Quantifying Changes in Carbon Pools with Shrub Invasion of Desert Grasslands using Multi-Angle Data from EOS Terra and Aqua
Mark J. Chopping (Principal Investigator), Lihong Su, Montclair State University; John V. Martonchik, NASA/JPL; Albert Rango, Debra P. C. Peters, USDA, ARS Jornada Experimental Range
This work addresses new approaches to exploiting data from the Multi-angle Imaging Spectro-Radiometer (MISR) and Moderate Resolution Imaging Spectroradiometer (MODIS) for mapping desert grasslands. Significant progress has been made in geometric optical canopy modeling, with good simulations of 441 red band pixels in the 9 MISR views in a 5.25 km2 area in the Jornada Experimental Range (N = 3969, r2 =0.78). Regression of the four background sub-model parameters on the kernel weights of a LiSparse-RossThin bidirectional reflectance distribution function (BRDF) model and nadir camera multi-spectral reflectance provided accurate measures of the background contribution, allowing inversion of the model for fractional shrub cover, using numerical methods. Absolute root-mean square error between retrieved and measured values was 0.03 and ~90% of the estimates were within 0.05 of the true value for a wide range of canopy configurations. Research was also performed on the use of maximum likelihood and support vector machine (SVM) algorithms for classification of MISR and MODIS data sets to plant community types in the Jornada Experimental Range and the Sevilleta National Wildlife Refuge, New Mexico (19 classes). Half of the samples were randomly selected as the training set and the other half as the testing set. A total of 66 classifications were performed with various combinations of data sets: the r0, k, and b parameters of the MRPV BRDF model; the isotropic, geometric and volume scattering kernel weights of a Li-Ross BRDF model; the structural scattering index; and MISR and MODIS surface reflectance estimates. This research found that multi-angular observations, surface anisotropy patterns and SVM algorithms can improve desert vegetation type differentiation importantly. Using multi-angle data raised the overall classification accuracy from 45.4% for nadir observations to 60.9%, and with surface anisotropy patterns derived from MRPV and RossThick-LiSparse-Reciprocal BRDF models an overall accuracy of 67.5% can be obtained with a maximum likelihood classifier. Using a non-parametric SVM algorithm the classification accuracy was raised to 76.7%.
Community Type Classifications for the Sevilleta National Wildlife Refuge using MISR data and model-derived multi-angle metrics (a) 1998 NSF LTER Vegetation Map (b) Maximum Likelihood method (c) Support Vector Machine method. View image with legend.
Effects of Logging, Plantation Conversion, Biomass Burning and Regrowth on Carbon Dynamics in Bornean Peat and Dipterocarp Forests: Implications for Global Carbon Cycle
Lisa Curran (Principal Investigator), Yale University; Simon Trigg, University of Maryland; Daniel Nepstad, Richard Houghton, Woods Hole Research Center
Cumulative forest loss within the Gunung Palung National Park (GPNP) in West Kalimantan boundary (yellow) and its surrounding 10-km buffer. Forest and nonforest classification are based on a Landsat Thematic Mapper time series (1988, 1994, 1997, 1999, 2001, and 2002). Classifications are shown for (A) 1988, (B) 1994, and (C) 2002. Lowland (green) and peat (olive) forests were converted to nonforest (red), first predominantly in the buffer and later within the park. Grayareas are montane forest (66 km2 more than 500 m a.s.l.) and were excluded from analyses. The well-defined nonforest area that appears northeast of GPNP in (B) has been clear-felled for an oil palm plantation.Download higher resolution image.
This research focuses on biodiversity-rich Indonesian Borneo (Kalimantan), a significant terrestrial reservoir for atmospheric carbon. With Kalimantan currently undergoing rapid forest conversion and habitat degradation, this research aims to develop carbon models based on: (1) a regional-scale database that can be used to quantify variations in terrestrial carbon storage as a function of forest cover and land-use type, and (2) new regionally-specific approaches to map the extent of peat forest, oil palm plantations, and areas burned, as well as assessments of the spatio-temporal patterns of degradation and land-cover change. Database development has been furthered by the compilation of essential data layers, such as district-level GIS census data, GIS lithology data, detailed GIS land system (cover) data, and Landsat-derived 1990-era roads, 2000-era roads, and oil palm plantation extent.
