Forest ecosystems contribute substantially to carbon (C) storage. The dynamics of litter decomposition, translocation and stabilization into soil layers are essential processes in the functioning of forest ecosystems, as these processes control the cycling of soil organic matter and the accumulation and release of C to the atmosphere. Therefore, the spatial distribution of litter and soil C stocks are important in greenhouse gas estimation and reporting and inform land management decisions, policy, and climate change mitigation strategies. Here we explored the effects of spatial aggregation of climatic, biotic, topographic and soil variables on national estimates of litter and soil C stocks and characterized the spatial distribution of litter and soil C stocks in the conterminous United States (CONUS). Litter and soil variables were measured on permanent sample plots (n = 3303) from the National Forest Inventory (NFI) within the United States from 2000 to 2011. These data were used with vegetation phenology data estimated from LANDSAT imagery (30 m) and raster data describing environmental variables for the entire CONUS to predict litter and soil C stocks. The total estimated litter C stock was 2.07 ± 0.97 Pg with an average density of 10.45 ± 2.38 Mg ha −1 , and the soil C stock at 0–20 cm depth was 14.68 ± 3.50 Pg with an average density of 62.68 ± 8.98 Mg ha −1 . This study extends NFI data from points to pixels providing spatially explicit and continuous predictions of litter and soil C stocks on forest land in the CONUS. The approaches described illustrate the utility of harmonizing field measurements with remotely sensed data to facilitate modeling and prediction across spatial scales in support of inventory, monitoring, and reporting activities, particularly in countries with ready access to remotely sensed data but with limited observations of litter and soil variables.
Bibliographical noteFunding Information:
This study was funded by USDA Forest Service, Forest Inventory and Analysis Program. The authors would like to thank Ronald McRoberts for early discussion on possible stratification techniques.
- Google Earth Engine
- Machine learning
- National Forest Inventory