Objectives
The primary objectives of our research are: (1) To independently quantify the relationship between biomass density and expected error from GEDI, NISAR and ICESat-2 in high AGB forests in Sonoma County, Costa Rica, and Gabon; (2) To identify the sources of error in high biomass forests for each mission, including from field estimates (GPS error, allometry), from errors in the airborne/spaceborne data (penetration to the ground), and from errors in empirical modeling; (3) To assess data fusion techniques in order to increase the accuracy of AGB estimation through the integration of the airborne simulators for the three missions; (4) To provide AGB stock and error maps to local stakeholders through a user-friendly web portal, enabling the estimation of total AGB and expected error specifically within areas of local interest.
Methods/Techniques
The proposed research focuses on establishing the relationship between AGB density and estimation error for each of three future active remote sensing NASA missions using three study areas with high AGB forests. We propose to use existing airborne datasets that have been collected over forests in Gabon, Costa Rica, and Sonoma County, and to process these datasets to simulate NISAR, ICESAT-2, and GEDI.

Field data have already been collected in all three study sites. New field data will also be collected in particularly high biomass areas of Sonoma County. Finally, Terrestrial Laser Scanning (TLS) data will be collected in Sonoma County, as well as provided to the research team from existing collections in Gabon. This TLS data will quantify existing or expected errors in field estimates of AGB.
LVIS and discrete return Airborne Laser Scanning (ALS) data are the data sources used to simulate GEDI, through a GEDI waveform simulator already under development at the University of Maryland. LVIS data has already been collected in Costa Rica and Gabon, and ALS has been collected in Sonoma County. ALS data will also be used to simulate ICESAT-2’s ATLAS dataset, through a photon counting simulation already tested using ALS data in Gabon. This simulation will be expanded to Sonoma County. Finally, UAVSAR will be used to simulate NISAR. Metrics gleaned from each simulation product will be used to build mission-independent AGB stock and error models for each of the three datasets.
Finally, a prototype design for future mission fusion will be developed to capitalize on the three independent sets of structural observations from GEDI, ICESAT-2 and NISAR. All AGB and error maps will be provided to local stakeholders via a cloud-based GIS software package, Ecometrica, which will enable the manipulation of maps to perform carbon accounting for locally relevant land management activities.
Perceived Significance
Through comparing future mission utility on a shared set of field observations, the proposed research will provide a precise and comparable quantification of expected errors from GEDI, ICESAT-2, and NISAR in high AGB forests. Additionally, methods will be tested to fuse these three future datasets with the intention of developing best practices for AGB and error MRV. By working with scientists from each of the three missions’science teams, this research will provide an unbiased analysis of the strengths and weaknesses of the future missions and inform the development of the next generation of NASA active RS instruments. Additionally, by working with local stakeholders both in the US and abroad, the proposed research will facilitate knowledge and data transfer from data developers to data users in the hopes that best practices can be developed to optimize the utility of future missions products for carbon monitoring initiatives, such as REDD+.
Contact Support to request an email list of project participants.
Project URL(s):
None provided.
Data Products:
Product Title: AGB stock and error maps with associated uncertainties for Costa Rica
Time Period: Corresponding to data acquisitions. Sonoma: 2014, Costa Rica: 2009, Gabon: 2016
Description: Fused AGB stock and error maps from simulated fused GEDI, NISAR & ICESAT-2 over La Selva Biological Research Station, Costa Rica
Status: Preliminary
CMS Science Theme(s): Land Biomass
Keywords:
Spatial Extent: Costa Rica
Spatial Resolution: 0.25 ha – 1 ha
Temporal Frequency: Stock maps (one time only)
Input Data Products: Airborne lidar (Sonoma: ALS, Costa Rica: LVIS, Gabon, LVIS), and UAVSAR
Algorithm/Models Used: GEDI, ICESAT2 and NISAR simulated from airborne proxies, and biomass is empirically derived algorithms developed in this research
Evaluation: Cross validation against field plot estimates of biomass
Intercomparison Efforts/Gaps: NA
Uncertainty Estimates: Error propagated through field data to plot level, through empirical models and mapped to produce 95th percentile confidence interval around pixel estimates.
Uncertainty Categories: Model-Data comparisons
Application Areas:
Relevant Policies/Programs:
Potential Users: Local stakeholders as identified in proposal
Stakeholders:
Current Application Readiness Level: 5
Start Application Readiness Level: 4
Target Application Readiness Level: 7
Future Developments: Web portal(s) to be propagated with biomass maps & uncertainties for stakeholders (Ecometrica in Sonoma, WRI-based web portal in Gabon)
Limitations: These are simulation-based results and represent only predicted mission data performance for biomass. Actual mission datasets may differ based on on-orbit performance, cloud cover, etc.
Date When Product Available:
Metadata URL(s):
Data Server URL(s):
Archived Data Citation:
Bounding Coordinates:
West Longitude:
0.00000
East Longitude:
0.00000
North Latitude:
0.00000
South Latitude:
0.00000
Product Title: CMS: LiDAR Biomass Improved for High Biomass Forests, Sonoma County, CA, USA, 2013
Start Date: 09/2013End Date: 09/2013 (2013)
Description: This data set provides estimates of above-ground woody biomass and uncertainty at 30-m spatial resolution for Sonoma County, California, USA, for the nominal year 2013. Biomass estimates, megagrams of biomass per hectare (Mg/ha), were generated using a combination of airborne LiDAR data and field plot measurements with a parametric modeling approach. The relationship between field estimated and airborne LiDAR estimated aboveground biomass density is represented as a parametric model that predicts biomass as a function of canopy cover and 50th percentile and 90th percentile LiDAR heights at a 30-m resolution. To estimate uncertainty, the biomass model was re-fit 1,000 times through a sampling of the variance-covariance matrix of the fitted parametric model. This produced 1,000 estimates of biomass per pixel. The 5th and 95th percentiles, and the standard deviation of these pixel biomass estimates, were calculated.
