Pith. sign in

REVIEW 4 major objections 5 minor 105 references

PIER scores retrieval candidates by physical consistency with the target lake, beating strong baselines across 356 lakes.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 10:23 UTC pith:365I7WQK

load-bearing objection Real gains from a novel physics-aware retrieval wrapper, but the cross-lake transfer of local verifiers needs direct validation before the abstract's 'consistent' outperformance claim holds. the 4 major comments →

arxiv 2607.20230 v1 pith:365I7WQK submitted 2026-07-22 cs.LG

PIER: Physics-Informed Environmental Retrieval for Time-Series Modeling

Shiyuan Luo , Runlong Yu , Chonghao Qiu , Yue Qin , Rahul Ghosh , Robert Ladwig , Paul C. Hanson , Yiqun Xie
show 1 more author
Xiaowei Jia
This is my paper
classification cs.LG
keywords retrieval-augmented time seriesphysics-informed retrievallocal verifiersflux-response consistencylake water temperaturedissolved oxygenknowledge transferscenario adaptation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

PIER sets out to fix a weakness in retrieval-augmented time-series modeling for environmental systems: embeddings can match scenarios that look alike statistically while behaving differently physically. Its answer is a second retrieval stream that scores each candidate by 'flux-response consistency'—applying the target lake's own physics-trained local verifier to the candidate's flux features and checking how well it predicts the candidate's observations. A learned gate then decides per scenario how much to trust this physics stream versus ordinary embedding similarity, and the merged, reranked candidates are used to fine-tune a scenario-specific predictor. The paper reports that on 356 Midwestern US lakes over 41 years, this framework lowers error for both water temperature and dissolved oxygen across all test settings, and that it works as a general augmentation on several model backbones. A sympathetic reader would take the central claim to be that physical-consistency filtering, not model complexity, is the key to transferring knowledge across sparse environmental systems.

Core claim

PIER's central discovery is that physical consistency between a target system and a candidate system can be measured directly, without needing a perfect global physical model: train a cheap local verifier on the target system's physics-derived flux variables alone, then score any candidate by the verifier's ability to predict the candidate's observed target values (Eq. 3.2). Candidates whose dynamics are well reproduced get high scores; candidates that behave differently are demoted even if their embeddings look similar. The paper shows this physics stream and the embedding stream are complementary, and that a per-scenario learned gate—trained to predict, from diagnostic features, the best b

What carries the argument

The load-bearing object is the local dynamics verifier bank {g_ℓ}: for each lake, a small model trained only on that lake's physics-derived flux variables. It powers the physical-consistency score s^phys_{q,i} = exp(−(1/τ)·MSE(g_{ℓ_q}(x^flux_i), y_i)), which rates a candidate by whether the target's dynamics can reproduce the candidate's observed behavior. This score forms the second retrieval stream, merged with embedding cosine similarity, and a lightweight gate network h_φ predicts the blend weight γ_q from three diagnostic features (local-verifier quality, candidate pass-rate, stream-agreement ratio). The gate is trained to minimize validation prediction error after fine-tuning on the re

Load-bearing premise

The load-bearing premise is that a local verifier trained on one lake's physics-flux variables is a reliable judge of whether another lake shares the target's dynamics; if cross-system predictability fails, the physics stream will systematically mis-rank candidates.

