmospheric CO2 at High Northern Latitudes,” a NASA Interdisciplinary Science (IDS) project affiliated with the NASA LCLUC program. The goal of this project is to quantify the role of boreal forest fires, globally and regionally, in the direct emission of trace gases (CO2, CO, and CH4) to the atmosphere. On a global scale, some 15 million ha of boreal forests and peatlands burned in 1998, releasing an estimated 188 to 440 Tg (1012 g) of carbon into the atmosphere (Conard et al. 2002; Kasischke and Bruhwiler 2002). The emission plumes from these fires were detected by at-

mospheric satellite remote sensing systems collected through surface sampling networks (Hsu et al. 1999; Fromm et al. 2000; Spichtinger et al. 2001). Boreal forest fire emissions in 1998 were also the source of anomalously high levels of CO (Forster et al. 2001; Wotawa et al. 2001; Kasischke and Bruhwiler 2002) and CH4 (Dlugokencky et al. 2001). Further analyses of satellite imagery, combined with field observations and verification, are providing key information for reducing uncertainties. Figure 3 presents a SPOT VEGETATION image collected over Sakhalin Island in September 1998. The 1998 fires are the dark red areas in the figure and comprise some 3 to 4 percent of the total land surface on Sakhalin Island (Kasischke et al. 1999). Consultation with local forestry officials revealed that large fires also occurred in 1989, with the scars from these fires still visible in the 1998 satellite imagery as areas of pink. Comparison of satellite imagery collected during the spring and fall of 1998 shows that much of the area that burned on Sakhalin Island occurred in one of the 1989 burned areas. Because of the large amounts of standing dead snags present in the 1989 burns, the 1998 fires in these stands resulted in high levels of trace gas emissions. Analyses of finer-scale satellite imagery are providing key insights on patterns of trace gas emissions released from fires in different regions of Russia. Use of remote sensing to study and resolve research questions related to landscape-scale fire behavior has recently become integrated as part of the LCLUC program. In the project “Modeling and Monitoring Effects of Area Burned and Fire Severity on Carbon Cycling, Emissions, and Forest Health and Sustainability in Central Siberia,” researchers at the USDA Forest Service, Canadian Forest Service, Sukachev Institute of Forests, and Institute of Chemical Kinetics and Combustion in Novosibirsk are collaborating to scale up from sitespecific data on fire behavior and fire effects to improve the application of new remote sensing technologies for monitoring and modeling burned

areas, fire intensity, and fire effects in the Siberian boreal forest. This project is focusing on the following activities: (1) combining ground data, aircraft data, and intermediate-resolution satellite data (Landsat-ETM+) to improve on current coarse-resolution satellite data (AVHRR) for estimating fire spatial extent and burn severity; (2) using ground data from experimental fires to refine models of fire severity and seasonality on fire behavior, fuel dynamics, carbon storage and emissions, and ecosystem health; and (3) combining the information on fire spatial extent and severity with the models calibrated from the experimental ground data to develop a regional model for monitoring and predicting capabilities. To date, this project has conducted 12 experimental surface and subcanopy fires at study sites near Yartsevo and Boguchani (fig. 4). Calculated fire intensities have ranged from less than 1,000 to more than 23,000 kW/m (McRae et al. 2002 and unpublished data). Estimated carbon release from these fires has varied from 4.8 to 15.4 t/ha. Based on sampling of canopy fuels, an additional 3 to 7 t/ha of carbon might be released if a crown fire were to occur (McRae et al. 2002 and unpublished data). These results point out the tremendous variability in carbon release from fires in dry Scotch pine types, and emphasize the need to quantify fire severity in addition to burned area to obtain accurate estimates of fire emissions. Scientists from the United States, Canada, and Russia are collaborating to quantify gaseous and aerosol emissions (Baker and Hao 2002; Koutzenogii et al. 2002), fire behavior and fuel consumption, and ecosystem effects of these fires, including effects on soil chemistry, respiration and microorganism populations, small animals, and plant communities. Strong correlations exist between components of the Canadian Forest Fire Weather Index System (Canadian Forestry Service 1987) and fuel consumption, suggesting that models can be developed to predict fire severity and carbon release. Investigators are developing a database of aerial infrared imaging and ground data for experimental

