Net uptake of carbon by forests provides a significant offset to anthropogenic carbon emissions at the global (Pan et al. 2011a), national (Turner et al. 1995, McKinley et al. 2011), and regional (Turner et al. 2007) scales. However, forest carbon sinks are vulnerable to disturbances in the form of harvesting, fire, and insects. At the stand level, these disturbances can alter rates of net primary production and heterotrophic respiration such that the ecosystems become strong carbon sources (Grant et al. 2010). At the regional scale, we have a poor understanding of the relative contribution of these disturbances to overall carbon budgets, but that knowledge is important in understanding how the carbon cycle is responding to on-going climate change and to developing policies for greenhouse gas mitigation. The U.S. Carbon cycle Science Plan (USCCSP 2011) calls for "a more comprehensive land and satellite based network to assess land use effects on the carbon cycle".
Satellite remote sensing, particularly from the Landsat series of sensors at ~ 30m resolution, offers the opportunity to monitor forest disturbances (Cohen and Goward 2004, Goward et al. 2008). In combination with spatially-distributed ecosystem process models that simulate carbon cycle responses to specific disturbances, remote sensing can be used to map and monitor forest carbon stocks and flux (Turner et al. 2004a,b). Here we propose to take advantage of a new Landsat-based time series analysis of forest disturbance (LandTrendr, Kennedy et al. 2010, In press) and a well-established modeling infrastructure for simulating regional carbon flux based on Biome-BGC (Turner et al. 2011a) to quantify carbon cycle impacts of harvesting, fire, and insects on forests of the Northwestern U.S. (WA, OR, ID, western MT). We will make extensive use of plot scale and aggregated USDA Forest Service Forest Inventory and Analysis (FIA) data (Bechtold and Patterson 2004, Wouldenberg et al. 2010) for calibration and validation of both LandTrendr and Biome-BGC.
Our study area provides excellent contrasts over which to characterize (as per the NRA) "impacts to and vulnerability of terrestrial ecosystems and critical terrestrial components of the global carbon cycle to global environmental change". Over the 25 year interval of our proposed study (1985-2010), the Northwest has experienced strong El Niño and La Niña years (the dominant mode of interannual variation in the region), extreme fire years (2002 in Oregon, 2006 in Montana), and a regional reduction in forest harvest driven by federal policy (Turner et al. 2011a). Studies at eddy covariance flux towers in the region suggest strong sensitivity of net ecosystem exchange (NEE) to ENSO related climate variation (Morgenstern et al. 2004, Wharton et al. 2009). In Oregon, extensive fires in 2002 generated relatively high direct carbon emissions and low net ecosystem production (NEP, the balance of net primary production and heterotrophic respiration) for that year (Turner et al. 2007). The 83% reduction in harvest on public land in the area of the Northwest Forest Plan has contributed to a major increase in NEP on those lands. Our proposed research will quantify the regional carbon budget and its possible trends over time. We will address questions about the sensitivity of the regional disturbance regime to interannual variation (IAV) in climate, as well as policy question about carbon benefits of harvesting or not harvesting on public lands. Our study will thus isolate the role of climatic and anthropogenic factors on the regional carbon budget. In addition, it will provide highly spatially resolved flux maps with well characterized uncertainty for use in evaluating alternative scaling approaches and assessing overall uncertainty in regional flux estimates.
Our approach relies heavily on the LandTrendr products, which have three key benefits over prior methods of detecting disturbance. Like other new algorithms emerging since the US Geological Survey (USGS) began providing Landsat imagery free of charge (Huang et al. 2010), 30-m resolution maps of disturbance can now be produced at the annual time step that is needed for carbon accounting. Second, the greater signal-to-noise ratio of using time-series data allows detection of more subtle disturbance than has been possible in the past (e.g. Cohen et al. 2002), particularly important for detecting the low-intensity partial forest harvest that is ubiquitous in the western U.S. Finally, the approach is unique among current methods in responding to both long-term trends and abrupt events, allowing capture of both chronic disturbance, such as that caused by forest insects (Meigs et al. in press), as well as anthropogenic activities and fire. The proposed work will take advantage of ongoing LandTrendr-based forest disturbance mapping for the entire states of Washington, Oregon, and California under as USDA-NIFA funded monitoring project (PI: Kennedy, Title: "Integrated, Observation-Based Carbon Monitoring for Wooded Lands of Washington, Oregon, and California"). The expected launch of the Landsat Data Continuity Mission (LDCM) in December 2012 will insure continuous delivery of data for regular updating.
