Pith. sign in

REVIEW 4 major objections 4 minor 28 references

Enabling Adoption of Regenerative Agriculture through Soil Carbon Copilots

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read An AI soil carbon copilot claims county-scale insight into how tillage, drought, and compost move soil organic carbon.

desk verdict Useful system paper with a nice tillage map, but the empirical claims about SOC change outrun the evidence and need validation or reframing. read the letter →

arxiv 2411.16872 v2 pith:WIATGRTL submitted 2024-11-25 cs.IR cs.AIcs.ET

Margaret Capetz , Swati Sharma , Rafael Padilha , Peder Olsen , Jessica Wolk , Emre Kiciman , Ranveer Chandra This is my paper · ORCID
classification cs.IRcs.AIcs.ET
keywords soilorganiccarbonregenerativeagriculturelargelanguagemodelsretrievalaugmentedgenerationcoherentchangedetectiontillageremotesensingclimateresilience
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Regenerative agriculture can raise soil organic carbon, but measuring SOC cheaply over time and isolating the effects of practices from weather is hard. This paper claims an AI-driven Soil Organic Carbon Copilot can automate the ingestion of satellite imagery, weather records, farm management data, and SOC predictions to answer localized questions at county scale. Using California as a testbed, it finds evidence that diverse agricultural activity may soften the negative effects of tillage, that drought and wildfire dominate SOC change, and that composting may buffer SOC loss under extreme conditions. The value claimed is that agronomists, consultants, and policymakers can get tailored, data-grounded answers instead of generic statements about soil health.

What carries the argument

The machinery is the SOC Copilot's tool-augmented LLM agent: a GPT-4-Turbo core with role-specific system prompts for agronomist, farm consultant, and policymaker personas, which chooses among tools for SOC prediction, drought conditions, wildfire incidents, crop types, tillage scale, and support arguments. The two specialized models carrying the quantitative load are a SOC prediction model from Sharma et al. (2023b) that maps Sentinel-2 imagery, location, and digital elevation data to 50m SOC values for 2016 and 2023, and a tillage detector using coherent change detection, where interferometric coherence between two Sentinel-1 radar passes drops when soil is disturbed, with bare-soil index filtering and road removal used to suppress false positives. A retrieval-augmented generation framework, meaning the LLM pulls from external tools before answering, grounds the responses in these tables and in hand-selected soil science papers, which is what lets the same prompt produce localized numbers rather than textbook generalities.

What would settle it

Measure actual soil organic carbon in a sample of fields in the analyzed counties for 2016 and 2023 using physical soil sampling, then compare the measured changes with the copilot's predicted SOC changes; if the predicted gains and losses do not reproduce the measured direction and magnitude at field or county level, the tillage, drought, and composting conclusions would be artifacts of the SOC model rather than agronomic evidence.

Watch

Extended reading notes

Core claim

The central discovery is that an LLM agent that pulls from external tools, wired to a 50m SOC prediction model run over Sentinel-2 imagery for 2016 and 2023, a pixel-level tillage detector based on coherent change detection of Sentinel-1 radar, drought and wildfire records, crop type layers, and curated soil-science literature, can deliver county-specific analyses that a general LLM cannot. On comparison queries, the copilot cites observed SOC values (for example, San Joaquin falling from 3.886% to 2.644%, Merced from 2.85 to 2.61, and Sonoma rising from 1.79 to 2.06) and pairs them with drought, wildfire, tillage, and crop data to explain the trends. The paper's headline findings are that Tulare County maintains relatively high SOC despite high-intensity tillage while Monterey loses SOC under no-till, interpreted as diverse cropping systems offsetting tillage, and that Marin County's composting appears to buffer SOC loss relative to Riverside's planting under broadly similar drought but heavier wildfire pressure.

Load-bearing premise

The county-level SOC differences between 2016 and 2023 are treated as real changes in soil organic carbon, which assumes the underlying SOC prediction model is accurate and stable across California and across years without ground-truth validation, uncertainty quantification, or correction for imagery and model drift.