Figure 4. Fireline intensity in controlled burns: (top)1,156 kW/m2; (middle) 9,018 kW/m2; (bottom) 23,824 kW/m2.

fires and wildfires to better quantify fire behavior and energy release at different scales and to better interpret thermal signals and fire scars on AVHRR and MODIS imagery. This project has produced one of the most comprehensive datasets of experimental fires ever gathered in the boreal zone. Carbon Budgets

The LCLUC program seeks to build on research quantifying landcover change, including forest dynamics and fire, to provide better estimates of regional and global carbon budgets. The carbon balance of northern midlatitude terrestrial ecosystems is uncertain, yet important for predicting future rates of CO2 increase in the atmosphere. Analyses based on atmospheric data and models show a net terrestrial sink that ranges between 3.5 and 0.7 PgC/yr in northern mid-latitudes. Analyses based on forest invenJune 2003

•

Journal of Forestry

37

Doug Oetter, Oregon State University

Figure 5. Example of ground data used to model carbon stores: forest inventory polygons (raw data and buffered polygons selected for modeling) from Tosno Leskhoz overlaid with Landsat TM image.

tories suggest a lower but still uncertain sink, especially for Russia and the former Soviet Union, where estimates of carbon balance range between a source of 0.5 PgC/yr and a sink of 1.0 PgC/yr (Shvidenko et al. 1996). Two LCLUC projects are focused primarily on carbon; one is regional and the other is Russia-wide. Carbon balance of forest ecosystems in the St. Petersburg region of European Russia is the focus of “Modeling Carbon Dynamics and Their Economic Implications in Two Forested Regions: Pacific Northwestern USA and Northwestern Russia.” This research project is part of a long-term NSF-Ecosystems and NASA-LCLUC funded collaboration between the Oregon State University, USDA Forest Service Pacific Northwest (PNW) Re38

Journal of Forestry

•

June 2003

search Station, Russian Federal Forest Service, Northwestern State Forest Inventory Enterprise, St. Petersburg Hydrological Institute, Komarov Botanical Institute, and St. Petersburg Forest Academy. To develop a comprehensive regional carbon budget, project scientists examined carbon stocks and their change in live forest biomass, coarse woody debris, peatlands, forest products, and soils. Landsat Thematic Mapper (TM) data from 1986 to 1995, in conjunction with the stand-level forest inventory data from 1992–93, were used to map land cover and carbon storage for a 76,850 km2 region around St. Petersburg. The forest inventory databases included a total of 12,791 mapped stand polygons in three separate locations within the region (fig. 5). Bio-

mass was calculated from field data collected for each polygon (Kukuev et al. 1997) using local allometric equations (Alexeev and Birdsey 1998). Approximately 1,500 polygons were selected based on size and spectral signals and randomly subdivided into training and testing sets to develop and test a model for calculating continuous biomass estimates directly from TM values. To reduce model bias, canonical correlation analysis was employed to produce indices of transformed TM values, which were then used with reduced major axis regression to produce linear models. The calculation of changes in regional carbon stores is based on changes in land cover detected on the sequence of Landsat TM scenes from early 1980s to the most recent Enhanced Thematic Mapper+ (ETM+) images. This result is now being compared to estimates based on historic summaries of forest inventory data showing that between 1973 and 1993, C stores in the St. Petersburg region increased from 185 to 250 million tons, or nearly 20 percent (Kobak et al. 1999). To highlight the diversity among forest regions in the patterns of carbon allocation and change over time, the carbon dynamics of the St. Petersburg region of Russia are being compared with the Pacific Northwest region of North America. The comparison is being synthesized into a book to be published in the Springer-Verlag Ecological Studies series. Following this comparison, a new phase of research began in 2001, the major goal of which is to fully integrate the socioeconomic drivers into the analysis of changes in carbon stores, both in the PNW and in the St. Petersburg regions of Russia. Project researchers are examining how the recent changes in land-use interact with changes that occurred earlier to shape the present pattern of C stores and flux, assessing the temporal patterns of ecosystem response to changes in socioeconomic conditions, and estimating to what degree humans can manipulate regional carbon stores on a decadal time scale. To complement the in-depth analysis of carbon dynamics carried out in the St. Petersburg region, a new LCLUC project, “Changes in Terres-