Our scaling approach also relies heavily on the Biome-BGC carbon cycle process model (Thornton et al. 2002). This model is widely used for regional scaling of NPP and NEP (e.g. Churkina et al. 2010). It treats the carbon, nitrogen, and hydrologic cycles. Its advantages include 1) a daily time step, which allows for calibration and validation with ecophysiological and micrometeorological measurements, 2) disaggregation of carbon stocks into pools that are commonly estimated such as wood mass and coarse woody debris, thus allowing forest inventory data to also be used for calibration and validation, 3) tracking of stand age and ecosystem recovery from disturbance, and 4) extensive previous research on parameterization in different vegetation types (e.g. White et al. 2000, Thornton et al. 2002, Wang et al. 2009). With previous support from EPA and DOE we have adapted the Biome-BGC code to simulate effects of harvests and fire (Law et al. 2006, Turner et al. 2007, Meigs et al. 2011). Here we will develop algorithms for thinning and insect outbreaks.
The third leg of our scaling approach is reliance on FIA data for model parameterization and uncertainty assessment. The FIA plot network is a design-unbiased sample of all forest land in the U.S. on a permanent 5.3 km grid with a remeasurement cycle of 10 years in the west (Bechtold and Patterson 2004). The representative FIA sample provides robust regional statistics of carbon stocks, but translating the variety of forest conditions encountered on each plot to model inputs and evaluating change from prior measurements and protocols can be complicated (Gray et al. in press). Co-I Gray has extensive experience with the origin, extraction, and application of this data (e.g., Gray and Azuma 2005, Gray et al. 2005, Gray et al. 2009). We will be using plot data for LandTrendr and Biome-BGC calibration and validation, and we will use aggregated plot data on carbon stocks at two points in time for evaluation of our regional flux estimates.
1. Update and compile field measurements from FIA permanent field plots over time to provide 1) consistent interpretations of disturbance and management events and their impacts on carbon stores, and 2) data for model development.
2. LandTrendr development and application. Apply LandTrendr over the 1985-2010 interval and the 4 state domain of our study. Outputs include annual maps of disturbance type and magnitude, with characterization of uncertainty.
3. Biome-BGC development and application. Update the Biome-BGC model to simulate partial disturbances. Apply it over the 4 state domain of our study in a spatially-distributed mode using LandTrendr disturbances. Characterize uncertainty in model inputs, model parameters, and model outputs.
4. Address relevant science questions. We will first develop a time series for a) areal extent, mortality, and magnitude of direct emissions from forest fires in our study region, b) areal extent and mortality from insect outbreaks in our study region, c) areal extent of clear-cut and partial harvests in the study region, d) mapped NEP. We will then examine the relationship of fire impacts to interannual variation (IAV) in climate, and the relationship of insect outbreaks to IAV in climate. We will compare the relative magnitudes of mortality from harvest, fire, and insects and how they change over time. These analyses will be stratified by ecozone.
5. Examine Relevant Policy Questions. We will develop a time series in each ecozone of the study region for NEP and NECB on public and private lands. We can then compare harvest removals, fire emissions, NEP, and NECB between public and private lands for selected intervals (e.g. Figure 1).
6. Address Scaling Questions. We will compare our LandTrendr/Biome-BGC estimates for carbon pools and fluxes in our study region to results from a variety of alternative scaling approaches.
Co-I Gray is a Research Ecologist at the USDA Forest Service PNW Research Station and is Leader of the Vegetation Monitoring Science and Applications team within the Station. The team is responsible for conducting research with, and providing direction to, the Forest Inventory and Analysis (FIA) program on the West Coast (AK, CA, OR,WA, and HI). He also has working relationships with FIA personnel in the Rocky Mountain Research Station. He is thus in a good position to assemble and analyze the historical and more recent FIA plot data, as well as aggregate the data for regional analysis (e.g. Campbell et al. 2010).
