Thomas Gumbricht
Researcher at the department of Physical Geography, Stockholm University with an interest in system science and AI based modeling applied to environmental studies. Coordinated the AI4SH in-situ data sampling methods and the development of databases and modeling tools for predicting soil health from different methods ranging from layperson to laboratory grade methods and instruments.
Sessions
Soils are complex ecosystems intrinsically interwoven with the surrounding environment. Local conditions such as topography, geomorphology, drainage, parent material, and microclimate are well-established drivers of soil formation, development, and function. Contemporary climate change constitutes a major global threat to soil functionality, although climatic variability has influenced pedogenesis throughout Earth history. Over millennia, and increasingly since the advent of mechanized agriculture and the widespread use of synthetic fertilizers and pesticides, agricultural intensification has become a dominant force shaping soil ecosystem processes. These transformations have degraded ecosystem integrity and functioning in many regions; within the European Union, an estimated 60–70% of soils are considered to be in poor health.
This article argues that understanding soil function and defining soil health must begin from two complementary perspectives: first, recognizing the hierarchical organization of patterns and processes operating both within soils and in relation to the broader environment; and second, applying a thermodynamic framework. The spatial scales relevant for soil ecosystem analysis span more than 50 orders of magnitude, from elementary particles to the observable universe, while temporal scales range from instantaneous radiation–matter interactions to the persistence of Precambrian paleosols. Across these spatial and temporal dimensions, soils have evolved through Darwinian selection favoring thermodynamically efficient dissipative systems capable of building internal organization while converting, regulating, and redirecting external energy and matter flows into negentropy. To improve the understanding, management, and optimization of soil health and ecosystem services, this article presents a conceptual framework linking soil functions to the spatial and temporal scales at which soils operate, together with the observational approaches required to assess relevant indicators across these scales.
Global soil degradation requires agricultural management practices that align with the soil processes evolved towards thermodynamic efficiency since the first soil ecosystems formed in the Precambrian. To empower farmers, advisors, and policymakers, there is a crucial need for rapid, cost-effective soil function and health assessment tools tailored to support agricultural management that make use of natural soil ecosystem efficiency for both enhancing productivity and improve farming profitability. The EU-funded AI4SoilHealth project explores a diverse set of monitoring methods across three user tiers - professionals, citizen scientists, and laypersons - tested in a dozen European countries. These methods address physical, chemical, and biological soil processes, ranging from traditional wet laboratory analyses of atomic and molecular content to novel, cutting-edge approaches including eDNA metabarcoding. Alternative, affordable, and rapid techniques evaluated include both laboratory and field spectrometry, enzymatic activity assays, single ring infiltration tests, microbial biomass quantifications, field- and home-adapted Ion-Selective Electrodes (ISE), penetrometers, soil respiration measurements, and app-based aggregate stability tests.
Using AI-driven modeling within an open framework, this study compares data from these diverse methods, assessing the accuracy and practical applicability of cost-effective, layperson, and citizen scientist approaches for in-field and home-based soil monitoring. Specific evaluations involve benchmarking wet chemistry, eDNA, and bulk density data against predictions derived from different spectrometric methods and varying sensor grades, assessing citizen scientist grade ISE sensors for pH and electrical conductivity, validating commercial kits for biomass estimations, correlating aggregate stability with bulk density, eDNA and infiltration, comparing field observed soil moisture with laboratory drying methods, etc. Several additional relationships are analyzed, summarized in an assessment of multi-tier methodologies for soil health indicators, emphasizing their varying accuracies and suitability.