Editorial extensions

If this is right

  • County-scale SOC trends can be produced for regions with public satellite, weather, and crop data, reducing the need for dense field sampling in every location.
  • Diverse crop rotations may act as a management lever that offsets SOC losses from tillage, so conservation programs should look beyond tillage alone.
  • Extreme weather can dominate short-term SOC change, so carbon incentive programs should account for drought and wildfire context when judging practice effectiveness.
  • Compost application may help protect SOC against loss under drought and wildfire, suggesting compost support could be targeted to high-stress regions.
  • Role-specific personas make the same data usable by agronomists, farm consultants, and policymakers, lowering the barrier to evidence-based soil management.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The county-level comparisons are associative rather than causal, yet the copilot frames them as explanations; a stricter comparison that matches counties and controls for soil type, elevation, and crop mix would test whether the tillage and compost conclusions hold.
  • Because the 2016 and 2023 SOC values come from the same model without uncertainty bounds, the pipeline would need those bounds; if model error is comparable to the observed declines, the practice-effect claims could reverse.
  • The tillage detector is validated qualitatively on Washington winter wheat and visually in California, so a field-level validation with farmer-reported tillage across many California crops would check whether the county tillage scale is reliable enough to support the tillage finding.
  • Extending the copilot to other states with different climates and data sources would test whether the pattern that extreme weather outweighs regenerative practice is specific to California or general.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper presents an LLM-based "Soil Organic Carbon Copilot" that integrates public data (drought, wildfire, crop type), specialized machine-learning models (SOC prediction and coherence-change-detection-based tillage detection), and retrieval-augmented generation over selected soil-science literature. The system is demonstrated on county-level queries in California, comparing its answers with those of GPT-4 and providing role-specific personas for agronomists, farm consultants, and policymakers. The abstract and conclusion claim to find evidence that diverse agricultural activity may mitigate tillage effects and that composting may buffer SOC loss under extreme weather, based on county-level SOC changes between 2016 and 2023.

Significance. If the empirical findings were supported, the system would be a valuable tool for county-scale soil-health analysis and stakeholder engagement. The paper has real strengths: it integrates heterogeneous public data sources, uses RAG to ground LLM answers in scientific literature, and includes a tillage-detection validation on known fields in Washington. However, the central agronomic conclusions are not established by the presented analysis because they rest entirely on unvalidated model outputs and post hoc qualitative interpretation by the LLM. The paper's own statements—deferring quantitative evaluation to future work and acknowledging the absence of ground-truth tillage data in California—confirm that the evidence base is missing.

major comments (4)
  1. [Appendix A.1; Section 3] The county-level SOC changes used throughout Section 3 are point predictions from the Sharma et al. (2023b) model with no reported validation against ground-truth SOC, no uncertainty intervals, and no correction for differences in Sentinel-2 acquisition, phenology, or model drift between 2016 and 2023. Tables 7 and 8 treat differences such as Tulare's 5.58 to 5.48 versus Monterey's 2.39 to 2.00 as evidence about tillage and composting, but these deltas are unsecured model outputs. The abstract's "we find evidence" claims therefore do not follow from the presented analysis; at minimum the paper needs independent validation or must label these as illustrative model-based outputs rather than evidence.
  2. [Appendix A.3.3; Table 7] The tillage values used to compare Monterey (0.0) and Tulare (1.0) are generated by the CCD pipeline, and the appendix states that there are no ground-truth data for California pixels. The Washington field validation in Appendix A.3.2 shows false positives and date errors even in the labeled setting, so the county-level tillage labels should carry uncertainty. Without this, the claim that diverse agricultural activity mitigates tillage effects cannot be supported.
  3. [Section 3, Tables 7-8] The comparisons are post hoc narratives produced by the LLM from the same SOC model outputs and the same literature that were provided as inputs. There is no statistical test, no control for baseline SOC or soil type, and no adjustment for confounding between counties (e.g., crop mix, climate, wildfire history). A difference in point estimates between two counties cannot identify the effect of a practice; the paper should present a formal analysis or explicitly downgrade these conclusions to hypotheses.
  4. [Conclusion; Appendix A.3.3] The paper's own statements—"In future work, we will consider quantitative evaluation metrics" and "we do not have ground truth data for the pixels in the region"—confirm that the quantitative evaluation needed to support the empirical findings is absent. A system demonstration with qualitative examples is a reasonable contribution, but it does not justify the evidential language in the abstract and conclusions.
minor comments (4)
  1. [Introduction] The Introduction contains a typo: "serveral" should be "several."
  2. [Appendix A.2] References in Appendix A.2 use incomplete author labels ("pat (2021)", "Sal (2024)", "et al (2019)") and should be expanded to full citations.
  3. [Section 3] The text refers to "Table 3" when discussing regenerative-practice comparisons, but the relevant comparison appears in Table 8; the cross-reference should be corrected.
  4. [Tables 7-8] The tables would benefit from consistent formatting of SOC values (units and decimal places) and from error bars or uncertainty ranges if any are available.