The summer field campaign of 2005 yielded aerial surveys across many land cover types in western West Kalimantan, and field surveys that derived the key measures of peat depth and above and below-ground biomass. The above ground biomass estimates are being tested to derive empirical relationships that will permit the use of ETM+ and MODIS reflectance measurements to extrapolate above ground biomass across Borneo. The below-ground biomass estimates are being compared with peat phasic zones classified from ETM+ data (as illustrated) to test phasic zone information as a predictor of peat depth. A further project component is using ETM+ data coupled with field survey to assess the carbon implications of the widespread conversion of residual forest stands to oil palm plantations. Preparations are also underway for our experimental peat burn (planned for 2006). Relationships derived from field parameters will assist time-series analysis of change and carbon model development; regional analyses will serve as the basis for scaling up mapping and modeling efforts.
Northern Eurasian C-land use-climate interactions in the semi-arid regions
Dennis Ojima (Principal Investigator), Colorado State University; Xiangming Xiao, University of New Hampshire; Chuluun Togtoghyn, Mongolian National University; Sayat Temirbekov, Kazakhstan National Institute of Botany; Svetlana Nikulina, UNEP; Muhtor Nasyrov, Mardonov Bakhtiyor, Samarkand State University; Kanat Akshalov, Research Institute of Grain Farming, Kazakhstan
Analysis of NPP trends from remote sensing observations and FAO statistics indicate similar increase cropland NPP during the past 20 years. This result will enhance our analysis of land use impacts on NPP and ecosystem responses.
The Northern Eurasian (from the Black Sea to the Mongolian Plateau) land use history and importance is unique within the global environmental science framework. The land-use systems are the most diverse among the temperate ecosystems due to the fertile soils, mountain fed rivers, and range of climate. The region is noted for the rapid increase in surface temperature, approximately a 1oC increase during the past 50 years. The recent changes in the climate and factors affecting land-use decision in the region have led to changes in cropland abandonment, destocking of certain rangelands and increased stocking of others, degradation of soils due to salinization and desertification, and damage to wetlands due to modifications of water regime.
Carbon stores and fluxes through out the region have been modified through land-use change over the past decades. The objective of this study is to investigate the roles and consequences of changes in climate and land-use intensity on the land carbon stores and fluxes, water vapor exchange, and CO2 exchange of the steppes of Eurasia during the past 100 years. Recent analysis of land-use changes indicate that socio-political changes driving land-use change have affected carbon dynamics more that mere climate changes in the region. However, the sensitivity of the rangeland and abandoned cropland ecosystems to changes in climate and land-use may alter the rate of carbon stored or emitted from these ecosystems.
Selective Logging in the Brazilian Amazon
Gregory Asner Principal Investigator), David Knapp, Eben Broadbent, Paulo Oliveira, Stanford University; Michael Keller, USDA Forest Service/University of New Hamshire; Jose Natalino Silva, Embrapa Amazonia Oriental
Spatial distribution of selective logging in five timber-production states of the Brazilian Amazon for the year intervals 1999–2000 (red), 2000–2001 (blue), and 2001–2002 (green). The states of Amazonas (AM), Amapa (AP), Tocantins (TO), Maranhao (MA), and the southern nonforested part of Mato Grosso were not included in the analysis. Light gray areas show the extent of indigenous reserves; dark gray areas delineate federal conservation lands as of 1999 (Science vol. 310, p. 480-482, 21 October 2005). Download higher resolution image.
Amazon deforestation has been measured by remote sensing for three decades. In comparison, selective logging has been mostly invisible to satellites. This team developed the first large-scale, high-resolution, automated remote sensing analysis of selective logging in the top five timber producing states of the Brazilian Amazon. Data were combined from Landsat 7 ETM+, Terra MODIS, and EO-1 Hyperion sensors to achieve this goal. Selective logging areas in the Brazilian Amazon ranged from 12,075-19,823 km2 yr-1 (+14%) between 1999 and 2002, equivalent to 60-123% of previously reported annual deforestation or “clear-cut” area. Up to 1,200 km2 yr-1 of logging was observed on conservation lands. Each year 27-50 million m3 of wood were extracted and a gross flux of up to 0.08 Gt C was destined for release to the atmosphere by logging. More information on this project can be found at: http://asnerlab.stanford.edu/projects/ amazon_logging2/amazon_logging.shtml
Detecting and characterizing fires in the Brazilian Amazon
Ivan Csiszar (Principal Investigator), University of Maryland; Jeff Morisette, GSFC; Douglas Morton, Wilfrid Schroeder University of Maryland; Joño Pereira IBAMA, Brazil; Louis Giglio SSAI/GSFC
As part of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) program, this project focuses on quantifying the uncertainty in satellite derived fire and burn scar products in the Brazilian Amazon, concentrating primarily on products derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensors flown aboard the Terra and Aqua satellites. The project is a collaborative effort between the NASA Goddard Space Flight Center, the University of Maryland and the Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis (IBAMA).
