B51B-0942 0800h
Mapping land cover change in South America using MODIS data from 2000-2004
Data from the MODerate Resolution Imaging Spectroradiometer (MODIS) sensor represent a unique information source for mapping land cover and land cover change. Data from MODIS exist at 250m, 500m and 1km spatial resolutions, with nearly daily global coverage. The finer spatial detail of MODIS compared to heritage global coverage instruments, such as AVHRR or VEGETATION, allows for the reasonable expectation of using the data to monitor land cover change dynamics. One of the standard products of the MODIS land science team is the Vegetation Continuous Field (VCF) data suite. These map layers represent percent cover maps of basic vegetation traits such as tree cover and crop cover. Because the VCF layers are mapped at the sub-pixel scale, successive depictions allow for the identification of land cover change. The current method uses a difference imaging approach with both thematic and spectral change indices. The combination of both inputs aids in limiting errors of commission in identifying change sites. Initial results for South America for two layers, percent tree cover and percent broadleaf crops, reveal a dynamic landscape with significant clearing on an annual basis across the continent. Results show a dramatic increase in clearing in the dry forest (largely Mato Grosso state) region of Brazil from 2002 to 2003. While the deforestation dynamic in the Legal Amazon as a whole is better understood, clearing of cerrado and chaco woodlands is comparable in scale to the area cleared within the forest domain. Spatially explicit change maps for woodland and parkland settings on an annual basis are new and should lead to a greater understanding of land use dynamics within these areas. Intensive agricultural use of the land follows much of the clearing episodes, with a shorter turnaround time from clearing to planting in the woodland settings. Final results are currently being tabulated to identify changed area per country, state and province, and across ecoregions for tree cover removal and agricultural expansion for the years 2000 to 2004.
B51B-0943 0800h
Changes in active layer ion distribution after short- and long- term forest disturbances, Eastern Siberia
Two intact Larch forests sites, two additional forests sites disturbed by severe fire and clear-cut respectively and a grassland site known as _eAlas_f (disturbed thousands of years ago) were soil sampled in order to assess changes in active layer ion distribution and in the underlying permafrost layer. Electric conductivity of the extract (ECe) of the intact soil in the forest revealed higher concentrations downward from the soil surface into the active layer and the underlying permafrost when the soil was silt-clay-loam with a maximum value of 4.7 mS cm$^{-1}$ at 2.1m, compared to lower values (less than 0.5 mS cm$^{-1}$) in the upper layers. In contrast, when the soil was sandy, ECe in the active layer did not exceed values higher than 0.5 mS cm$^{-1}$. Ion concentration was higher in the upper layers of the burnt site compared to the clear-cut site due to differences in vegetation. The ECe in the active layer of the Alas site between 30 and 80cm was high with values exceeding 4 mS cm$^{-1}$.The concentration of ions, especially Na$^{+}$, Mg$^{2+}$, Ca$^{2+}$, SO$_{4}$$^{2-}$as well as HCO$_{3}$$^{-} in the upper layers of this long-term disturbed site not only indicates the upward movement of ions but also shows a stronger process of alkalinization over time. Based on the movement of ions from lower to upper layers and the difference in soil texture in forest soils, the adverse implications of tree roots eventually uptaking water from thawing permafrost during dry years is discussed.
B51B-0944 0800h
An approach to decision aid of boreal forest fire control using both of ground observation and remote sensing
Burned area of boreal forest fires is increasing in these decades. Two thirds of forest fires are judged as man-made in Siberia. On the other hand, for boreal forest fire emits global warming gas due to combustion and to change of land coverage, forest fire may accelerate global warming. In 2003 summer, 17million hectares are burned in Siberia and CO2 emission is estimated as 3 hundred million tons. Thus, it is important to control forest fire. Toward this aim, we collected data of boreal forest fire in Alaska and east Siberia in summer fire seasons for two years. Data were acquired from each of ground observation, observation from aircraft and remotely sensed fire detection in June and July. Remotely detected fire using some algorisms were compared with observed data to evaluate the accuracy and earliness of automatic detection. Study areas are Alaska and East Siberia in this year and squares of 1000km centered on Yakutsk, Irkutsk and Krasnoyarsk for each in 2003. Daily NOAA and MODIS satellite images are corrected and used for fire detection. 750 ground observation reports are corrected from Russian agency including location, weather and fire front size and severity. 178 reports are corrected from JAL aircraft flying across Siberia including location and time. Comparison between ground truth data and satellite images was done for validation of automatic forest fire detection. Almost all location of ground and aircraft observation data of forest fires as large as 1 hectare were automatically detected at almost same time using satellite images where whether permitting. We are developing connection of fire detection algorithm and fire expansion simulation model to forecast the possible burned area. On the basis of fire expansion forecast, risk analysis of possible fire expansion for decision aid of fire-fighting activities will be analyzed.@@On the basis of these analyses, we will discuss some possible utilizations of remotely sensed forest fire to control them.