Circularity Check

2 steps flagged · score 5.0 of 10

The agronomic 'findings' are presented as evidence but rest on the authors' own SOC model outputs and on RAG-ingested literature that already states the conclusions; the copilot architecture itself is not circular, but the empirical claims are largely a restatement of its inputs.

  1. self citation load bearing [Appendix A.1, used in Section 3 and the abstract (Tables 5-8)]
    "We run inference on a SOC prediction model Sharma et al. (2023b) for the state of California for the years 2016 and 2023 at a resolution of 50m. The model’s inputs include satellite imagery data from Sentinel-2, location information (latitude and longitude) and topology (DEM)."

    The paper's central empirical claims ('diverse agricultural activity may mitigate the negative effects of tillage', 'composting may mitigate SOC loss') are supported only by county-level SOC values produced by Sharma et al. (2023b), a model from the same research group (authors Swati Sharma, Rafael Padilha, Emre Kiciman, Ranveer Chandra). The present paper provides no ground-truth validation, no uncertainty quantification, and no correction for model drift between 2016 and 2023; Appendix A.1 simply averages per-image predictions into yearly pixel values. The 'evidence' is therefore not an independent measurement but an output of the authors' own prior model, which is a load-bearing self-citation rather than a minor one.

  2. renaming known result [Section 3 (Results and Discussion), Table 8 caption and surrounding text]
    "In this comparison, the SOC Copilot suggests compost is effective in protecting against SOC loss amidst extreme environmental conditions, which aligns with et al (2019)."

    The 'compost may mitigate SOC loss' insight is presented as a copilot discovery, but the supporting 'alignment' is Tautges et al. (2019), one of the hand-selected articles ingested into the copilot's RAG context (Appendix A.2, Table 1). The Riverside-vs-Marin SOC numbers are the authors' own model predictions, and the copilot's 'Support Arguments' tool retrieves literature to support a hypothesis. The conclusion is thus a restatement of the system's input literature and input model outputs rather than an independent empirical result; the retrieval of a known result is reported as evidence, which is a form of renaming the input as a finding.

full rationale

The copilot architecture itself is not circular: it genuinely integrates public drought, wildfire, and crop data with specialized models to produce localized answers, and the system demonstration has independent value. The circularity burden falls on the paper's agronomic 'findings'. The abstract and Section 3 treat differences in county-level SOC predictions from Sharma et al. (2023b) as though they were measurements of real SOC change, but Appendix A.1 gives no validation against ground truth, no error bars, and no adjustment for imagery or model drift between years. This is partly a correctness risk, but it is also a self-citation problem because the SOC model comes from the same authors and is the sole source of the SOC values used in every empirical comparison. In addition, the 'compost buffers SOC loss' conclusion is explicitly aligned with a Tautges et al. (2019) paper that was hand-selected and supplied to the copilot as RAG context, making the claimed discovery a restatement of the system's own curated literature. The paper honestly notes future quantitative evaluation and the lack of tillage ground truth in California (Appendix A.3.3), but those limitations do not remove the circular support for the headline findings. Given that the central claims are not derived from independent measurements and are substantially forced by the authors' model outputs and RAG inputs, a moderate circularity score is appropriate.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claims rest on two model pipelines (SOC prediction and tillage detection) with several hand-chosen thresholds and no released validation. The agronomic inferences add domain assumptions about what county-level differences mean. No new physical entities are introduced.