The team is building on previous efforts by quantifying the spatial extent of fire in the Amazon and characterizing fire types based on the nature of land-use changes. The project is producing data and tools to determine where fire leads to land-cover change and where it maintains land cover in a state of equilibrium. A series of field campaigns in the Brazilian Amazon are providing coincident field, airborne, and Advanced Spaceborne Thermal Emission and Reflection (ASTER) sensor data to evaluate the quality of the MODIS fire product. Ground temperature measurements and fire perimeters from field and airborne data are compared with coincident satellite images to refine the detection algorithms. Characteristics of fires from MODIS are combined with time series of the MODIS vegetation index (VI) product to differentiate between “conversion” fires, the result of recent deforestation, and “maintenance” fires, used to rehabilitate degraded pasture areas or to clear woody material from agricultural fields. Quantifying the timing and area of new clearing and maintenance fires provides critical insight into carbon fluxes from land-use change in Amazonia. Analysis of NPP trends from remote sensing observations and FAO statistics indicate similar increase cropland NPP during the past 20 years. This result will enhance our analysis of land use impacts on NPP and ecosystem responses.
Reducing Uncertainties of Carbon Emissions from Land Use-Related Fires with MODIS Data: From Local to Global Scale
Ruth DeFries (Principal Investigator), Simon Trigg, Doug Morton, University of Maryland College Park; G.J. Collatz, NASA GSFC; J. Randerson, University of California Irvine; G. Van der Werf, USDA-FAS, NASA GSFC; Louis Giglio, SSAI, NASA/GSFC; Lisa Curran, Yale University
Tropical deforestation is a major source of carbon to the atmosphere, primarily through fire used to clear forests for cropland or pasture. Previous estimates of carbon emissions from fire are based on coarse resolution satellite data and do not account for varying fire regimes associated with different land uses or for variations in biomass. This project uses MODIS data and the CASA biogeochemical model at the MODIS 250m resolution in two test areas, each covering the extent of a MODIS tile (approximately 1000 x 1000 km). The test areas are the southern Amazon and Kalimantan, two regions of rapid land-use change where fire is used extensively for land management. Detailed analyses of these two test areas allows for assessment of sources of uncertainties in the coarser scale estimates. It also provides a means of partitioning carbon emissions from different land-use types, i.e. initial forest clearing vs. maintenance of previously cleared pasture or oil palm plantations. Using the high-resolution model results, it will be feasible to develop approaches to realistically scale up estimates of carbon emissions from land use-related fires to regional and global scales. The high-resolution model results also provide a basis for assessing emissions from possible future land-use trajectories in the rapidly-changing tropics.
Other Carbon/Biogeochemical Cycle LCLUC Projects:
- Conard, Susan - USDA Forest Service. Wildfire Impacts on Carbon Stocks and Exchanges in Forests of Central Siberia: Quantifying Effects of Fire Intensity, Fire Severity, and Burning Conditions
- Lettenmaier, Dennis - University of Washington/JPL. Diagnosis and Prognosis of Changes in Lake and Wetland Extent on the Regional Carbon Balance of Northern Eurasia
- Li, Changsheng - University of New Hampshire. Quantifying CO2 Fluxes from Boreal Forests in Northern Eurasia: An Integrated Analysis of Flux Tower Data, Remote Sensing Data and Biogeochemical Modeling
- Nepstad, Daniel - Woods Hole Research Center. Integration of land use, fire, and carbon flux in critical Amazon landscapes: the Xingu River headwaters and the BR163 highway corridor
- Qi, Jiaguo - Michigan State University. Land Use and Land Cover Dynamics of China in Support of GOFC/GOLD and NEESPI Sciences
- Saatchi, Sassan - JPL//BU/USDA. Forest woody biomass carbon estimates of N. America from synergistic analysis of MODIS, MISR and JERS data in support of NACP
- Shugart, Herman - University of Virginia. Modeling the carbon dynamics of the Eurasian Boreal Forest Soja Amber, NASA Langley Research Center. Wildfire, ecosystems, and climate in Siberia
- Sun, Guoqing/Masek, Jeff - UMD/NASA. Comparative Studies on Carbon Dynamics in Disturbed Forest Ecosystems: Eastern Russia and Northeastern China
- Tubiello, Francesco - Columbia University. Carbon, Climate and Managed Land in Ukraine: Integrating Data and Models of Land Use for NEESPI
- Woodcock, Curtis - Boston University. Quantifying the Effects of Land-Use Change on Carbon Budgets in the Black Sea Region and China
