A key challenge in a carbon monitoring system is scaling thematically rich but highly localized information to the broad spatial scales needed for carbon accounting and management. This is particularly true for wooded ecosystems, where carbon storage potential is high, but actual carbon status is highly determined by local-scale environmental and forest management conditions.
Through a USDA-NIFA funded project entitled Integrated, observation-based carbon monitoring for wooded lands of Washington, Oregon and California , our team is developing a system to integrate Landsat satellite imagery, maps of environmental characteristics, Forest Inventory and Analysis (FIA) plot data, small-footprint lidar data, and aerial photos to characterize key carbon dynamics in forested ecosystems across all ownerships in the states of Washington, Oregon, and California from 1985 to 2010.
Key characteristics of our system include:
' Operational scaling of local-scale dynamics to all forests in Washington, Oregon, and California
' Yearly mapping of forest biomass and change in biomass from 1990 to 2010
' Explicit characterization of cause of change
' Integration of USDA Forest Service Forest Inventory and Analysis (FIA) plot data
' Linkage of small-footprint lidar data with regional scale biomass maps
' Explicit quantification of methodological uncertainties for all estimates
Because our approach addresses key challenges faced by the current NASA Carbon Monitoring System (CMS), we believe it has the potential to complement and aid NASA s mandate for operational carbon monitoring. To help reach that potential, we propose three activities.
-- 1. We will utilize the products from our own carbon monitoring program in forests of Washington, Oregon, and California to evaluate, understand, and improve performance of the NASA CMS products, and compare a variety of national-scale products both to each other and to FIA plot estimates.
-- 2. We will work with collaborators within the USDA FIA to extend our approaches to a different forest system, linking explicitly with the local-scale NASA CMS efforts in eastern forests.
-- 3. Finally, we will bring our data, methods, and lessons-learned to NASA CMS Science Definition Team, and work closely with other SDT members to link our approaches into those analytical and modeling frameworks to further the overarching goals of the CMS.
The following characteristics of our project are relevant to NASA s need to evaluate and improve its CMS:
- Evaluating the utility and characterizing uncertainties in CMS products
- Understanding scaling issues needed to link local to national scale products
- Developing and demonstrating feasibility of alternative approaches to monitoring
- Illustrating capabilities of satellite-based monitoring for science and management
Contact Support to request an email list of project participants.
Project URL(s):
None provided.
Data Products:
Product Title: Forest biomass maps.
Time Period: 1990-2010
Description: - Create a forest carbon monitoring system using Landsat, airborne Lidar, and field plot data for evaluation of other CMS biomass products.; - Test the carbon monitoring system (originally developed in western forests) in eastern U.S. forests.
Status: Planned
CMS Science Theme(s): Land Biomass
Keywords: Carbon Stocks (; terrestrial)
Spatial Extent: Washington, Oregon, and California
Spatial Resolution: 30 m
Temporal Frequency: Annually
Input Data Products: Landsat time series; Forest Inventory and Analysis plot data (all occasions); small-footprint, airborne discrete return Lidar data and associated field data.; ; Airborne Lidar data collection sites (2001 to 2010, depending on site, variable instruments): Cedar River, WA; Colville, WA; Coos Bay, OR; Deschutes National Forest, OR; Ellsworth, WA; Savanna River, GA/SC; Wind River, WA; Yosemite, CA; ; Area of Lidar data collection: Variable depending on the site, 2349 ha (Ellsworth) to 48,000 ha (Coos Bay)
Algorithm/Models Used: Landsat analysis: LandTrendr time series algorithms; Lidar: various regression-based approaches; Imputation: canonical correspondence analysis, k-neighbor imputation
Evaluation: Cross-validation of imputation models with observations using leave-one-out approaches; Comparison of Landsat-scale with lidar-scale biomass estimates at selected locations
Intercomparison Efforts/Gaps: - Comparison to national scale maps (NBCD, FIA, CMS P1); ; - Comparison at select sites to lidar-based estimates
Uncertainty Estimates: To characterize uncertainty in our core imputation steps, we will use the cross-validation results. That measure of uncertainty is aspatial, however. For spatially-explicit estimates of uncertainty, we will produce multiple runs of the entire prediction system for all pixels, and use the variability as an estimate of uncertainty. The multiple runs will vary in three categories: 1. different strategies for time-series analysis of Landsat imagery; 2. different approaches to drawing plot data in imputation space; 3. different allometric equations to convert plot-level tree data to plot-wide biomass estimates.