free parameters (4)
  • SOC model learned parameters = Not given in this paper (from Sharma et al. 2023b)
    All county SOC values and changes come from this model; no validation or error bars for California 2016/2023.
  • Bare Soil Index threshold = 0.06
    Hand-chosen in Appendix A.3.3 to flag bare soil for tillage detection; directly controls which pixels are called tilled.
  • Minimum tillage patch dimension = 3 pixels
    Hand-chosen filter in Appendix A.3.3 to remove road-like line artifacts from the tillage map.
  • Maximum SAR baseline for coherence pairs = 100 m
    Hand-chosen in Appendix A.3.1 to limit geometric decorrelation; affects which Sentinel-1 pairs are used.
assumptions (5)
  • standard math Coherent Change Detection coherence loss indicates physical change in the scattering surface (Eq. 1).
    Relied on in Appendix A.3.1 to justify tillage detection; standard in InSAR/CCD literature.
  • domain assumption The SOC prediction model (Sharma et al. 2023b) generalizes to California for 2016 and 2023 without local revalidation.
    Assumed throughout Section 3 and Appendix A.1; the paper provides no validation of these specific predictions.
  • domain assumption County-level differences in model-predicted SOC between 2016 and 2023 represent real changes in soil organic carbon.
    This assumption underlies every comparison in Tables 5-8; no uncertainty or calibration is reported.
  • domain assumption Tillage detection via BSI > 0.06 plus CCD identifies actual tillage events in California fields.
    Appendix A.3.3 explicitly states no ground truth is available for California pixels; the map is visually inspected instead.
  • domain assumption The hand-selected scientific articles in Table 1 are sufficient to ground causal claims about SOC.
    The RAG system retrieves arguments from these papers, and the paper treats those as support for the copilot's conclusions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enabling Adoption of Regenerative Agriculture through Soil Carbon Copilots." pith.science (2026). https://pith.science/paper/WIATGRTL

@misc{pith2026241116872,
  author       = {Pith},
  title        = {Pith review of: Enabling Adoption of Regenerative Agriculture through Soil Carbon Copilots},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WIATGRTL}},
  note         = {Machine review of arXiv:2411.16872}
}
read the original abstract

Mitigating climate change requires transforming agriculture to minimize environ mental impact and build climate resilience. Regenerative agricultural practices enhance soil organic carbon (SOC) levels, thus improving soil health and sequestering carbon. A challenge to increasing regenerative agriculture practices is cheaply measuring SOC over time and understanding how SOC is affected by regenerative agricultural practices and other environmental factors and farm management practices. To address this challenge, we introduce an AI-driven Soil Organic Carbon Copilot that automates the ingestion of complex multi-resolution, multi-modal data to provide large-scale insights into soil health and regenerative practices. Our data includes extreme weather event data (e.g., drought and wildfire incidents), farm management data (e.g., cropland information and tillage predictions), and SOC predictions. We find that integrating public data and specialized models enables large-scale, localized analysis for sustainable agriculture. In comparisons of agricultural practices across California counties, we find evidence that diverse agricultural activity may mitigate the negative effects of tillage; and that while extreme weather conditions heavily affect SOC, composting may mitigate SOC loss. Finally, implementing role-specific personas empowers agronomists, farm consultants, policymakers, and other stakeholders to implement evidence-based strategies that promote sustainable agriculture and build climate resilience.

Figures

Figures reproduced from arXiv: 2411.16872 by the authors.

Figure 1
Figure 1. Tillage detection for a region of central California for a Sentinel-2 tile (11SKA) which is [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. SOC Copilot Architecture. The copilot processes user queries containing location data (e.g., county names or coordinates) and customizes responses using role-based personas. It retrieves and analyzes relevant data using multi-resolution, multi-modal tools. loss, shown in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Comparison of SOC Copilot and GPT-4 Responses. For a query comparing regenerative [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Excerpts of stakeholder-specific SOC Copilot responses. The Agronomist, Farm Consultant, [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Tillage detection for 4 fields in Washington state. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: The textual data includes recent scientific literature on how different factors influence SOC. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 24 canonical work pages

  1. [1]

    Soil texture and environmental conditions influence the biogeochemical responses of soils to drought and flooding. 2021. URL https://doi.org/10.1038/s43247-021-00198-4

    work page doi:10.1038/s43247-021-00198-4 2021

  2. [2]

    Impact of wildfire recurrence on soil properties and organic carbon fractions. 2024. doi:10.1016/j.jenvman.2024.120293

    work page arXiv 2024

  3. [3]

    Sustainability ai copilot: Analyze & ideate at scale to enable positive impact

    Rajagopal A, Nirmala V, Immanuel Raja, and Arun V. Sustainability ai copilot: Analyze & ideate at scale to enable positive impact. In NeurIPS 2023 Workshop on Tackling Climate Change with Machine Learning, 2023. URL https://www.climatechange.ai/papers/neurips2023/118

    work page 2023

  4. [4]

    Farmer perspectives on carbon markets incentivizing agricultural soil carbon sequestration , 2023

    Barbato, C.T., Strong, A.L. Farmer perspectives on carbon markets incentivizing agricultural soil carbon sequestration , 2023. URL https://doi.org/10.1038/s44168-023-00055-4

    work page doi:10.1038/s44168-023-00055-4 2023

  5. [5]

