This interactive web GIS tool presents 30-m resolution estimates of topsoil (0–20 cm) soil organic carbon (SOC) stocks across Florida’s grazing lands, together with associated prediction uncertainty expressed as a 90% Prediction Interval (PI) The 90% prediction interval indicates the range within which the true SOC stock is expected to fall with 90% confidence. . The maps provide a spatially explicit baseline for understanding SOC storage patterns and their uncertainty in subtropical grazing systems.

The mapped products support applications in carbon monitoring, land management, conservation planning, and research, and are intended to complement field observations and decision-making.

How the Maps Were Developed

SOC stocks were mapped using a parsimonious, uncertainty-aware digital soil mapping framework grounded in the STEP–AWBH conceptual framework. This approach explicitly operationalizes time-invariant soil-forming factors, grazing management intensity, and temporally dynamic environmental controls through spatiotemporal feature construction.

SOC stock predictions and associated uncertainty estimates were generated using Quantile Regression Forests, a probabilistic machine learning approach that enables estimation of both point predictions and prediction intervals. Model performance and uncertainty were evaluated using spatial cross-validation to account for spatial dependence in soil observations.

Key SOC Stock Estimates

• Florida’s grazing lands store an estimated 33.95 Tg of topsoil (0–20 cm) SOC.
• The mean SOC stock is 19.12 t ha^-1.

These estimates represent regional patterns and should be interpreted in conjunction with the uncertainty information provided.

How to Use the Map

This web GIS features a synchronized side-by-side map display to support transparent interpretation of SOC estimates and their uncertainty:

• Left panel: Topsoil SOC stock point estimates (0–20 cm) at 30 m resolution.
• Right panel: Corresponding 90% prediction interval (PI) width, representing spatial uncertainty in SOC estimates.

The two panels are spatially linked, allowing users to pan, zoom, and explore locations simultaneously. This design enables direct comparison of where SOC stocks are high or low and where uncertainty is greater or lower, supporting informed interpretation and decision-making.

Understanding Uncertainty

The 90% prediction interval (PI) represents the range within which the true SOC stock value is expected to fall with 90% probability, given the available data and model structure. Areas with wider PI values indicate higher uncertainty, which may arise from sparse sampling, high environmental variability, or complex soil–landscape interactions.

Users are encouraged to consider both SOC stock estimates and uncertainty information when applying these maps for analysis or decision support.

About the Data

This web GIS displays SOC stock and uncertainty maps that were developed using the following datasets:

• Legacy soil observations from the Florida Soil Carbon Project ( Metadata – Florida Carbon Project Lab Data - Pedometrics, Landscape Analysis and GIS Laboratory )
• Contemporary soil sampling was conducted across grazing lands in Florida.
• Environmental covariates representing soil, terrain, climate, water, vegetation, and management influences.

These datasets were used to train and evaluate the SOC prediction models. The web GIS provides interactive access to the resulting mapped products; raw input datasets are not directly visualized.

Funding sources

This research is supported by the 2022–2023 Florida State Legislative Budget AI-HARVEST program; Florida Milk Checkoff; Florida Cattle Enhancement Board (P0326003); USDA-NIFA Hatch Funds (FLA-AGR-006393); the UF/IFAS Archer Early Career Seed Grant (P00133052); and startup funds from the Florida Agricultural Experiment Station, UF/IFAS, University of Florida. The Florida Soil Carbon Project Dataset (FSCPD) was supported by USDA-CSREES-NRI grant 2007-35107-18368 (PI: Grunwald).

Acknowledgements

We thank the livestock producers and ranchers in the state of Florida for their generous permission to conduct the grazing lands soil sampling.

How to Cite This Map

If you use this map, database, or associated data products in your work, please cite:

Zhao, C., Song, J., Dubeux, J., Grunwald, S., Bretas, I. L., Liao, H.-Y., Tziolas, N., Harley, J. B., Zare, A., Babaeian, E., Garcia, L., Queiroz, L., & Mendes, C. T. E. (2026). Spatiotemporal controls on soil organic carbon stocks in subtropical grazing lands: An uncertainty-aware digital soil mapping approach. Available at SSRN: https://ssrn.com/abstract=6459841 or http://dx.doi.org/10.2139/ssrn.6459841

The dataset is archived on Zenodo with a DOI, and the corresponding code is hosted on GitHub. If you use this work, please cite it appropriately:

Zenodo Database

Zhao, C., Song, J., Dubeux, J., Grunwald, S., Bretas, I. L., Liao, H.-Y., Tziolas, N., Harley, J. B., Zare, A., Babaeian, E., Garcia, L., Queiroz, L., & Mendes, C. T. E. (2026). Topsoil Organic Carbon Stocks and Uncertainty in Florida Grazing Lands Derived from Quantile Regression Forest (30 m Resolution) (v1.0) [Dataset]. Zenodo. https://doi.org/10.5281/zenodo.19192952

GitHub

https://github.com/Ecosystem-Services-GeoAI/florida-grazing-soc-qrf

Web GIS Platform Ownership

• Ecosystem Services GeoAI Lab at Agronomy Department, UF/IFAS, University of Florida