http://bffci.net/
B51B-0945 0800h
Human Impacts on Wildfires in Interior Alaska
The effects of human activities on the fire regime of high latitude ecosystems, which has not been well investigated, has the potential to influence water, energy, and carbon dioxide exchange with the atmosphere by influencing land cover and ecosystem dynamics. In this study we assessed the potential footprint of human presence on fire regime in Interior Alaska by investigating three research questions: 1) Does the type of fire ignition (human or lightning) have a significant impact on fire size?; 2) Does human impact on fire regime vary with population size?; and 3) Does distance from towns, roads or rivers affect fire size and ignition? To evaluate these questions, we overlaid the large-firescar database (fires $<$0.4 km2 for 1988-2002) and the fire ignition database (1956-2000) of the Alaska Fire Service with towns (all named settlements), major roads, and major rivers in Interior Alaska. Currently, humans are responsible for high fire frequency near towns and roads; however, human caused fires are generally much smaller than lightning ignited fires. Human impact on fire regime is a function of town size, and distance to roads and to a lesser extent rivers play an important role as they allow humans access to remote areas. Thus, it is clear that human activities influence fire regime in localized areas of Interior Alaska. Our next challenge is to evaluate if these effects of humans on the fire regime influence water, energy, and carbon dioxide exchange at the regional scale.
B51B-0946 0800h
Large Scale Mapping of Disturbance and Recovery Using Fusion of Radar and Lidar: An Assessment
Forested landscapes are generally composed of a heterogeneous mixture of patches that reflect the complex interaction of processes occurring at many spatial and temporal scales. Whether caused by natural disturbances, such as blow downs and fire, or management practices, such as logging and agriculture, ecosystem structure and carbon fluxes will vary strongly as a result of differences in successional stage. Identifying and tracking these dynamics through space and time has been extraordinarily difficult, given the burden and limited scope of field-based methods, and the limited efficacy of most remote sensing approaches. Lidar remote sensing has proven to be exceptionally effective for mapping forest structure and successional status, providing critical initialization for carbon modeling approaches. However, application of lidar-based methods has been hindered by the relative expense and sparse spatial coverage of existing data sets. Regional and continental scale studies may thus require the integration of sparse lidar observations with data from sensors that may reasonably be expected to provide continuous coverage of some aspect of forest structure at high spatial resolution over large areas. One such technology is radar remote sensing. In this paper we assess the potential of integrating p-band radar with waveform lidar for the mapping of disturbance and recovery. Lidar data acquired over the tropical forests of La Selva, Costa Rica, are used to derive canopy height, biomass, and forest vertical structure. These data are then combined with p-band radar acquired several years later to develop and evaluate approaches to fusion that facilitate the determination of ecosystem dynamics.
B51B-0947 0800h
Modeling ecosystem disturbance over regional scales with the Ecosystem Demography model
The Ecosystem Demography (ED) model is a mechanistic individual-based terrestrial ecosystem model that is capable of explicitly incorporating ecosystem disturbance and recovery in regional and continental scale simulations. The ED model is a momentum approximation of a traditional `gap model', a stochastic individual model of forest dynamics. As such, ED models both the age and the size structures of the terrestrial ecosystems it simulates. The age structure directly accounts for the distribution of ages since last disturbance in each simulation grid cell, and captures the disturbance impact on the ecosystem's carbon stocks. The size structure is used to model the impact of different kinds of disturbance on ecosystem dynamics and the recovery processes following disturbance. We present a set of ED simulations of the carbon dynamics of New England for the period 1700-2100 that account for both natural and anthropogenic ecosystem disturbance. We simulate tree-fall, land-use change, forest harvesting, and the impact of the Hemlock Wooly Adelgid, an introduced pathogen of eastern hemlock. While all these disturbances involve tree mortality, their simulated impact on the carbon dynamics is substantially different depending on the timing, frequency, and selectivity of the disturbance. In particular the temporal patterns of net ecosystem productivity during ecosystem recovery from Hemlock Wooly Adelgid infestation is not only influenced by the amount of hemlock present, but also by the ecological role that hemlock played in that specific ecosystem, and by the species that replace it after removal.