    Humberto Blanco-Canqui and R. Lal. No-tillage and soil-profile carbon sequestration: An on-farm assessment. Soil Science Society of America Journal, 72 0 (3): 0 693--701, 2008. doi:https://doi.org/10.2136/sssaj2007.0233. URL https://acsess.onlinelibrary.wiley.com/doi/abs/10.2136/sssaj2007.0233

    work page arXiv 2008

  6. [6]

    Monitoring us agriculture: the us department of agriculture, national agricultural statistics service, cropland data layer program

    Claire Boryan, Zhengwei Yang, Rick Mueller, and Mike Craig. Monitoring us agriculture: the us department of agriculture, national agricultural statistics service, cropland data layer program. Geocarto International, 26 0 (5): 0 341--358, 2011

    work page 2011

  7. [7]

    Healthy soils program - incentives program

    California Department of Food and Agriculture . Healthy soils program - incentives program. https://www.cdfa.ca.gov/oefi/healthysoils/IncentivesProgram.html, 2024. Accessed: 2024-08-16

    work page 2024

  8. [8]

    Fire incidents

    California Department of Forestry and Fire Protection . Fire incidents. https://www.fire.ca.gov/incidents, 2024. Accessed: 2024-08-22

    work page 2024

Show all 28 references
  1. [9]

    Deep soil inventories reveal that impacts of cover crops and compost on soil carbon sequestration differ in surface and subsurface soils

    Tautges et al. Deep soil inventories reveal that impacts of cover crops and compost on soil carbon sequestration differ in surface and subsurface soils. Global Change Biology, 2019. URL https://doi.org/10.1111/gcb.14762

    2019 doi

  2. [11]

    Daly, Thomas K

    Keunbae Kim, Erin J. Daly, Thomas K. Flesch, Trevor W. Coates, and Guillermo Hernandez-Ramirez. Carbon and water dynamics of a perennial versus an annual grain crop in temperate agroecosystems. Agricultural and Forest Meteorology, 314: 0 108805, 2022. ISSN 0168-1923. doi:https...

    2022

  3. [13]

    R. Lal. Soil carbon sequestration to mitigate climate change. Geoderma, 123 0 (1): 0 1--22, 2004. ISSN 0016-7061. doi:https://doi.org/10.1016/j.geoderma.2004.01.032. URL https://www.sciencedirect.com/science/article/pii/S0016706104000266

    2004 doi

  4. [14]

    Impact of wildfire on soil carbon and nitrogen storage and vegetation succession in the nanweng'he national natural wetlands reserve, northeast china

    Xiaoying Li, Huijun Jin, Ruixia He, Hongwei Wang, Long Sun, Dongliang Luo, Yadong Huang, Yan Li, Xiaoli Chang, Lizhong Wang, and Changlei Wei. Impact of wildfire on soil carbon and nitrogen storage and vegetation succession in the nanweng'he national natural wetlands reserve, ...

    2023

  5. [15]

    Lowder, Jakob Skoet, and Terri Raney

    Sarah K. Lowder, Jakob Skoet, and Terri Raney. The number, size, and distribution of farms, smallholder farms, and family farms worldwide. World Development, 87: 0 16--29, 2016. ISSN 0305-750X. doi:https://doi.org/10.1016/j.worlddev.2015.10.041. URL https://www.sciencedirect.c...

    2016 doi

  6. [16]

    Sentinel-2 l2a

    Microsoft Planetary Computer . Sentinel-2 l2a. https://planetarycomputer.microsoft.com/dataset/sentinel-2-l2a, n.d. Accessed: 2024-08-29

    2024

  7. [17]

    National Drought Mitigation Center, University of Nebraska-Lincoln . U.s. drought monitor. https://droughtmonitor.unl.edu/CurrentMap.aspx, 2024. Accessed: 2024-08-16

    2024

  8. [18]

    Coherent change detection: Theoretical description and experimental results

    Mark Preiss and Nicholas JS Stacy. Coherent change detection: Theoretical description and experimental results. Technical report, 2006

    2006

  9. [19]

    How much of the world's food do smallholders produce? Global Food Security, 17: 0 64--72, 2018

    Vincent Ricciardi, Navin Ramankutty, Zia Mehrabi, Larissa Jarvis, and Brenton Chookolingo. How much of the world's food do smallholders produce? Global Food Security, 17: 0 64--72, 2018. ISSN 2211-9124. doi:https://doi.org/10.1016/j.gfs.2018.05.002. URL https://www.sciencedire...

