B54A-01 INVITED 16:00h
Large-Scale Disturbance Events in Terrestrial Ecosystems Detected using Global Satellite Data Sets
Studies are being conducted to evaluate patterns in a 19-year record of global satellite observations of vegetation phenology from the Advanced Very High Resolution Radiometer (AVHRR), as a means to characterize large-scale ecosystem disturbance events and regimes. The fraction absorbed of photosynthetically active radiation (FPAR) by vegetation canopies worldwide has been computed at a monthly time interval from 1982 to 2000 and gridded at a spatial resolution of 8-km globally. Potential disturbance events were identified in the FPAR time series by locating anomalously low values (FPAR-LO) that lasted longer than 12 consecutive months at any 8-km pixel. We can find verifiable evidence of numerous disturbance types across North America, including major regional patterns of cold and heat waves, forest fires, tropical storms, and large-scale forest logging. Based on this analysis, an historical picture is emerging of periodic droughts and heat waves, possibly coupled with herbivorous insect outbreaks, as among the most important causes of ecosystem disturbance in North America. In South America, large areas of northeastern Brazil appear to have been impacted in the early 1990s by severe drought. Amazon tropical forest disturbance can be detected at large scales particularly in the mid 1990s. In Asia, large-scale disturbance events appear in the mid 1980s and the late 1990s across boreal and temperate forest zones, as well as in cropland areas of western India. In northern Europe and central Africa, large-scale forest disturbance appears in the mid 1990s.
http://geo.arc.nasa.gov/sge/casa/
B54A-02 16:15h
LEDAPS: A North American Disturbance Record from Landsat Imagery
The disturbance regime of forests exerts a strong control on North American terrestrial carbon dynamics. Disturbance events themselves (fire, harvest, insect damage) emit carbon directly to the atmosphere; regrowth following disturbance tends to sequester carbon from the atmosphere. In general, the spatio-temporal patterns of disturbance control the age structure and successional patterns of forests, and thus influence net ecosystem productivity on a regional basis. The Landsat remote sensing program has collected high-resolution (30-80 meter resolution) imagery since 1972, and provides an excellent tool for assessing recent disturbance patterns. To date, however, this archive has not been comprehensively analyzed for such information. To remedy this, the LEDAPS (Landsat Ecosystem Disturbance Adaptive Processing System) project at NASA GSFC is processing decadal Landsat imagery centered on 1975, 1990, and 2000 epochs to map forest disturbance rates and patterns across North America. Landsat imagery is calibrated and atmospherically corrected to surface reflectance using the 6S radiative transfer model. Disturbance patterns are extracted in two phases. First, a simple Disturbance Index (S. Healey, personal communication) is used to map the location of significant biomass loss following a `tasseled cap' radiometric transformation. Second, we are experimenting with the integration of canopy reflectance models to relate pixel radiometry directly to stand attributes (canopy cover, forest type) to better characterize regrowth state. Initial "Beta" products have been released and feedback from the land science community is invited. This presentation will give an overview of the LEDAPS project, and present initial results from reflectance processing and disturbance mapping.
B54A-03 16:30h
Beyond Potential Vegetation: Combining Lidar Remote Sensing and a Height- Structured Terrestrial Ecosystem Model for Improved Estimates of Carbon Stocks and Fluxes at a Set of Sites in North and Central America
Because of natural disturbance events, human land use, and land-use history, terrestrial ecosystems globally are generally not in a "potential" state. However, the information required to adequately describe the current state of terrestrial ecosystems is lacking and currently limiting our understanding of the carbon cycle. To address this challenge, we combined airborne lidar remote sensing of vegetation structure and the height-structured terrestrial ecosystem model ED to produce lidar-initialized model estimates of ecosystem structure, carbon stocks, and carbon fluxes at a set of 10 study sites in North and Central America. Field data were used to test model results and form the basis for improved model parameterizations. Resulting lidar-initialized ED estimates of ecosystem structure and above-ground biomass compared favorably to field-based estimates at key study sites, and corresponding model estimates of net carbon fluxes differed substantially from estimates based on bracketing alternatives. The results of this multi-site study build on earlier published results from La Selva, Costa Rica and provide additional evidence of the power of combining remote sensing data on vegetation structure with a height-structured ecosystem model to go beyond potential vegetation and address the heterogeneity in terrestrial ecosystems caused by disturbance. Extending these analyses to larger scales will require the development of regional and global lidar data sets, and the continued development and application of height-structured ecosystem models.