Status: Archived
CMS Science Theme(s): Land Biomass
Keywords:
Spatial Extent: Sonoma County, California, USA
Spatial Resolution: The data set has no explicit temporal component. Data are nominally for the year 2013
Temporal Frequency: Grid cells at 30-meter resolution
Input Data Products: LiDAR data were acquired over Sonoma County by Watershed Sciences Inc (WSI) in September – November of 2013 covering ~440,000 ha (44 flights). Airborne discrete return LiDAR instrument - Leica ALS70 sensor was mounted on a Cessna Grand Caravan at 14 points m-2 (Dubayah et al., 2013). Field plot data included the 166 field plots from Dubayah et al. (2017) and 30 new field reference plots that were randomly sampled in tall (>30 m) forests across the County. Field plots were measured as variable radius plots that were distributed as a stratified random sample across the county. The additional 30 plots were sampled randomly from tall forests (>30m) from a layer of land accessibility.
Algorithm/Models Used: The relationship between field estimated and airborne LiDAR estimated aboveground biomass density used a parametric model that predicts biomass as a function of %Canopy Cover (Dubayah et al., 2017), and 50th percentile and 90th percentile LiDAR heights at a 30-m resolution.
Evaluation:
Intercomparison Efforts/Gaps: This revised product was compared to the original product both in terms of the model fit, in areas of known high biomass (e.g. redwood groves), and per pixel across the full County.
Uncertainty Estimates: To estimate per-pixel uncertainty, the biomass model was re-fit 1,000 times through a sampling of the variance-covariance matrix of the fitted parametric model. This produced 1,000 estimates of biomass per pixel. The 5th and 95th percentile, as well as the standard deviation of these pixel estimates, were calculated. Note that error was not propagated from field estimations
Uncertainty Categories:
Application Areas: Forest conservation, land management, GHG accounting, Forest Management
Relevant Policies/Programs: California Environmental Protection Agency Air Resources Board Compliance Offset U.S. Forest Projects
Potential Users: Sonoma County Agricultural Preservation and Open Space District, California State Parks, Redwood National and State Parks
Stakeholders:
Current Application Readiness Level: 6
Start Application Readiness Level: 5
Target Application Readiness Level: 6
Future Developments:
Limitations: This product is primarily focused on improving high biomass forest carbon estimated for 1) those interested in forest carbon accounting and forest conservation in Sonoma County and 2) those interested in comparing satellite biomass products to higher quality reference datasets. Note that the uncertainties reported in the product do not include uncertainties from allometric models (i.e. from field estimates), and thus are underestimates of true uncertainties.
Archived Data Citation: Duncanson, L., R.O. Dubayah, J. Armston, M. Liang, A. Arthur, and D. Minor. 2020. CMS: LiDAR Biomass Improved for High Biomass Forests, Sonoma County, CA, USA, 2013. ORNL DAAC, Oak Ridge, Tennessee, USA. DOI: 10.3334/ORNLDAAC/1764
Bounding Coordinates:
West Longitude:
-123.54000
East Longitude:
-122.34000
North Latitude:
38.85000
South Latitude:
38.11000
Publications:
Disney, M., Burt, A., Wilkes, P., Armston, J., Duncanson, L. 2020. New 3D measurements of large redwood trees for biomass and structure. Scientific Reports. 10(1). DOI: 10.1038/s41598-020-73733-6
Duncanson, L., Armston, J., Disney, M., Avitabile, V., Barbier, N., Calders, K., Carter, S., Chave, J., Herold, M., Crowther, T. W., Falkowski, M., Kellner, J. R., Labriere, N., Lucas, R., MacBean, N., McRoberts, R. E., Meyer, V., Naesset, E., Nickeson, J. E., Paul, K. I., Phillips, O. L., Rejou-Mechain, M., Roman, M., Roxburgh, S., Saatchi, S., Schepaschenko, D., Scipal, K., Siqueira, P. R., Whitehurst, A., Williams, M. 2019. The Importance of Consistent Global Forest Aboveground Biomass Product Validation. Surveys in Geophysics. 40(4), 979-999. DOI: 10.1007/s10712-019-09538-8
Duncanson, L., Neuenschwander, A., Hancock, S., Thomas, N., Fatoyinbo, T., Simard, M., Silva, C. A., Armston, J., Luthcke, S. B., Hofton, M., Kellner, J. R., Dubayah, R. 2020. Biomass estimation from simulated GEDI, ICESat-2 and NISAR across environmental gradients in Sonoma County, California. Remote Sensing of Environment. 242, 111779. DOI: 10.1016/j.rse.2020.111779
Silva, C. A., Duncanson, L., Hancock, S., Neuenschwander, A., Thomas, N., Hofton, M., Fatoyinbo, L., Simard, M., Marshak, C. Z., Armston, J., Lutchke, S., Dubayah, R. 2021. Fusing simulated GEDI, ICESat-2 and NISAR data for regional aboveground biomass mapping. Remote Sensing of Environment. 253, 112234. DOI: 10.1016/j.rse.2020.112234
Archived Data Citations:
Duncanson, L., R.O. Dubayah, J. Armston, M. Liang, A. Arthur, and D. Minor. 2020. CMS: LiDAR Biomass Improved for High Biomass Forests, Sonoma County, CA, USA, 2013. ORNL DAAC, Oak Ridge, Tennessee, USA. DOI: 10.3334/ORNLDAAC/1764