What would settle it

Across all 356 lakes, rank candidates by the change in actual prediction error they produce when added to the target's fine-tuning set, and compare that ordering to PIER's physical-consistency scores; if the scores do not positively rank-order the beneficial candidates, the physics stream is not the mechanism behind the reported gains.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Matching lakes by shared physical dynamics, not just embedding similarity, is what drives the main error reductions; the case studies show phys-promoted candidates fixing failures that embedding retrieval introduces.
  • Because the framework is model-agnostic, it improves several backbone architectures (an LSTM, Transformer variants, and an MLP mixer) when wrapped as a plug-in augmentation.
  • On sparse observational data, data usage beats model complexity: a simple LSTM with PIER beats more complex models without retrieval.
  • Physics-derived simulation labels and flux features become substantially more useful when routed through local verifiers than when simply concatenated into inputs.
  • No fixed mixture of the two retrieval streams is best; the learned per-scenario weight is required for peak performance across tasks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Because the flux-response consistency test only needs a process model that emits flux variables, the same design could transfer to agriculture, hydrology, or carbon-cycle forecasting—domains the paper does not test.
  • The local-verifier cross-prediction is a pairwise similarity measure; one could assemble it into a full system-to-system graph and do retrieval or clustering without per-domain gate training.
  • A side diagnostic: the learned gate weight γ_q may itself be an interpretable index of how trustworthy a lake's physics simulations are—near 1 for well-observed lakes with reliable fluxes, near 0 for sparse or biased ones.
  • An untested but natural ablation would be to replace the physics flux features in the local verifiers with raw forcing features; if gains vanish, the internal physics variables are the true carriers of the transfer signal.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. PIER is a retrieval-augmented framework for environmental time-series prediction. It augments embedding-based retrieval with a physics-aware stream: local verifiers trained on physics-derived flux features score candidate scenarios by how well the target's verifier predicts the candidate's observations (Eq. 3.2). A learned gate (Eqs. 3.7–3.8) combines the two retrieval scores per scenario, and the top-ranked scenarios are used to fine-tune a scenario-specific predictor initialized from a global model. The experiments cover 356 Midwestern U.S. lakes over 41 years for dissolved oxygen and water temperature. Table 1 shows PIER achieving the lowest RMSE among the selected non-retrieval baselines in all six settings, and the paper claims this is a general augmentation strategy across backbones.

Significance. If the local-verifier transfer assumption is valid, PIER is a practically useful, model-agnostic way to inject physics consistency into retrieval for sparse environmental data. The paper benefits from a substantial real-world benchmark, a model-agnostic design, a clear ablation structure, and case studies. The gate avoids expensive second-order optimization, so the method is deployable. However, the load-bearing assumption that a local verifier trained on one lake transfers as a physical-consistency detector to other lakes is not directly validated, and the baseline set omits retrieval-augmented competitors. The current evidence supports a narrower claim than the abstract states: PIER improves over the specific non-retrieval baselines in Table 1, but not yet that physics-aware retrieval is the cause of the gains.