The LandTrendr analysis of Landsat Thematic Mapper time series (Kennedy et al. 2010, In Press) allows capture of both abrupt events, such as fire and harvest, and ongoing disturbance processes, such as spreading mortality caused by insects and drought. The core step is a temporal segmentation: partitioning the temporal "life history" of each pixel into straight-line segments that describe periods when a consistent process is occurring. Relative to historical methods that relied on two or a handful of images to detect abrupt change, the trajectory approach capitalizes on both the change event and the conditions before and after the event to better separate signal from noise. Additionally, the trajectory approach can detect long-duration processes that cause persistent degradation of the forest canopy, a process observed consistently for most insect-related mortality events in Oregon and Washington. Technically, the approach is attractive in being able to mosaic on the fly multiple images in a year, allowing seamless use of many partly cloudy Landsat 5 images as well as 'gappy' Landsat 7-SLC off images. Thus, the method is not contingent on the availability of cloud-free imagery.
Conceptually, the primary advantage of temporal segmentation is that it simplifies the complex yearly signal from Landsat data, allowing direct mapping of the disturbance year or disturbance onset, the magnitude of change, and, for long-duration mortality events, the duration of that mortality process. Under support from the USDA Forest Service, this process has been applied for the period 1985 to 2007 to the entire area of the Northwest Forest Plan (NWFP), which encompasses all forests in western Washington, Oregon, and parts of Northern California. We currently also have USDA funding to extend this coverage to eastern WA and eastern OR through 2010. For this project, we will expand the geographic scope to forests in Idaho and western Montana.
Critical to parameterization in Biome-BGC are robust labels for the cause, or agent, of each disturbance. On smaller projects working with the National Park Service and the National Marine Fisheries Service, we have developed methods to assign these labels. The approach uses the Random Forest statistical technique (Breiman 2001) to link a suite of spectral, contextual, and patch-shape characteristics with agent-labels defined at a sample of disturbances by a trained interpreter. Interpreters assign these labels using information from the imagery itself, from high resolution photos, and from ancillary databases from the Monitoring Trends in Burn Severity (MTBS) program (Eidenshink et al. 2007) and the USDA Forest Service's Aerial Detection Survey within the Forest Health Monitoring program (http://fhm.fs.fed.us/dm/maps/aerial.shtml). Under the support of our USDA-funded project, these change agent labels will already be assigned for all disturbances in Washington and Oregon. We will extend the same methodology to Idaho and western Montana. Additionally, we will use repeat-measurement observations at the FIA plots to aid calibrating these algorithms for Idaho and Montana. A large proportion of the FIA plots in the region have been measured multiple times over the last 30 years, and information on the nature and severity of disturbance, and estimates of changes in carbon stocks, are available. These estimates can be used to calibrate the LandTrendr algorithms for characterizing disturbance type and magnitude (plot location are available for research purposes).
Accuracy Assessment. Assessing the accuracy of all maps is necessary to place uncertainty bounds on the map-wide estimates of disturbance rate. We will use TimeSync, a stand-alone open source software tool developed at OSU to allow visual interpretation of time series data from both Landsat image time series and from other airphoto-based images (Cohen et al. 2010). Under the auspices of the NWFP project, we have already collected validation data for more than 2000 plots that has been used in validation efforts. New plot data from Idaho and Montana will be added to our validation data base.
In addition to the requirement for information on recent disturbances, our Biome-BGC based carbon cycle modeling approach also requires an indication of stand age for those areas undisturbed in the last 25 yrs. Previously we have used single Landsat images and developed spectral signatures for different stand age classes using FIA plots (Cohen et al. 1995, Duane et al. 2010). The signal in these statistical models of spectral vegetation indices against stand age is rather weak, thus we will employ an improved approach for this study in which stand age class is estimated using a combination of LandTrendr and Gradient Nearest neighbor Analysis (GNN, Ohmann et al. 2002, In press). With this approach, each pixel is assigned the attributes of the FIA plot it most closely resembles in a statistical analysis of the spectral data and other geospatial data. Through our USDA-funded project, Kennedy and Ohmann will extend the LandTrendr/GNN analysis to eastern Oregon and Washington, and from these maps age will be extracted. The effort involved in creating these maps is substantial, however, and for the currently-proposed project, only a single variable (age) would be needed. Therefore, for simplicity and efficiency, we will use age maps estimated using GNN for eastern WA and OR to train simple statistical relationships that we can then extend to the spectral data in ID and MT. Because the LandTrendr approach removes most spurious spectral features from image stacks, it also allows seamless cross-scene mosaicking, which facilitates cross-scene model extension of this sort. We will then evaluate the resultant map data by comparison of the stand age distributions to the distributions of stand age in the FIA database for those states as well as the Landsat based successional stage maps from the USDA Landfire Project (Landfire 2007).