trial Carbon Storage in Russia as a Result of Recent Disturbances and LandUse Change,” will quantify carbon change over the entire Russian Federation at coarser spatial and temporal scales. Researchers at the Woods Hole Research Center, Russian Federation State Forest Inventories, and Oregon State University are determining the current distribution of carbon storage in Russia and changes in that storage over the past decades using an approach that integrates forest inventory data, results of ecological studies, data on land-use change, and a combination of Landsat and MODIS (Moderate Resolution Imaging Spectrometer) data and products. This project has three goals. The first goal is to construct a map of the forest biomass of the Russian Federation. The approach will use forest inventory data on wood volumes at the vydel (or polygon) level, to calibrate Landsat data (30 m spatial resolution). Once this is complete, the Landsat data will be used to calibrate countrywide MODIS data (250 m resolution). Theoretically, this will result in a biomass map for all of Russia. The second goal is to construct a map of forest age. The approach will be similar to that described for the first goal: Forest inventory data on age-since-last-disturbance will be used to calibrate Landsat TM data, and the latter will be used to calibrate MODIS data. A map showing where forests have been disturbed over the past 30 to 50 years will be the result. The third goal is to construct a map of sources and sinks of carbon. For any particular set of environmental conditions, rates of carbon accumulation are a function of forest age (or current biomass). Rates of forest growth and decay following disturbance, determined from forest inventory data and the ecological literature for the major ecosystems of Russia, will be applied to the maps of forest ages and biomass to determine the sinks and sources of carbon from living vegetation, woody debris, and soil carbon. A dynamic bookkeeping model (Houghton et al. 1999) is being used to calculate the annual flux of carbon to or from the atmosphere as a result of disturbances over the past few decades.

Figure 6. High-resolution classifications of forest disturbance from Landsat-7 ETM+ and radars (left, with fire scars in red) were used to develop regional classifications using MODIS and Radarsat data (right).

New Remote Sensing Methods

Russia’s great size and the continual dynamics of its forests are the reasons remotely sensed data is used in studying forest and carbon dynamics over its territory. In one project, “Combined Satellite Mapping of Siberian Landscapes,” there has been a particularly strong focus on the development of new methods of extracting landcover change information from remote sensing data. The aim of this study carried out by researchers at NASA’s Goddard Space Flight Center and Sukachev Institute of Forest Research is to evaluate the utility of several different remote sensing instruments, alone and in combination, to produce land-cover and land-cover change maps that can be used to determine the extent and rate of natural and anthropogenic impacts on the Siberian boreal forest. This central Siberian study is centered at 94 degrees East and 58 degrees North and is within the International GeosphereBiosphere Program (IGPB) Western Siberian Transect. In a two-level approach, classification procedures were developed for moderate-coarse resolution MODIS data (250 m – 1 km) in combination with Radarsat ScanSAR data (1 km) and areas of change were delineated. Following this step, areas of known change were further analyzed using higher spatial resolution (~30 m) satel-

lite sensors, including Landsat ETM+, Radarsat medium-fine resolution modes, ERS SAR (the European satellite synthetic aperture radar), and JERS SAR (the Japanese satellite synthetic aperture radar). Analysis methods focused on determining characteristic features related to the several forest disturbances using image analysis techniques including decision tree classification and spatial texture analysis (Ranson et al. 2001; Sun et al. 2002). The accuracy of the results was determined by comparing results with areas of known forest types based on ground studies, forestry maps, and high-resolution IKONOS data (1–4 m). Results indicated that, when available, high-resolution radar data improved classification of disturbance sites over Landsat ETM+ data alone. SAR data were especially suitable for detailed mapping of the complex landscape created by the juxtaposition of logged areas in different stages of regeneration and fire scars. These data and procedures were then used to classify land cover and disturbance on moderate-coarse resolution MODIS and Radarsat ScanSAR images (fig. 6). Texture measures of the Radarsat data (such as homogeneity, contrast, dissimilarity, mean, variance, and entropy) increased the information content of the Radarsat ScanSAR data. The combination of the radar June 2003