B51B-0948 0800h
Historical Changes in Carbon Storage of the Eastern United States: Uncertainties Associated With Forest Harvest and Agricultural Activities
We conducted simulations with the Terrestrial Ecosystem Model (TEM) to estimate historical changes in carbon (C) storage of the eastern United States (US) and to evaluate uncertainties in these estimates associated with forest harvest and agricultural activities. The use of alternative data sets for forest harvest since 1600 resulted in different trajectories of C storage dynamics prior to the middle of the 20th Century, but these differences in carbon storage disappear by the 1990s under the same scenarios of harvest activity during the second half of the 20th Century. In contrast to the aforementioned convergence, the TEM simulations revealed large uncertainties in the latter half of the 20th Century associated with the growth of forests after agricultural abandonment. The variability in growth of forests after agricultural abandonment in the TEM simulations primarily depended on the degree to which soil nitrogen (N) was depleted by historical agricultural activities. In the northeastern US, annual changes in vegetation C storage increased from 22 to 34 Tg C per year from 1988 to 1992 between simulations assuming maximum vs. minimum depletion of soil N. In the southeastern US, annual changes in vegetation C storage from 1988 to 1992 decreased by 5 Tg C per year in the maximum-depletion simulations and increased by 16 Tg C per year in the minimum-depletion simulations. Our analyses suggest that historical land-use practices may be affecting the current trajectories of C storage dynamics in the US. Analyses of future forest carbon sequestration may need to consider the legacy of this influence as well as interactions between agricultural land use and forest regrowth.
B51B-0949 0800h
Using Landsat Imagery to Detect Forest Disturbance in the Pacific Northwest
A spatially and temporally accurate record of forest disturbance is essential in regional carbon accounting. Forests of the Pacific Northwest (PNW) have the potential to store large amounts of carbon, leading to significant carbon losses at the time of disturbance and significant sequestration during subsequent regeneration. Landsat data have been used to comprehensively map stand-replacing harvests and fires in western Oregon and Washington from 1972 to 2002. This map product has been used as input to several carbon accounting and process models operating at the regional scale, including Landcarb, Biome-BGC, and C-Quest. The disturbance data were also used to support evaluation of the effects of Northwest Forest Plan. Efforts to automate the mapping process led to the development of the Disturbance Index, a transformation of Landsat data that facilitates the identification of disturbed forest through multi-temporal composite analysis. We are also developing methodology to map partial canopy removal on a regional scale. Shifts in forest management practices over the last decade have led to a greater role for partial harvesting in the PNW, and we are completing a project that maps canopy removal as a continuous variable over several thousand hectares of forest in central Washington. This process involved the use of data from aerial photos, management records, and field plots in the interpretation of biennial Landsat imagery. Our continually updated and improved disturbance maps, showing the history of stand-thinning and stand-replacing disturbance in the PNW region, are enabling more realistic and spatially explicit accounting of the region's carbon stores and emissions.
B51B-0950 0800h
Mapping and Monitoring Boreal Forest Regrowth Dynamics using Satellite Data Products
We used field measurements of canopy light interception (Fpar) and indirectly estimated LAI at numerous burn regrowth sites in interior Alaska to assess satellite data products of these same variables. Fire severity varied across the range of burn sites depending on a number of conditions, and aspen, willow, spruce and herbaceous vegetation has regrown in heterogeneous mixtures, heights and densities. Fpar and LAI estimates were made using handheld instruments in order to characterize spatial and temporal variability in regrowth and to produce Fpar maps using IKONOS and Landsat imagery. A set of custom manufactured solar cells also operated nearly continuously through the active growing seasons. A sensitivity analysis of the indirect canopy LAI estimates with different instruments revealed those factors most critical to accurate retrievals, and allowed us to provide error estimates. Substantial temporal and spatial variability in Fpar across the sites and through time was largely captured by the MODIS Fpar products, but there were notable differences associated with vegetation cover types, ground cover and regrowth stage. There were also differences in vegetation indices based on the higher resolution maps. The MODIS tree cover ("continuous fields") products were effective at capturing differences in burns of various ages across interior Alaska. We also analyzed two AVHRR time series, extracting 10-day Fpar values for the larger burn scars within a Canadian fire database and comparing anomalies with unburned areas in the same ecoregions. Results show significant fire-related anomalies, and associated recovery trajectories. The linkages between boreal fire disturbance and associated regrowth carbon dynamics have direct relevance to numerous ongoing research programs, including the USGCRP North American Carbon Program.