Our primary tool for scaling carbon flux across the regional domain is the Biome-BGC carbon cycle process model. The model has a daily time step and is run over multiple years to simulate succession. Simulated carbon cycle processes include photosynthesis, autotrophic respiration, heterotrophic respiration, plant C allocation, and mortality. Simulated C pools include stemwood, coarse roots, fine roots, foliage, litter, CWD, and soil organic matter. A water balance is also calculated. For each point-mode run, the model is spun-up with a 25-year repeating loop of climate data, and disturbance events are prescribed by year, type, and intensity to bring the simulation up to the current condition. A look-up table determines the partitioning of the extant carbon stocks at the time of disturbance into removals, direct fire emissions, and necromass, thus maintaining ecosystem mass balance. Partitioning factors are based on observations within the region (Turner et al. 1995, Law et al. 2003, Sun et al. 2004, Campbell et al. 2007). Our simulations include all cover types, notably agricultural systems which have prescribed harvests (as a proportion of NPP) hence tend to be carbon sinks.
A significant limitation of the current model version, especially with respect to simulating thinning events, is the lack of a distinction between trees and shrubs. Our field observations (Meigs 2009, Campbell et al. 2009) suggest that in the drier eastside forests, stand level productivity is restored quickly after fire, a response initially dominated by shrub production. To better capture that dynamic, which has important effects on the temporal trends in NEP, we will begin using a multiple life form version of the model (i.e. two life forms may co-exist in a given grid cell).
Our current version of Biome-BGC was adapted from version 4.1.2 (Thornton et al. 2002) to simulate stand-replacing disturbances (Law et al. 2004), dynamic allocation over the course of succession (Law et al. 2006), and mixed severity fire (Meigs et al. 2011). Bond-Lamberty et al. (2007) have produced versions of Biome-BGC with multiple life forms that we have tested, and the two life form logic there will be adapted for our purposes. We will have data on understory vegetation from the FIA data to aid in model building/validation.
Our NEP/NECB scaling methods and initial validation results for applications in Oregon, Washington, and California have been reported in Turner et al. (2004a,b, 2007, 2011a,b), and Law et al. (2004, 2006). We have completed model runs for all of OR and WA through 2006. For this proposal we will revise and update all model inputs and extend our domain to cover Idaho and western Montana.
Land Cover Input. Our base land cover dataset will be the 2006 National Land Cover Database (Vogelmann et al. 2001), which is at Landsat resolution (~30 m).
Disturbance Histories. LandTrendr will be used to develop annual maps (1985-2010) of disturbance over the study domain. Disturbance attributes mapped at the 30m grain will include year of disturbance, attribution (thinning, clear-cut, fire, insects), and magnitude. As noted in the previous section, older disturbances are prescribed on the basis of mapped stand age class from LandTrendr/GNN. Stands > 25 yrs of age will be binned into 3 age classes (young, mature, old) and assigned the age class midpoint to keep the number of unique disturbance histories manageable.
Climate Data Input. The meteorological inputs to Biome-BGC are daily minimum and maximum temperature, precipitation, humidity, and solar radiation. We currently have a 27-year (1980-2006) time series at 1 km resolution over the 4 state region that was developed with the Daymet model (Thornton et al. 1997, Thornton & Running 1999, Thornton et al. 2000). The data are based on interpolations of meteorological station observations using a digital elevation model and general meteorological principles (http://www.daymet.org/). Uncertainty in Daymet interpolations has previously been evaluated in our region and elsewhere (Hasenauer et al. 2003, Daly et al. 2008). For this analysis we will obtain updated Daymet meteorology through 2010. That data product is under development by P. Thornton (ORNL) as a contribution to the North American Carbon Program (Thornton et al. 2011) and will be available through the NACP data center (MAST-DC).
Soil Data Input. Soil texture and depth are specified from US Geological Survey soil maps (CONUS 2007).
Model Parameterization. Biome-BGC has ~20 cover type specific ecophysiological parameters that must be specified. We primarily adopt the recommendations in White et al. (2000). A representative set of parameters is given in Turner et al. (2007). In the case of the evergreen needle leaf cover type, we do a final ecoregion specific adjustment to two of the parameters (the fraction of leaf nitrogen as rubisco, FLNR, and the mortality fraction) that have been identified in our sensitivity analyses as heavily impacting wood mass and NEP. Reference data for these parameter selections are site- and age-specific estimates of live biomass from our database of USDA Forest Service FIA plots (e.g. Waddell and Hiserote 2005, VanTuyl et al. 2005, Hudiburg et al. 2009). We run the model at each FIA plot location (available for research purposes) to the stand age in the plot data and compare predicted and observed live biomass.