major comments (4)
  1. [§3.2, Eq. (3.2)] The physical-consistency score assumes that a local verifier g_l trained on lake l's flux features and observations transfers to candidate lake i as a detector of shared dynamics. This assumption is not tested. Because all flux features come from a shared process-based model, low Eq. (3.2) error may reflect common simulator bias or verifier artifact rather than mechanistic similarity, and sparse |T_i| makes the RMSE noisy. Please add a control experiment: e.g., compare Eq. (3.2) rankings against known physical similarity (morphometry, stratification regime, climate), or permute flux features/observations and show that the physics score loses signal. Without this, the E5-vs-E4 comparison cannot establish that gains arise from physics awareness rather than from the retrieval/fine-tuning machinery.
  2. [§4.4, Table 1] No retrieval-augmented baseline is included in the main comparison. Table 1 compares PIER against models trained without retrieval, so the improvement could be due to fine-tuning on retrieved candidates rather than to retrieval quality. The strongest controlled contrast, E1 (simple retrieval), appears only as a figure in §4.6 with no numeric values. Report numeric RMSE for E0, E1, E4, and E5 (and preferably also a standard embedding retrieval baseline) in Table 1 or a companion table. This is necessary to separate the contribution of retrieval from the contribution of the physics stream.
  3. [§3.3.3, Eqs. (3.7)–(3.8)] The gate h_phi is trained to reproduce gamma* obtained by minimizing the outer loss on each training scenario's own labels. At test time only diagnostic features are available, so the central question is whether h_phi generalizes. The paper does not report gate prediction error, the distribution of gamma*, or how sensitive test RMSE is to gamma. If the gate is poorly calibrated, the claimed per-scenario adaptation is not demonstrated. Please report gate accuracy/error on held-out scenarios and a sensitivity analysis over gamma in [0,1] for representative cases.
  4. [§4.4.2, Fig. 3; Abstract] The abstract and conclusion claim that PIER 'consistently outperforms' baselines and 'serves as a general augmentation strategy,' but §4.4.2 states that 'In tasks such as DO prediction with iTransformer, PIER does not improve over the physics-enhanced setting.' Fig. 3 is reported only graphically, without numeric RMSE values. This makes the scope of the claim unverifiable. Provide the numeric backbone-by-backbone results, and either soften the abstract/conclusion to 'almost all settings' or explain why the iTransformer-DO exception does not contradict the stated generality.
minor comments (5)
  1. [Table 1] The header contains a typo: 'W ater temperature' should be 'Water temperature.'
  2. [§3.2.2, Eq. (3.2)] The score is undefined or extremely noisy when |T_i| is very small. Please state the minimum observation count required for a candidate to be scored, or describe how sparse candidates are handled (e.g., ignored, pooled, or penalized).
  3. [§3.3, Eq. (3.3)] The representation e_q used in s_rep is not precisely defined. Specify which hidden-state layer or pooling operation produces e_q from the global model.
  4. [§4.6, Fig. 6] The ablation results are presented only as figures without numeric values. Please include a table with the exact RMSE and standard deviations for E0–E5, at least in an appendix.
  5. [§4.1] More detail on the local verifier architecture and training (e.g., input flux variables per target variable, optimization, and observation availability) would aid reproducibility. The paper says only that each g_l is trained 'independently for each system' without specifying the model class.

Circularity Check

0 steps flagged

No circular derivation found: the physics-aware stream is an empirical transfer measure and the gamma gate is standard amortized hyperparameter selection, not a reduction of the target to its inputs.

full rationale

PIER's derivation chain is not circular. The physical-consistency score (Eq. 3.2) is an out-of-sample transfer error: the target lake's local verifier g_lq, trained on that lake's flux features and observations, is applied to candidate flux features and scored against candidate observations. This is an empirical statistic, not a quantity defined in terms of the final prediction target. Claiming high scores denote physical consistency is an interpretation, and it is tested against external baselines (Table 1) and ablations E0-E5, so the central claim is not tautological. The two-phase gamma selection (Sec. 3.3.3) is also not circular in the prohibited sense: gamma* is obtained by evaluating the full inner loop on training scenarios' own labels (Eq. 3.7), and a gate is trained to predict those values from diagnostic features z_q that do not require target observations at test time (Eq. 3.8). This is amortized hyperparameter optimization, not 'prediction equals fit'; the final test predictions come from the fine-tuned predictor d_theta on the retrieved set, not from the gate's training target. The model-quality safeguard uses held-out validation error of the simulator versus the learned verifier, again an empirical control. Self-citations (e.g., refs. 15,16,31,33) contextualize related work and do not carry the method's correctness; no uniqueness theorem is imported. One acknowledged limitation--'In tasks such as DO prediction with iTransformer, PIER does not improve over the physics-enhanced setting' (Sec. 4.4.2)--qualifies the abstract's 'consistently outperforms' and 'general augmentation strategy,' but this is a performance-claim issue, not a circularity issue. No step of the derivation reduces, by definition or by self-citation, to its own input.