http://www.whrc.org
B51B-0951 0800h
Scale Impacts in Net Ecosystem Productivity Estimations
Net ecosystem production (NEP) estimations play a key role in the terrestrial carbon cycle assessment, both at regional and global scales studies. The emergence of remote sensing greatly improved NEP estimation methods and analysis domain. Yet, spatial and temporal resolution of sensors and remote sensing products often imply adjustments to NEP calculation methods. The Carnegie Ames Stanford Approach (CASA) terrestrial biogeochemical model (Potter et al., 1993; Friedlingstein et al., 1999) simulates plant and soil processes allowing the estimation of NEP through the difference between net primary productivity and soil respiration. CASA inputs include climatic data: precipitation, temperature and solar radiation; soil texture; vegetation type and percentage cover; as well as leaf area index (LAI), fraction of photosynthetically active radiation absorbed by vegetation (FPAR) and normalized difference vegetation index (NDVI). With a research interest in regional vegetation dynamics in the Iberian Peninsula (IP), estimations of NEP were compared with local measurements over a Quercus ilex and Quercus suber with perennial grassland ecosystem, representing a region characteristic land cover. The CASA calibration process aimed the tuning of efficiency scalars directly related to net primary productivity and soil respiration calculations, maximum light use efficiency (*) and temperature effect on soil fluxes (Q10). To this end local weather station data was used as climatic inputs, with remotely sensed LAI, FPAR and NDVI products from MODIS sensor. In a first approach the NEP calculations were performed at a finer spatial and temporal resolution of 1 km and 8 days, respectively, for the periods of 2002 and 2003 (years of available NEP measurements). A confident correlation is found, although local extremes tend to differ and affect the annual balance concordance between estimations and measurements of NEP. Consequently, calibrated * and Q10 values were used at coarser temporal and spatial resolutions, whose varying correlation results will be presented, as well as results for the scalars calibration at different temporal and spatial scales. Accordingly, NEP modeling results for the IP are presented, as well as an analysis of the impact of temperature and precipitation fields in the dynamic trends of vegetation spatial patterns and intra-annual behavior.
B51B-0952 0800h
Detection of human induced degradation of dry lands at regional scales
Numerous reports over the past 20 years have disputed the extent of desertification at local, regional and global scales, and there is a pressing need for an objective, repeatable, systematic and spatially explicit measure of human-induced land degradation in global dry lands. Our recent studies in southern Africa have made use of the heavily populated areas known as homelands or communal lands that are indisputably degraded in order to test a series of hypotheses about the existence of a distinct state of degradation that might justify the term "desertification". These hypotheses concern the spatial extent and temporal persistence of degradation and the fundamental significance of primary production. Satellite monitoring of primary production has been used as a tool to address these hypotheses regarding desertification and to map large, affected regions. The primary production of non-degraded land is affected by rainfall, soils, topography, human management, and land use, but regional databases of these factors are either generally not available or have very coarse spatial resolution. We explore here various procedures that hold promise for controlling these factors and allowing reductions in productivity caused by humans to be detected. These methods are applied to sites in southern Africa that are known from extensive field surveys to be degraded as a result of human utilization.
B51B-0953 0800h
Estimation of Community Metabolisms in Coral Reef by Using the Water Flow Model
Global environmental change has been a great issue for the past several decades. Coral reefs are under a serious threat of degradation in the world ecosystems. Coral reefs have a biological diversity of benthic organisms and are characterized by the high productivity of community metabolisms. Photosynthesis-respiration and calcification are the main processes of the metabolisms controlling carbonate system in reef water. To prevent coral reefs from the degradation, it is necessary to observe the long-term community metabolisms that can be used for detecting initial irregular change in coral reef ecosystem. However carbonate system is affected by the input and mixing of ground water and outer reef ocean, and the increasing CO$_{2}$ from the atmosphere. For the purpose of long-term monitoring of community metabolism, we introduce a model to estimate the daily organic and inorganic carbon production under the natural water flow condition. The observation has been carried out at the edge of a pier of Sesoko Station, Tropical Biosphere Research Center, University of the Ryukyus, Okinawa, Japan. Dissolved oxygen (DO), pH, temperature, salinity and water current were monitored every 15 minutes for over a year. Total alkalinity was measured every 3 weeks by the Gran titration. The model to estimate the carbon productions was constructed using the difference of concentration between outer and inner reef water, and residence time of the reef water that was derived from water current. The community metabolisms calculated from the carbon productions of the model were within the range of the other coral reefs in the literature. This result indicates that the model can be used for the estimation of community metabolisms of natural reef environment.