Model Algorithm Development. We have previously adapted Biome-BGC for simulating harvest events (Law et al. 2004) and variable intensity fire (Meigs et al. 2011). Here we will add code for variable intensity harvest and insect attack. Disturbance intensity is a continuous variable from LandTrendr. This proportional disturbance intensity will be applied directly to the live carbon pools when a partial disturbance is prescribed by LandTrendr. Transfers to dead carbon pools will follow existing logic in the case of harvest (thinning) removals. In the case of insect attack, we will make transfers to a standing dead pool and parameterize subsequent transfers to coarse woody debris on the ground based on field studies in the literature (Busse 1994, Mitchell and Preisler 1998, Brown et al. 2011). We will also implement an optional substrate limitation algorithm to address the issue of possible overestimation of autotrophic respiration in Biome-BGC (Mitchell et al. 2011).
Model Application. It is not computationally feasible to run Biome-BGC for each 30 m Landsat pixel. Thus we take the 5 most frequent combinations of cover type and disturbance history and report their weighted mean for the 1 km fluxes. Using this scheme we generally cover > 90% of the forest area under consideration.
Uncertainty analysis. We will evaluate point mode daily flux simulations (GPP and NEE) and their and seasonality with eddy covariance flux tower data (3 active tower sites in the region). When optimizing model parameters based on FIA plot data, we will conduct a cross-validation exercise to characterize uncertainty in simulated biomass. Regional changes in wood volume (e.g. Figure 6) will be used to estimate changes in C stocks and NEP as simulated by LandTrendr/Biome-BGC (e.g. Turner et al. 2011b).
Bechtold, W.A., Patterson, P.L., 2004. The enhanced Forest Inventory and Analysis Program--national sampling design and estimation procedures. General Technical Report SRS-80. USDA Forest Service, Southern Research Station, Asheville, NC, p. 85.
Bond-Lamberty, B., Peckham, S. D., Ahl, D. E. & Gower, S. T. (2007) Fire as the dominant driver of central Canadian boreal forest carbon balance. Nature, 450, 89-+.
Breiman, L. (2001), Random forests, Machine Learning, 45(1), 5-32.
Brown, M., Black, T. A., Nesic, Z., Foord, V. N., Spittlehouse, D. L., Fredeen, A. L., Grant, N. J., Burton, P. J. & Trofymow, J. A. (2011) Impact of mountain pine beetle on the net ecosystem production of lodgepole pine stands in British Columbia. Agricultural and Forest Meteorology, 150, 254-264.
Busse, M. D. (1994) Downed bole-wood decomposition in lodgepole pine forest of central Oregon. Soil Science Society of America Journal, 58, 221-227.
Campbell, J., Alberti, G., Martin, J. & Law, B. E. (2009) Carbon dynamics of a ponderosa pine plantation following a thinning treatment in the northern Sierra Nevada. Forest Ecology and Management, 257, 453-463.
Campbell, J., Donato, D., Azuma, D. L. & Law, B. (2007) Pyrogenic carbon emission from a large wildfire in Oregon, United States. Journal of Geophysical Research - Biogeosciences, 12, G04014.Campbell, S., Waddell, K., Gray, A., 2010. Washington's Forest Resources, 2002 to 2006. General Technical Report PNW-800. U.S.D.A. Forest Service, Pacific Northwest Research Station, Portland, OR.
Campbell, Sally, Karen Waddell, and Andrew Gray, tech. ed's. 2010. Washington's Forest Resources, 2002 to 2006. General Technical Report PNW-GTR-800. Portland, OR:
Churkina, G., Zaehle, S., Hughes, J., Viovy, N., Chen, Y., Jung, M., Heumann, B. W., Ramankutty, N., Heimann, M. & Jones, C. (2010) Interactions between nitrogen deposition, land cover conversion, and climate change determine the contemporary carbon balance of Europe. Biogeosciences, 7, 2749-2764.
