Predicted Soil Organic Carbon Stocks for Pacific Northwest Watersheds

ORNL_CLOUD provides predicted soil organic carbon (SOC) stocks and percentages for 2021-2022 across four Pacific Northwest watersheds. Data includes predictor layers from lidar and satellite imagery (2012-2022) and is based on field samples from 114 locations. The dataset supports wetland carbon storage studies using linear mixed effects and quantile random forest models.

Use Cases

• Predict SOC stock using Wetland Intrinsic Potential (WIP) probability and Sentinel-2 vegetation spectral indices.
• Model SOC percent from predictor layers related to topography, lithology, and climate.
• Compare linear mixed effects model coefficients against quantile random forest outputs for feature importance.
• Analyze spatial patterns of carbon storage across the Heen Latinee, Hoh River, Mashel River, and Colville watersheds.

Strengths

• Includes field sample observations from 114 specific locations for model validation.
• Predictor data spans a 10-year temporal range from 2012 to 2022.
• Covers four distinct watersheds across Alaska and Washington state.

Limitations

• Predicted values are nominal for 2021-2022; temporal coverage for predictions is limited.
• Sample size of 114 field locations may be insufficient for highly granular sub-watershed analysis.
• Urban zones and surface water areas are masked out, limiting analysis in those land covers.

Provenance

Source

ORNL_CLOUD via NASA EarthData.

Collection Method

Field sample collection combined with predictor layers from lidar and satellite imagery processed in Google Earth Engine.

Time Range

Predictor data: 2012-2022; Field samples and nominal predictions: 2021-2022.

Freshness

Data reflects a nominal period ending in 2022.

Geography

Four watersheds in the Pacific Northwest: Heen Latinee Experimental Forest (Alaska), Hoh River Watershed, Mashel River Watershed, and Colville Watershed (Washington).