B54A-04 16:45h
Contributions of Land Inventory and Biometrics for Characterizing Disturbance in Ecosystem and Carbon Accounting Models
Nearly all forest lands of the U.S. are disturbed or affected by disturbance. Each decade about half of the total forest land area is disturbed by harvesting, grazing, wildfire, pests, and other natural causes. Because of the patchy nature of forest disturbances and the amount of edges, a large proportion of undisturbed interior forest area is affected by disturbances. Some ecosystem and carbon accounting models require spatially explicit input data about frequency or effects of disturbance. Current and historical information about forest disturbances includes both geospatial and statistical data. Examples of geospatial data sets from land inventories include maps of insect defoliation and N deposition. Examples of primarily statistical data sets that can be made into geospatial data sets include county-level statistics from forest inventory and other census data about land use or cover. A variety of techniques are available to manipulate statistical data sets into geospatial data sets. Many ecosystem and carbon accounting models lack ability to simulate the dynamics of disturbance and instead only represent potential forest vegetation. Several techniques are being developed in large-scale ecosystem models to address this issue. One approach is illustrated by Production Efficiency Models (PEMs) that use satellite-derived information to estimate vegetation productivity and C changes affected by landscape changes and climatic variability. Models that rely primarily on remote-sensing information lack ability to separate kinds of disturbances that may have similar canopy impacts, such as forest management and land use change. Also, PEMs lack the capacity to detect impacts of global change stressors such as CO2, N deposition and ozone. Another approach is to combine disturbance models with ecosystem models. Such model combinations use different integration approaches. Whether using ecosystem models to provide growth information to parameterize disturbance models, or using disturbance models to pass the age cohort information to ecosystem models, ground-based disturbance information is required to verify remote sensing information and initialize succession stages. Besides spatially explicit characterization of disturbance, accurate characterization of forest processes in ecosystem and carbon accounting models requires parameters from disturbed sites. However, study areas with sufficient data available for model parameterization are typically undisturbed interior forest. Thus there is a great need for establishment of intensive monitoring in landscapes that include disturbances. Protocols for landscape-scale intensive monitoring are being developed and applied in a pilot mode under the North American Carbon Program.
B54A-05 17:00h
Historical Land Use Modeling of Agriculture and Forestry
We developed a land use model that converts native ecosystems to agriculture, harvests forests with and without land use conversion, and tracks age cohorts following harvest and cropland abandonment annually from 1600 to 2002. These land use data sets were used to drive the Terrestrial Ecosystem Model (TEM) to estimate historical changes in carbon storage and to evaluate uncertainties in these estimates associated with forest harvest and agricultural activities. The land use model integrates temporally and spatially explicit agricultural land use information from Ramankutty and Foley with temporally and spatially explicit estimates of forest area disturbed through harvest within 0.5 degree by 0.5 degree grids. Because little inventory data on forest area harvested is available, we developed a statistical model to estimate forestland harvested over the 1952 to 1997 based on available inventory data. From 1600 to 1952, we assumed no harvest disturbance until the first permanent settlement, and interpolated the 1952 statistical estimate to the date of first settlement, modifying interpolated estimates by human population trends. Total US forestland area dynamics over 1600-2002 parallel inventory trends and are within 6 to 10 percent of inventory estimates. The model slightly overestimates total US forestland harvested over the 1980-1990 period (2.12 percent versus inventory estimate of 2.0 percent). At the regional level, the inventory estimate of southeastern forestland harvested (2.51) was less than the model estimate (2.85) and in the Northeast, the model estimate (1.86 percent) slightly over predicted actual harvest area (1.95 percent). When the estimated cohort distributions are compared with an independent data set on forest age class distributions, the model generally overestimates the youngest and the oldest age classes as a percent of the total. For example, the model estimate for 0-19 years US total is 28 percent versus 18 percent in the inventory, and the very old age classes, 23 versus 4 percent inventory estimate. In regions where timber harvest is the predominant disturbance, age class comparisons are closer; the model estimated 20.2 percent in the 40-59 age class in the Southeast versus the inventory estimate of 21.8 percent. Uncertainties include fire as a disturbance, assumptions about harvest methods, and the relationship between wood use and population numbers.