Cohen, W. B., Spies, T. A. & Fiorella, M. (1995) Estimating the age and structure of forests in a multi-ownership landscape of western Oregon, U.S.A. International Journal of Remote Sensing, 16, 721-746.
Cohen, W. B., Spies, T. A., Alig, R. J., Oetter, D. R., Maiersperger, T. K. & Fiorella, M. (2002) Characterizing 23 years (1972-95) of stand replacement disturbance in western Oregon forests with Landsat imagery. Ecosystems, 5, 122-137.
Cohen, W. B. & Goward, S. N. (2004) Landsat's role in ecological applications of remote sensing. BioScience, 54, 535-545.
Cohen, W. B., Yang, Z. G. & Kennedy, R. (2010) Detecting trends in forest disturbance and recovery using yearly Landsat time series: 2. TimeSync - Tools for calibration and validation. Remote Sensing of Environment, 114, 2911-2924.
CONUS. 2007. Conterminous United States multi-layer soil characteristics data set for regional climate and hydrology modeling. http://www.soilinfo.psu.edu/index.cgi?soil_data&conus
Daly, C., Halbleib, M., Smith, J. I., Gibson, W. P., Doggett, M. K., Taylor, G. H., Curtis, J. & Pasteris, P. P. (2008) Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. International Journal of Climatology, 28, 2031-2064.
Duane, M. V., Cohen, W. B., Campbell, J. L., Hudiburg, T., Weyermann, D. & Turner, D. P. (2010) Implications of two different field-sampling designs on Landsat-based forest age maps used to model carbon in Oregon forests. Forest Science 65, 405-416.
Eidenshink, J., Schwind, B., Brewer, K., Zhu, Z. L., Quayle, B. & Howard, S. M. (2007) A project for monitoring trends in burn severity. Fire Ecology, 3, 3-21.
Goward, S. N., Masek, J. G., Cohen, W. B., Moisen, G., Collatz, G. J., Healey, S. P., Houghton, R. A., Jhuang, C., Kennedy, R. E., Law, B., Powell, S., Turner, D. & Wulder, M. A. (2008) Forest disturbance and North American carbon flux. Eos, 89, 105-116.
Grant, R. F., Barr, A. G., Black, T. A., Margolis, H. A., Mccaughey, J. H. & Trofymow, J. A. (2010) Net ecosystem productivity of temperate and boreal forests after clearcutting-a Fluxnet-Canada measurement and modelling synthesis. Tellus Series B-Chemical and Physical Meteorology, 62, 475-496./font>
Gray, A.N., Azuma, D.L., 2005. Repeatability and implementation of a forest vegetation indicator. Ecological Indicators 5, 57-71.
Gray, A.N., Veneklase, C.F., Rhoads, R.D., 2005. Timber Resource Statistics for Nonnational Forest Land in Western Washington, 2001. In: Resource Bulletin PNW-RB-246. Pacific Northwest Research Station, U.S. Department of Agriculture, Forest Service, Portland, Oregon, p. 117 pp.
Gray, A.N., Monleon, V.J., Spies, T.A., 2009. Characteristics of remnant old-growth forests in the northern Coast Range of Oregon and comparison to surrounding landscapes. General Technical Report PNW-GTR-790. U.S.D.A. Forest Service, Pacific Northwest Research Station, Portland, Oregon. 45 p.
Gray, Andrew N., Thomas J. Brandeis, John D. Shaw, William H. McWilliams. in press. Forest inventory vegetation database of the United States of America. Biodiversity & Ecology./font>
Hasenauer, H., Merganicova, K., Petritsch, R., Pietsch, S. A. & Thronton, P. E. (2003) Validating daily climate interpolations over complex terrain in Austria. Agricultural and Forest Meteorology, 119, 87-107.
Huang, C., S. N. Goward, J. G. Masek, N. Thomas, Z. Zhu, and J. E. Vogelmann (2010), An automated approach for reconstructing recent forest disturbance history using dense Landsat time series stacks, Remote Sensing of Environment, 114, 183-198.
Hudiburg, T., Law, B., Turner, D. P., Campbell, J., Donato, D. & Duane, M. (2009) Carbon dynamics of Oregon and Northern California forests and potential land-based carbon storage. Ecological Applications, 19, 163-180.
Kennedy, R. E., Yang, Z. G. & Cohen, W. B. (2010) Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr - Temporal segmentation algorithms. Remote Sensing of Environment, 114, 2897-2910.
