Our research focuses on understanding and modeling how micro-scale processes modulate macro-scale soil carbon cycling, and how soils respond to climate and anthropogenic disturbances (e.g., fire, deforestation, agriculture). We combine dynamical systems theory, numerical simulations, climate model outputs, and data analytics, across spatiotemporal scales, towards improving the accuracy of regional and global carbon cycle projections.

---

THEMES & TOOLS

Process understanding
​across scales
- theory, data collection & synthesis, machine learning -

Model development &
benchmarking
- theory, numerical methods, machine learning, climate models -

Global change & 
​management implications
- data collection & synthesis, model projections, diverse disturbances -

---

RESEARCH AREAS & PROJECTS

Mineral-organic associations

Field observations suggest that more than half of soil organic carbon (SOC) is chemically or physically associated with soil minerals. These associations often limit microbial access to otherwise decomposable substrates, and consequently, mineral-associated organic carbon (MOC) can have much longer turnover times than particulate organic carbon. However, MOC can be vulnerable to change under novel conditions, and carbon storage in this pool may have a finite capacity that depends on the soil mineralogy. We use data synthesis and machine learning to explore the mineralogical capacity and controls of soil carbon storage, and to develop novel, spatially-explicit global estimates of soil carbon pools, which are essential for benchmarking process-based soil models.

Relevant publications: Georgiou et al. Nature Comms. (2022), Abramoff et al. Biogeochem. (2021), Sokol et al. Functional Ecol. (2022), Dwivedi et al. RiMG (2019), Georgiou et al. Biogeochem. (2021), Pellegrini et al. GCB (2021), Georgiou et al. Nature Geosci. (2024). 
Also see my 2022 AGU/ISCN webinar on soil carbon saturation.

Microbial community dynamics

Microbial communities play a dynamic and integral role in soil carbon cycling. At the pore-scale, complex interactions within and between the soil microbial community and mineral matrix can result in emergent dynamics and nonlinearities (e.g., density-dependent microbial mortality) not captured at broader scales. Trait- and individual-based models are increasingly being used to explore such interactions, but their complete representation in global climate models is not tractable. Machine learning-enabled surrogate models play an important role in bridging this gap towards computationally efficient and accurate projections across spatiotemporal scales. We explore how microbial community structure and function influence macro-scale soil carbon decomposition, and what effective representations emerge from upscaling micro-scale interactions in models.

Relevant publications: Georgiou et al. Nature Comms. (2017), Abramoff et al. SBB (2022), He et al. Matters Arising in Nature (2024).

Process-based soil models

Process-based soil models within global climate models remain highly uncertain and computationally expensive, especially at the high resolution needed for decision making. Our research spans theory-based structural improvements, model diagnostics and parameterization, and numerical methods. We develop data-driven surrogate models and machine learning (ML) enabled diagnostic tools to identify shortcomings of soil models and inform top-down improvements. The use of ML is crucial for revealing and interpreting higher-order and non-linear interactions that advance our understanding of soil carbon cycling and persistence. These insights are essential for refining process-based model formulations and parameterizations, guiding future experiments, and improving the accuracy of regional and global carbon cycle projections.

Relevant publications: Georgiou et al. Nature Comms. (2017), Georgiou et al. Comp. Geosci. (2018), Georgiou et al. Biogeochem. (2021), Sulman et al. Biogeochem. (2018), Abramoff et al. SBB (2022), Georgiou et al. Nature Geosci. (2024).

Global change & management

Soils are vulnerable to novel and increasingly severe environmental disturbances. Management practices that promote soil carbon sequestration are urgently needed to mitigate and adapt to climate change. How will soils respond to changes in environmental conditions, disturbance regimes, and management practices? Our research topics include the response of soil carbon and its underlying components to warming temperatures, changes in plant inputs (i.e., through management or elevated CO2), repeated fires, rewilding, and forest management.

Relevant publications: Georgiou et al. GCB (2015), Abramoff et al. GBC (2019), Walker et al. New Phyt. (2020), Pellegrini et al. GCB (2021) and Nature Geosci. (2022), Kristensen et al. TREE (2022).