Global Waveform Structural Complexity Index (WSCI) estimates provide a spatially continuous measure of forest structure at 25-meter resolution with 3-monthly frequency. The dataset was produced by ORNL_CLOUD using deep learning to fuse GEDI lidar, Sentinel-1 and ALOS-2 PALSAR-2 SAR data, and Copernicus DEM. It covers the period from 2015 to 2022, with model training on approximately 133 million GEDI footprints from 2019 to 2022.
Use Cases
- Model forest carbon storage by analyzing the Waveform Structural Complexity Index (WSCI) across global biomes.
- Assess forest disturbance and recovery over time using the 3-monthly WSCI time series from 2015 to 2022.
- Quantify prediction uncertainty in structural metrics using the provided aleatoric and epistemic standard deviation fields.
- Validate sparse lidar transects by comparing GEDI footprint data with the continuous 25-meter pixel WSCI estimates.
- Map forest structural heterogeneity by analyzing spatial patterns in the mean WSCI and associated data quality indicators.
Strengths
- High spatial resolution of 25 meters per pixel for wall-to-wall global coverage.
- Training utilized approximately 133 million GEDI lidar footprints for model development.
- Provides quantified uncertainty metrics, including aleatoric and epistemic standard deviations.
- Temporal coverage spans 8 years (2015-2022) with a 3-monthly update frequency.
Limitations
- Spatial coverage is limited to latitudes between 51.6 degrees North and 51.6 degrees South, excluding boreal and high-latitude forests.
- Model predictions beyond the GEDI operational domain and pre-2019 rely on sensor fusion, introducing potential extrapolation error.
- The 25-meter resolution may be too coarse for analyzing fine-scale forest structural patterns.
Provenance
- Source
- ORNL_CLOUD, leveraging NASA GEDI, ESA Sentinel-1, JAXA ALOS-2 PALSAR-2, and Copernicus DEM data.
- Collection Method
- Deep learning fusion using an adapted EfficientNet version 2 architecture on multi-sensor satellite data.
- Time Range
- 2015 to 2022
- Freshness
- 3-monthly temporal frequency; last known update was in 2023.
- Geography
- Global, from 51.6°N to 51.6°S latitude.