    2018 doi

  10. [20]

    Monitoring autumn agriculture activities using synthetic aperture radar (sar) and coherence change detection

    Laura Dingle Robertson, Heather McNairn, Marco van der Kooij, Xianfeng Jiao, Samuel Ihuoma, and Pamela Joosse. Monitoring autumn agriculture activities using synthetic aperture radar (sar) and coherence change detection. Heliyon, 9 0 (6), 2023

    2023

  11. [21]

    Gmtsar: An insar processing system based on generic mapping tools

    David Sandwell, Rob Mellors, Xiaopeng Tong, Matt Wei, and Paul Wessel. Gmtsar: An insar processing system based on generic mapping tools. 2011

    2011

  12. [22]

    Sentinel-1 & sentinel-2 data for soil tillage change detection

    Giuseppe Satalino, Francesco Mattia, Anna Balenzano, Francesco P Lovergine, Michele Rinaldi, Angelo Pio De Santis, Sergio Ruggieri, DA Nafr \' a Garc \' a, V Paredes G \'o mez, Eric Ceschia, et al. Sentinel-1 & sentinel-2 data for soil tillage change detection. In IGARSS 2018-...

    2018

  13. [23]

    Perennial cropping systems increased topsoil carbon and nitrogen stocks over annual systems—a nine-year field study

    Yiwei Shang, Jørgen Eivind Olesen, Poul Erik Lærke, Kiril Manevski, and Ji Chen. Perennial cropping systems increased topsoil carbon and nitrogen stocks over annual systems—a nine-year field study. Agriculture, Ecosystems & Environment, 365: 0 108925, 2024. ISSN 0167-8809. doi...

    2024

  14. [24]

    Knowledge guided representation learning and causal structure learning in soil science, 2023 a

    Somya Sharma, Swati Sharma, Licheng Liu, Rishabh Tushir, Andy Neal, Robert Ness, John Crawford, Emre Kiciman, and Ranveer Chandra. Knowledge guided representation learning and causal structure learning in soil science, 2023 a . URL https://arxiv.org/abs/2306.09302

    2023 arXiv

  15. [25]

    Domain adaptation for sustainable soil management using causal and contrastive constraint minimization

    Somya Sharma, Swati Sharma, Rafael Padilha, Emre Kiciman, and Ranveer Chandra. Domain adaptation for sustainable soil management using causal and contrastive constraint minimization. In NeurIPS 2023 Workshop on Tackling Climate Change with Machine Learning, 2023 b . URL https:...

    2023

  16. [26]

    Drought effects on soil organic carbon under different agricultural systems

    Pedro R Soares, Matthew T Harrison, Zahra Kalantari, Wenwu Zhao, and Carla S S Ferreira. Drought effects on soil organic carbon under different agricultural systems. Environmental Research Communications, 5 0 (11): 0 112001, nov 2023. doi:10.1088/2515-7620/ad04f5. URL https://...

    2023 doi

  17. [27]

    Tzachor, M

    A. Tzachor, M. Devare, C. Richards, P. Pypers, A. Ghosh, J. Koo, S. Johal, and B. King. Large language models and agricultural extension services. Nature Food, 4 0 (11): 0 941--948, 2023. doi:10.1038/s43016-023-00867-x. URL https://doi.org/10.1038/s43016-023-00867-x

    2023 doi

  18. [28]

    Department of Agriculture, National Agricultural Statistics Service

    U.S. Department of Agriculture, National Agricultural Statistics Service . Cropscape - cropland data layer. https://catalog.data.gov/dataset/cropscape-cropland-data-layer, 2024. Accessed: 2024-08-16

    2024

  19. [29]

    Exploring the impact of artificial intelligence on sustainable agriculture: A case study of precision farming

    Anil Walke, P.K.Srivastava, and Deepak Sharma. Exploring the impact of artificial intelligence on sustainable agriculture: A case study of precision farming. 2022. URL https://api.semanticscholar.org/CorpusID:264351864

    2022

  20. [30]

    Amgain, Abul Rabbany, Jay Capasso, Kevin Korus, Stewart Swanson, and Jehangir H

    Nan Xu, Naba R. Amgain, Abul Rabbany, Jay Capasso, Kevin Korus, Stewart Swanson, and Jehangir H. Bhadha. Interaction of soil health indicators to different regenerative farming practices on mineral soils. Agrosystems, Geosciences & Environment, 5 0 (1): 0 e20243, 2022. doi:htt...

    2022 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.