Kennedy, R. E., Y. Zhiqiang, W. Cohen, E. Pfaff, J. Braaten, and P. Nelson (In press), Spatial and temporal patterns of forest disturbance and regrowth within the area of the Northwest Forest Plan, Remote Sensing of Environment.
Landfire. (2007) Landscape Fire and Resource Management Planning Tools Project. http://www.landfire.gov/
Law, B. E., Sun, O. J., Campbell, J., Van Tuyl, S. & Thornton, P. E. (2003) Changes in carbon storage and fluxes in a chronosequence of ponderosa pine. Global Change Biology, 9, 510-524.
Law, B. E., Turner, D., Campbell, J., Van Tuyl, S., Ritts, W.D. & Cohen, W. B. (2004) Disturbance and climate effects on carbon stocks and fluxes across Western Oregon USA. Global Change Biology, 10, 1429-1444.
Law, B. E., Turner, D. P., Lefsky, M., Campbell, J., Guzy, M., Sun, O., Vantuyl, S. & Cohen, W. B. (2006) Carbon fluxes across regions: observational constraints at multiple scales, edn. Columbia University Press, New York.
Mckinley, D. C., Ryan, M. G., Birdsey, R. A., Giardina, C. P., Harmon, M. E., Heath, L. S., Houghton, R. A., Jackson, R. B., Morrison, J. F., Murray, B. C., Pataki, D. E. & Skog, K. E. (2011) A synthesis of current knowledge on forests and carbon storage in the United States. Ecological Applications, 21, 1902-1924.
Meigs, G. W., Turner, D. P., Ritts, W. D., Yang, Z. Q. & Law, B. E. (2011) Landscape-Scale Simulation of Heterogeneous Fire Effects on Pyrogenic Carbon Emissions, Tree Mortality, and Net Ecosystem Production. Ecosystems, 14, 758-775.
Mitchell, R.G., Priesler, H.K. (1998) Fall rate of lodgepole pine killed by the mountain pine beetle in Central oregon. Western Journal of Applied forestry 13:23-26.
Mitchell, S. R., Harmon, M. E. & O'connell, K. E. B. (2009) Forest fuel reduction alters fire severity and long-term carbon storage in three Pacific Northwest ecosystems. Ecological Applications, 19, 643-655.
Morgenstern, K., Black, T. A., Humphreys, E. R., Griffis, T. J., Drewitt, G. B., Cai, T., Nesic, Z., Spittlehouse, D. L. & Livingston, N. J. (2004) Sensitivity and uncertainty of the carbon balance of a Pacific Northwest Douglas-fir forest during an ElNino/LaNina cycle. Agricultural and Forest Meteorology, 123, 201-219.
Ohmann, J. L. & Gregory, M. J. (2002) Predictive mapping of forest composition and structure with direct gradient analysis and nearest-neighbor imputation in coastal Oregon, U.S.A. Canadian Journal of Forest Research, 32, 725-741.
Ohmann, J. L., Gregory, M. J., Roberts, H. M., Cohen, W. B., Kennedy, R. E. & Yang, Z. (In Review) Spatial monitoring of older forest with nearest-neighbor imputation and Landsat time-series.
Pan, Y., Chen, J. M., Birdsey, R., Mccullough, K., He, L. & Deng, F. (2011b) Age structure and disturbance legacy of North American forests. Biogeosciences, 8, 715-732.
Sun, O. J., Campbell, J., Law, B. E. & Wolf, V. (2004) Dynamics of carbon stocks in soils and detritus across chronosequences of different forest types in the Pacific Northwest, USA. Global Change Biology, 10, 1470-1481.
Thornton, P. E., Running, S. W. & White, M. A. (1997) Generating surfaces of daily meteorological variables over large regions of complex terrain. Journal of Hydrology, 190, 214-251.
Thornton, P. E. & Running, S. W. (1999) An improved algorithm for estimating incident daily solar radiation from measurements of temperature, humidity, and precipitation. Agricultural and Forest Meteorology, 93, 211-228.
Thornton, P. E., Hasenauer, H. & White, M. A. (2000) Simultaneous estimation of daily solar radiation and humidity from observed temperature and precipitation: an application over complex terrain in Austria. Agricultural and Forest Meteorology, 104, 255-271
Thornton, P. E., Law, B. E., Gholz, H. L., Clark, K. L., Falge, E., Ellsworth, D. S., Goldstein, A. H., Monson, R. K., Hollinger, D., Falk, M., Chen, J. & Sparks, J. P. (2002) Modeling and measuring the effects of disturbance history and climate on carbon and water budgets in evergreen needleleaf forests. Agricultural and Forest Meteorology, 113, 185-222.