•

Journal of Forestry

39

and optical data provided better classification results of the central area than either data type alone. The ongoing utility of this study is that robust sensor-based techniques have been developed for mapping forest disturbance types in the Russian boreal forest, and that once baseline disturbance is established, disturbance dynamics may be assessed through annual satellite data analysis. Summary and Future Directions

Several themes are common to these NASA LCLUC projects. • Results underscore the importance of remotely sensed data to assess forested regions as geographically vast and as dynamic as the Russian boreal forest. Remote sensing capabilities are being enhanced by new sensors such as Landsat ETM+, MODIS, and radar sensors; growing databases of multidecadal imagery such as that now present through the Landsat and AVHRR programs; new methods for information extraction and coupling of remote sensing–derived information with models; and greater data sharing efforts between North American and Russian organizations. • Considerable forest inventory and forest ecology data and forest science expertise are present in Russia. Coupling of inventory data with remote sensing observations significantly expands the information content of the remote sensing data, especially when forest age and cutting history is determined from inventory data. • Results of scientific research on disturbance, including fire behavior and the interaction of logging and fire in the Russian boreal forest, is contributing to a greater understanding of the role of Russian forests in global carbon exchange. Analysis of natural and experimental fires is contributing to the implementation of remote sensing–based algorithms for mapping and monitoring fires. • The link between forest dynamics and socioeconomic factors is now being integrated. The NASA LCLUC projects have fostered growing international collaboration on the individual, institutional, and national government levels. 40

Journal of Forestry

•

June 2003

Collaborations at these several scales are now making it possible for Russian and US scientists to work together to further our knowledge on the influence of land-cover and land-use change on the global boreal forest. Literature Cited ALEXEEV, V.A., and R.A. BIRDSEY. 1998. Carbon storage in forests and peatlands of Russia. Washington DC, USDA Forest Service. BAKER, S.P., and W. HAO. 2002. Emissions of trace gases from experimental fires in central Siberia. Paper presented at American Geophysical Union (AGU) Spring Meeting, Washington, DC. BERGEN, K.M., T. ZHAO, et al. (2003). Forest dynamics in the East Siberian boreal forest: Analysis using timeseries statistical and satellite data. Paper presented at annual conference, International Association of Landscape Ecology, Banff, Canada. BEUKEMA, S.J., and W.A. KURZ. 1998. Vegetation Dynamics Development Tool User’s Guide, Version 3.0. Vancouver, BC, Canada: ESSA Technologies. BLAM, Y., L. CARLSSON, et al. 2000. Institutions and the emergence of markets—Transition in the Irkutsk forest sector. Laxenburg, Austria: International Institute for Applied Systems Analysis. BONDAREV, A. 1990. Dynamics of forest regeneration in the Central Pryangar’ye. Forest Inventory and Forest Regulation 94–100. CANADIAN FORESTRY SERVICE. 1987. Tables for the Canadian Forest Fire Weather Index System. Forestry Technical Report 25. Ottawa. CONARD, S.G., AND G.A. IVANOVA. 1997. Wildfire in Russian boreal forests—Potential impacts of fire regime characteristics on emission and global carbon balance estimates. Environmental Pollution 98(3): 305–13. CONARD, S.G., A.L. SUKHININ, et al. 2002. Determining effects of area burned and fire severity on carbon cycling and emissions in Siberia. Climatic Change 55:197–211. DLUGOKENCKY, E.J., B.P. WALTER, et al. 2001. Measurements of an anomalous global methane increase during 1998. Geophysical Research Letters 28:499–502. FORSTER, C., U. WANGDINGER, et al. 2001. Transport of boreal forest fire emissions from Canada to Europe. Journal of Geophysical Research 106(22):22887–906. FROMM, M., J. ALFRED, et al. 2000. Observations of boreal forest fire smoke in the stratosphere by POAM III, SAGE II, and lidar in 1998. Geophysical Research Letters 27:1407–10. HOUGHTON, R.A., J.L. HACKLER, et al. 1999. The US carbon budget: Contributions from land-use change. Science 285:574–78. HSU, N.C., J.R. HERMAN, et al. 1999. Satellite detection of smoke aerosols over a snow/ice surface by TOMS. Geophysical Research Letters 26:1165–68. KASISCHKE, E.S., K. BERGEN, et al. 1999. Satellite imagery gives clear picture of Russia’s boreal forest fires. EOS Transactions 80(13):141,147. KASISCHKE, E.S., and L.M. BRUHWILER. 2002. Emissions of carbon dioxide, carbon monoxide and methane from boreal forest fires in 1998. Journal of Geophysical Research 108(D1): art.no. 8146, December 6. KHARUK, V.I., A.G. KOZUKHOVSKAYA, et al. 2001. Siberian silk moth outbreak monitoring based on NOAA/

