An updated assessment of US Soil Organic Carbon to enhance Earth system simulations

The carbon content of the Soil is approximately twice that of the atmosphere and plants put together. With the ability to take in more carbon dioxide from the atmosphere than it emits, it is a significant carbon sink. Not only is soil carbon management essential to soil health and agricultural output, but it is also a critical component of efforts to mitigate climate change.

However, measuring soil carbon is a time-consuming and costly procedure. Upscaling measurements on a broad spatial scale is difficult since samples need to be excavated and transferred to a lab for analysis.

Environmental scientists have now estimated soil Organic Carbon at the continental level of the United States by combining machine learning techniques with data collected at the field level. The revised estimate of soil organic carbon, which provides new information on the effects of environmental variables on soil organic carbon and enhances the estimate for the United States overall, was published in the Journal of Geophysical Research: Biogeosciences.

"Investing in the enhancement of soil organic carbon through sustainable land management practices is becoming increasingly important," states Debjani Sihi, an assistant professor of environmental sciences at Emory University and the study's senior author. "Our estimate is more accurate than previous estimates and provides a better baseline to guide policymakers and managers of land in adopting environmentally conscious practices."

Sihi points out that land retains carbon significantly more effectively than the ocean does, and this suggests one potential natural way to slow down climate change.

"We could potentially create conditions," she continues, "that are favorable for soil to capture carbon dioxide from the atmosphere and lock it there for a really long time—over millennia."

Sihi is a biogeochemist who focuses on sustainability and environmental problems at the intersection of soil and climate.

Zhuonan Wang, a former postdoctoral fellow in Sihi's group who is currently at Colorado State University, is the first author of the current paper.

Examining soil data in depth

The organic carbon found in soil is composed of decomposed plant and animal debris in different stages of breakdown. Though carbonate minerals contain inorganic carbon as well, organic carbon often makes up the majority and is the primary factor influencing soil biology and quality.

The soil characterization database for the National Cooperative Soil Survey is kept up to date by the U.S. Department of Agriculture. Over several decades, this data was collected by both exploring the terrain and making observations, as well as by excavating core samples and shipping them to labs for examination. For example, to measure soil organic carbon, a core must be dug to the root zone, approximately 30 centimeters deep to acquire a topsoil profile, and deeper until the core reaches bedrock to produce a complete soil profile.

Other regions in the world also sample their soil. Over 430,000 soil profiles from all around the world are included in the International Soil Organic Carbon Network. These data are used by scientists to produce "soil maps," or assessments of the properties of the soil in different locations. The Harmonized World Soil Database, created by the Food and Agricultural Organization of the United Nations and its partners, is one well-known soil map. Another is SoilGrids, which is backed by the Netherlands-based International Soil Reference and Information Center.

There are notable disparities in the soil organic carbon values found in SoilGrids and the Harmonized World Soil Database. Sihi and her team set out to explore if they could identify more efficient methods of scaling up the soil-sampling data, and thereby address these discrepancies within the US estimations.

The US, comprising all 50 states as well as Puerto Rico, was split up into 20 regions by the researchers, who then developed machine-learning models for each one. They collected around fifty thousand soil samples from various areas, ranging in depth from thirty centimeters to one meter. These data samples for soil organic carbon, paired with exact geographic information system locations, were used to build their algorithms.

To provide their models with 36 environmental variables, such as information about the climate, topographical features of the land, soil biogeochemical qualities, and the amount of flora on the terrain, they also drew from additional open-source data.

An improved reference point for Earth system modeling

The results demonstrated that, for the top 30 centimeters of soil, where the majority of biologically active soil organic carbon tends to be concentrated, the novel method produced estimates that were more accurate than those supplied by SoilGrids and the Harmonized World Soil Database.

The novel approach also demonstrated the regional variations in the impacts of environmental factors on soil organic carbon. In the majority of the locations, soil organic carbon was most commonly predicted by climate; however, in the arid southwest, the vegetation index tended to be more significant. In areas that included a large river delta or were mountainous, elevation was a crucial factor.

The researchers expect that additional nations and continents with sufficient amounts of field data will adopt their methodology.

"The beauty of our approach is that it gives us the power to identify regions with high uncertainty in our estimates and that helps us to guide future sampling efforts," Sihi explains.

The new model is more flexible when environmental factors are taken into account since rising global temperatures due to climate change cause soils to warm and modify rainfall patterns. Sihi points out that it's still uncertain if soils will keep acting as carbon sinks or turn into carbon sources.

To comprehend how the carbon content of the soil will change under climate change, we first need accurate estimates of present soil organic carbon levels and the key variables that influence them," explains Sihi. "Our new estimate is a step towards getting more accurate baseline information to improve Earth system simulations for climate change."

Umakant Mishra (Sandia National Laboratories), Katharine Todd-Brown (University of Florida), Samantha Weintraub-Leff (National Ecological Observatory Network), and Jitendra Kumar (Oak Ridge National Laboratory) are co-authors of the updated estimate.