B54A-06 17:15h
Simulating Disturbance Impacts With the Carbon Budget Model of the Canadian Forest Sector (CBM-CFS)
The Carbon Budget Model of the Canadian Forest Sector (CBM-CFS) is a forest carbon (C) accounting system that estimates C stocks and stock changes in biomass and dead organic matter pools. The model operates at four spatial scales from stand-level to national. Recent advances have improved the generalized framework for the simulation of disturbances such as fires, forest insects, and management activities by expanding the disturbance impact types that can now be represented. The new approach accommodates all types of impacts from growth reduction to partial mortality to stand replacement, and these can span single- to multi-year periods. The amount of disturbance can be defined by area, proportion of eligible area, or C removed. Their impacts are modeled as the immediate effect on C stocks (disturbance matrices define the transfer, release and export of C) and on growth dynamics. The model simulates the post-disturbance growth trajectory and dead organic matter dynamics of the stand. Disturbance data obtained from monitoring programs are combined with inventory and growth and yield data to estimate past C stock changes. Projections of future natural disturbances, based on probability distribution functions, can be used to estimate the distribution of future C stock changes. Example simulations demonstrate the approach for different disturbance types and regions.
http://carbon.nrcan.gc.ca
B54A-07 17:30h
Variations in the fire regime in the North American boreal forest between 1990 and 2004 and their potential impacts on terrestrial carbon storage
Fires in the North American boreal region play an important role in regulating the the levels of carbon stored in the terrestrial ecosystems of this region, both directly and indirectly. Biomass burning not only consumes carbon present in the aboveground vegetation and litter layers of boreal forests and peatlands (as is common during fires in temperate and tropical ecosystems), but also large amounts of carbon present in the organic layer that lies on top of mineral soil (consisting of moss, lichen, dead woody debris and organic soil). Understanding the factors controlling consumption of ground-layer organic matter during fires in boreal ecosystem is central to quantifying the terrestrial carbon budget in this region. The 1950-2004 period can be divided into 3 distinct epochs in terms of fire activity in the North American boreal region. The early epoch of 1950-1968 experienced the lowest fire activity, 1.2 million ha or Mha per yr, and increased to 2.1 Mha per yr during 1969-1986 epoch and 3.0 Mha per yr during 1987-2004 epoch. The end result of this steep rise in fire activity is an increase in the average amount of carbon released during fires. A key question that needs to be addressed is how much carbon has actually been released through the burning of ground-layer organic matter. The observed increases in average area burned are due to a combination of increases in the frequency of large fire years, as well as increases in average area burned during large fire years. Analyses of fire databases show that as the burned area increases during a given year, the percent of area burned in large fire events increases as well. The amount of fires occurring later in the growing season also increases. Recent and ongoing studies have integrated field observations with satellite observations on fire location and fire severity to provide more detailed assessments of how fires impact carbon budgets of boreal systems. These studies, along with theoretical models, indicate that organic layers common in boreal forests and peatlands burn deeper and release more carbon during large fire events and during late season fires. In this paper, we will present the results of a study on how recent changes in the North American boreal fire regime have influenced carbon storage in the ground-organic layer in this region based on different assumption regarding levels of consumption of organic layer burning.
B54A-08 INVITED 17:45h
Lessons from high-resolution satellite case studies for global-scale models of biomass burning: An approach for assessing the component of net deforestation linked with fire
Besides fossil fuel emissions, the largest anthropogenic flux of carbon to the atmosphere stems from tropical deforestation. One of the most important tools used to clear land is fire, but it remains uncertain what fraction of the net deforestation flux is lost as direct fire emissions as opposed to onsite or offsite decomposition of slash and non-merchantable timber. Most of the uncertainty can be attributed to the large temporal and spatial variations in burned areas, fuel loads, and how much of the fuel is actually combusted in the process of land clearing. This variability is especially large in areas undergoing rapid deforestation, where frequently burning pastures and grasslands often border fire prone closed canopy tropical forest. Here we investigate the links between net deforestation and fire in three active biomass-burning areas: the southern Amazon, southern Africa, and southern Borneo (Kalimantan). In each area, these links are explored using MODIS 250m satellite data of fire activity and vegetation characteristics. Results demonstrate the need to preserve spatial variability of fire processes when aggregating site level data to regional or global level; they also reveal how interannual variability in precipitation affects the rate of deforestation, which is then shown to be one of the main drivers of interannual variability in the growth rates of many atmospheric trace gases.