AVHRR images. Russian Journal of Remote Sensing 1:80–86. KHARUK, V.I., K.J. RANSON, et al. In press. Landsat based analysis of insect outbreaks in southern Siberia, 2003. Canadian Journal of Remote Sensing. KOBAK, K.I., Y.A. KUKUEV, et al. 1999. The role of forests in changing the carbon content in the atmosphere (example from Leningrad oblast). Lesnoye Khoziajstvo 5:43–45. KOUTZENOGII, K.Y., V. SAMSONOV, et al. 2002. Ecological consequence of wildland fire in Siberian boreal forest. In Proceedings, Macro and Trace Elements Workshop, Jena, Germany. KUKUEV, Y.A., O.N. KRANKINA, et al. 1997. The forest inventory system in Russia: A wealth of information for western researchers. Journal of Forestry 95(9): 15–20. MCRAE, D.J., S.G. CONARD, et al. 2002. FIRE BEAR project: A ground-based validation and remote-sensing project for forest fire assessment in Russia. Paper presented at American Geophysical Union (AGU) Spring Meeting, Washington, DC. RANSON, K.J., G. SUN, et al. 2001. Characterization of forests in Western Sayani Mountains, Siberia from SIR-C SAR data. Remote Sensing of Environment 75(2):188–200. SHUGART, H.H., R. LEEMANS, et al., eds. 1992. A systems analysis of the global boreal forest. New York: Cambridge University Press. SHVIDENKO, A., S. NILSSON, et al. 1996. Forest management and the global carbon cycle: Carbon budget of the Russian boreal forest: A systems analysis approach to uncertainty. In Forest ecosystems, forest management and the global carbon cycle, eds. M.J. Kapps and D.T. Price, 142–62. Berlin: Springer-Verlag. SPICHTINGER, N., M. WENIG, et al. 2001. Satellite detection of a continental-scale plume of nitrogen oxides from boreal forest fires. Geophysical Research Letters 28(24):4579–82. SUN, G., K.J. RANSON, et al. 2002. Radiometric slope correction for forest biomass estimation from SAR data in the Western Sayani Mountains, Siberia. Remote Sensing of Environment 79:279–87. WOTAWA, G., P.C. NOVELLI, et al. 2001. Inter-annual variability of summertime CO concentrations in the Northern Hemisphere explained by boreal forest fires in North America and Russia. Geophysical Research Letters 24:4575–79.

Kathleen M. Bergen (kbergen@umich. edu) is assistant research scientist, School of Natural Resources and Environment, and faculty research associate, Center for Russian and East European Studies, University of Michigan, 430 East University, Ann Arbor, MI 48109-1115; Susan G. Conard is national program leader for fire ecology research, USDA Forest Service, Washington, DC; R.A. Houghton is senior scientist, Woods Hole Research Center, Woods Hole, Massachusetts; Eric S. Kasischke is associate professor, Department of Geography, University of

Maryland, College Park; Vyacheslav I. Kharuk is professor, Sukachev Institute of Forest, Krasnoyarsk, Russia; Olga N. Krankina is assistant professor, Department of Forest Science, Oregon State University, Corvallis; K. Jon Ranson is research scientist, Goddard Space Flight Center, Greenbelt, Maryland; Herman H. Shugart is Corcoran Distinguished Professor, Environmental Sciences, University of Virginia, Charlottesville; Anatoly I. Sukhinin is head, Remote Sensing Facility, Sukachev Forest Research Institute, Krasnoyarsk, Russia; Rudolf F. Treyfeld is chief engineer, Northwestern State Forest Inventory Enterprise, St. Petersburg, Russia.

June 2003

•

Journal of Forestry

41