Thornton, M.L., Thornton, P.E., Cook, R.B., Wei, Y., Vannan, S.K.S., Eby, P.I., Devarakonda, R. 2011. Daymet: Gridded subdaily weather data for North America. Poster for NASA Carbon Cycle & Ecosystems Joint Science Workshop, Oct. 3-7, 2011.
Turner, D. P., Koerper, G. J., Harmon, M. E. & Lee, J. J. (1995) A carbon budget for forests of the conterminous United States. Ecological Applications, 5, 421-436.
Turner, D. P., Guzy, M., Lefsky, M., Ritts, W., Vantuyl, S. & Law, B. E. (2004a) Monitoring forest carbon sequestration with remote sensing and carbon cycle modeling. Environmental Management, 4, 457-466.
Turner, D. P., Ollinger, S. V. & Kimball, J. S. (2004b) Integrating remote sensing and ecosystem process models for landscape to regional scale analysis of the carbon cycle. BioScience, 54, 573-584.
Turner, D. P., Ritts, W. D., Law, B. E., Cohen, W. B., Yang, Z., Hudiburg, T., Campbell, J. L. & Duane, M. (2007) Scaling net ecosystem production and net biome production over a heterogeneous region in the western United States. Biogeosciences, 4, 597-612.
Turner, D. P., Ritts, W. D., Yang, Z. Q., Kennedy, R. E., Cohen, W. B., Duane, M. V., Thornton, P. E. & Law, B. E. (2011a) Decadal trends in net ecosystem production and net ecosystem carbon balance for a regional socioecological system. Forest Ecology and Management, 262, 1318-1325.
Turner, D. P., Gockede, M., Law, B. E., Ritts, W. D., Cohen, W. B., Yang, Z., Hudiburg, T., Kennedy, R. E. & Duane, M. (2011b) Multiple constraint analysis of regional land surface carbon flux. TellusB 63:207-221.
Van Tuyl, S., Law, B. E., Turner, D. P. & Gitelman, A. I. (2005) Variability in net primary production and carbon storage in biomass across forests -- an assessment integrating data from forest inventories, intensive sites, and remote sensing. Forest Ecology and Management, 209, 273-291.
USCCSP. 2011. A U.S. Carbon Cycle Science Plan. http://www.carboncyclescience.gov/
Vogelmann, J. E., Howard, S. M., Yang, L., Larson, C. R., Wylie, B. K. & Van Driel, N. (2001) Completion of the 1990s National Land Cover Data Set for the conterminous United States from Landsat Thematic Mapper data and ancillary data sources. Photogrammetric Engineering and Remote Sensing, 67, 650-652.
Waddell, K.L., Hiserote, B. 2005. The PNW-FIA Integrated Database, Version 2.0. USDA Forest Service Pacific Northwest Research Station, Forest Inventory and Analysis program, Portland, OR. http://www.fs.fed.us/pnw/fia/publications/data/data.shtml
Wang, W., Ichii, K., Hashimoto, H., Michaelis, A. R., Thornton, P. E., Law, B. E. & Nemani, R. R. (2009) A hierarchical analysis of terrestrial ecosystem model Biome-BGC: Equilibrium analysis and model calibration. Ecological Modelling, 220, 2009-2023.
Wharton, S., Chasmer, L., Falk, M. & U, K. T. P. (2009) Strong links between teleconnections and ecosystem exchange found at a Pacific Northwest old-growth forest from flux tower and MODIS EVI data. Global Change Biology, 15, 2187-2205.
White, M. A., Thornton, P. E., Running, S. W. & Nemani, R. R. (2000) Parameterization and sensitivity analysis of the BIOME-BGC terrestrial ecosystem model: net primary production controls. Earth Interactions, 4, 1-85.
Woudenberg, S.W., Conkling, B.L., O'Connell, B.M., LaPoint, E.B., Turner, J.A., Waddell, K.L. 2010. The Forest Inventory and Analysis Database: Database description and users manual version 4.0 for Phase 2. General Technical Report RMRS-245. U.S.D.A. Forest Service, Rocky Mountain Research Station, Fort Collins, CO. 339 p.