Calibration/Validation

GEDI depends on world-wide, crowd-sourced in situ and airborne datasets to develop representative pre-launch calibration equations for predicting above ground biomass across the GEDI observation domain. These equations are used in the generation of a spatially continuous biomass map of the worlds tropical and temperate forests. To meet the requirements for in situ biomass calculation, coincident airborne laser scanning (lidar) and ground plot field inventory datasets are collated and standardized (see the information box below for general data specifications). Lidar data can be collected with an airborne/drone lidar system or a terrestrial laser scanning instrument. These lidar data are used to simulate GEDI-like waveforms and derived metrics using a simulator developed by the GEDI Science Definition Team (SDT). These simulated waveform metrics are then used to calibrate the biomass equations with the field biomass values as reference data.

General requirements:

  1. Ground inventory plots that are at least the size of a GEDI footprint (25 m diameter)
  2. Geolocation information for inventory plots and lidar data in order to match the two datasets
  3. Temporally coincident data – no more than 2 years difference between ground and airborne data collections
  4. Either individual tree information, OR at the very least both field biomass estimates and a report of the allometric equations and procedure used to estimate the plot level above ground biomass.

 

Field dataset requirements:

  1. Reported geolocation accuracy
  2. Fixed area plots (as opposed to variable radius plots)
  3. Individual tree measurements (diameter, height of measurement, live/dead, species, ideally top height)
  4. A report of how top height was measured
  5. A shapefile of field plot geometry, or plot centroid/orientation/shape/size

 

Lidar dataset requirements:

  1. Airborne lidar
  2. If full lidar dataset unavailable, data clipped to match field plots with a buffer:
         a) 100 m diameters around field plot locations for plots <= 1 ha
         b) lidar clipped to match plots plus a 50 m buffer for plots >1 ha
  3. Preferably greater than 4 pulses per square meter

Currently we are using datasets kindly shared by collaborators. These data are distributed over the world, covering many different biomes, as shown in the map below. Do you have any coincident small footprint lidar data and field measured plot inventory data you would like to contribute to the GEDI project? Please contact the GEDI Cal/Val team (nasagedi@umd.edu) for more information on how you can contribute these data and participate in the GEDI mission.

Locations at which lidar and field data are available for calibration of GEDI biomass equation. Blue indicates publicly available datasets, orange indicates private datasets. The size of each marker (circle) indicates the number of GEDI footprint size field plots that are available at the calibration location.

NASA’s Land Vegetation and Ice Sensor (LVIS) plays a key role in the calibration and validation of the GEDI lidar waveform simulator and data product algorithms. LVIS is deployed post-launch along GEDI ground tracks for assessment of the quality of GEDI Level 1 to Level 4 data products. LVIS is a full-waveform airborne laser scanning instrument that can be configured to acquire data with near-identical specifications to the GEDI lidar instrument. LVIS science and algorithms for vertical forest structure and biomass retrieval over the past twenty years underpin the development of GEDI data products.

Field and lidar data of different scales is processed for calibration and validation of the GEDI biomass product. Reference biomass data is collected in field plots of varying size. LVIS, airborne laser scanning (ALS) and terrestrial laser scanning (TLS) data can all be used to simulate GEDI waveforms and calibrate biomass equations.

NASA LVIS has also participated in opportunistic cross-mission campaigns, such as the February 2016 AfriSAR campaign. AfriSAR was a collaborative ground and airborne campaign between NASA, ESA, DLR, ONERA and AGEOS. Multiple field datasets were collected coincident with airborne and terrestrial laser scanning as well as different modes of Synthetic Aperture Radar (SAR) data. These campaigns are very important for cal/val of the GEDI Level 1 to Level 4 products, as well as those from the NASA NISAR and ESA BIOMASS missions. Coincident lidar and radar data provides further opportunities to explore data fusion activities for improved global mapping of vertical forest structure and biomass.

Simple Share Buttons