Uncertainty Categories: model-model comparison
Application Areas: - Fire management; - Forest inventory; - Land management ; - Invasive species; - Air quality protection
Relevant Policies/Programs: USDA-NIFA project called Integrated, observation-based carbon monitoring for wooded lands of Washington, Oregon, and California, FIA, FLPMA
Potential Users: USFS, Oregon Department of Forestry, Oregon Department of Fish and Wildlife, Washington State Department of Natural Resources, California Department of Forestry and Fire Protection, California Clean Air Resources Board
Stakeholders:
Current Application Readiness Level: 3,4
Start Application Readiness Level: 1
Target Application Readiness Level: 6
Future Developments: Most products now complete (March 2015), with validation and change agent mapping the last steps to complete
Limitations: - Only focused on forests, incomparable to agricultural and urban areas.; - Empirical as opposed to mechanistic, constricted by FIA data.; - Biomass estimates are not likely to capture high biomass situations since they are fundamentally based in optical
Date When Product Available: Anticipated posting on FTP sites by July 2015
Assigned Data Center: ORNL DAAC
Metadata URL(s):
Data Server URL(s):
Archived Data Citation:
Product Title: Maps of forest disturbance by agent, severity, and timing.
Time Period: 1990-2010
Description: - Test the carbon monitoring system (originally developed in western forests) in eastern U.S. forests.
Status: Planned
CMS Science Theme(s): Land Biomass
Keywords: Disturbance (agent; severity; timing)
Spatial Extent: Harvard Forest and environs (Massachusetts), Savanna River Forest and environs (South Carolina & Georgia)
Spatial Resolution: 30 m
Temporal Frequency: Annually
Input Data Products: Landsat time series
Algorithm/Models Used: Landsat analysis: LandTrendr time series algorithms; Lidar: various regression-based approaches
Evaluation: Point-based validation of disturbance
Intercomparison Efforts/Gaps: None
Uncertainty Estimates: Overall estimates of error from point-based disturbance methods
Uncertainty Categories: model-data comparison
Application Areas: - Fire management; - Forest inventory; - Land management ; - Invasive species; - Air quality protection
Relevant Policies/Programs: USDA-NIFA project called Integrated, observation-based carbon monitoring for wooded lands of Washington, Oregon, and California, FIA, FLPMA
Potential Users: USFS, Oregon Department of Forestry, Oregon Department of Fish and Wildlife, Washington State Department of Natural Resources, California Department of Forestry and Fire Protection, California Clean Air Resources Board
Stakeholders:
Current Application Readiness Level: 1,2
Start Application Readiness Level: 1
Target Application Readiness Level: 6
Future Developments: Basic mapping to be completed later this summer
Limitations: Only maps of disturbance in forest
Date When Product Available: Anticipated posting by September 2015
Assigned Data Center: ORNL DAAC
Metadata URL(s):
Data Server URL(s):
Archived Data Citation:
Publications:
Bell, D. M., Gregory, M. J., Kane, V., Kane, J., Kennedy, R. E., Roberts, H. M., Yang, Z. 2018. Multiscale divergence between Landsat- and lidar-based biomass mapping is related to regional variation in canopy cover and composition. Carbon Balance and Management. 13(1). DOI: 10.1186/s13021-018-0104-6
Kennedy, R. E., Ohmann, J., Gregory, M., Roberts, H., Yang, Z., Bell, D. M., Kane, V., Hughes, M. J., Cohen, W. B., Powell, S., Neeti, N., Larrue, T., Hooper, S., Kane, J., Miller, D. L., Perkins, J., Braaten, J., Seidl, R. 2018. An empirical, integrated forest biomass monitoring system. Environmental Research Letters. 13(2), 025004. DOI: 10.1088/1748-9326/aa9d9e
Neeti, N., Kennedy, R. 2016. Comparison of national level biomass maps for conterminous US: understanding pattern and causes of differences. Carbon Balance and Management. 11(1). DOI: 10.1186/s13021-016-0060-y
2015 NASA Carbon Cycle & Ecosystems Joint Science Workshop Poster(s)
An integrated, observation-based system to monitor aboveground forest carbon dynamics in Washington, Oregon, and California
-- (Robert E Kennedy, Matthew Gregory, Janet L. Ohmann, Heather Roberts, Neeti Neeti, David Miller, Zhiqiang Yang, Warren B. Cohen, Van Kane, Jonathan Kane, Scott L. Powell)
[abstract]