Axiom & Free-Parameter Ledger

5 free parameters · 4 axioms · 0 invented entities

The central claim rests on the fidelity of simulator-derived flux features, the transferability of single-lake verifiers, and generalization of the learned gate; none of these are independently verified, and all three are domain assumptions rather than standard mathematics.

free parameters (5)
  • Retrieval set size K = 50
    Tuned on validation (§4.3); controls the size of the candidate pool from each retrieval stream.
  • Scaling parameter tau = 10
    Tuned on validation (§4.3); temperature of the exponential physical-consistency score in Eq. 3.2.
  • Gamma grid G = not specified
    Discrete grid over [0,1] used to select per-scenario gamma* in Eq. 3.7; the specific grid values are not reported.
  • Gate h_phi parameters = learned
    MLP trained to predict gamma* from three diagnostic features (§3.3.3).
  • Model-quality safeguard threshold = 0
    Physics stream is disabled when mu_l <= 0 (§3.2.3); the threshold is a hand-chosen decision boundary.
axioms (4)
  • domain assumption Process-based model flux features faithfully represent lake physics for all systems and times.
    Invoked in §2.2 and §4.1: flux features are the sole physics signal; if they are biased, the physics stream mis-ranks candidates.
  • ad hoc to paper A local verifier trained on one lake's flux features and sparse observations predicts other lakes' observations when dynamics are shared.
    Core scoring assumption in Eq. 3.2; the claim that verifier prediction skill is a physical-consistency measure is not independently established.
  • ad hoc to paper The gate h_phi generalizes from training scenarios (where gamma* is obtained with label access) to test scenarios.
    Needed for adaptive weighting to work at test time (§3.3.3); no generalization gap analysis is provided.
  • domain assumption Retrieving training scenarios and fine-tuning a global model on them improves target predictions.
    Standard retrieval-augmented modeling assumption, relied on in §3.3.2 and §3.4.

pith-pipeline@v1.3.0-alltime-deepseek · 11659 in / 16350 out tokens · 128943 ms · 2026-08-01T10:23:57.772537+00:00 · methodology

0 comments
read the original abstract

Accurate modeling of environmental systems is fundamental to scientific understanding and decision-making, yet remains challenging because observations are limited and physical dynamics vary across systems. Retrieval-augmented approaches offer a natural path to transfer knowledge across systems, but standard embedding-based retrieval does not guarantee consistency of underlying physical processes, since scenarios with similar embeddings may arise from different underlying mechanisms. We propose Physics-Informed Environmental Retrieval (PIER), a model-agnostic framework that augments embedding-based retrieval with a physics-aware stream that scores candidates by flux-response consistency with the target, using local verifiers trained on physics-derived flux features. A weight adjustment mechanism then learns per-scenario weights that adaptively balance the two retrieval streams based on diagnostic features summarizing physics-stream reliability. Experiments on 356 lakes across the Midwestern United States spanning 41 years show that PIER consistently outperforms baselines for water temperature and dissolved oxygen prediction, and serves as a general augmentation strategy across diverse backbones.

Figures

Figures reproduced from arXiv: 2607.20230 by Chonghao Qiu, Paul C. Hanson, Rahul Ghosh, Robert Ladwig, Runlong Yu, Shiyuan Luo, Xiaowei Jia, Yiqun Xie, Yue Qin.

Figure 1
Figure 1. Figure 1: Overview of PIER. where x flux i denotes the physics-derived flux features for candidate scenario i and τ is a scaling parameter, and Ti denotes the set of time steps for which ground￾truth observations are available for candidate scenario i. The exponential mapping converts the prediction error into normalized similarity range s phys q,i ∈ (0, 1], ensuring that small errors yield scores close to 1 and lar… view at source ↗
Figure 2
Figure 2. Figure 2: Map of 356 tested lakes. cost of differentiating through the inner loop while still optimizing the gate for downstream prediction quality. At test time, the gate produces γˆq in a single forward pass, making the full pipeline efficient at inference. Specifically, the weight parameter is predicted from diagnostic features zq: (3.8) γˆq = hϕ(zq), Lgate = [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison under three training settings. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Time-series predictions for two case study [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Results of ablation study. spectively); retrieval actually worsens the base model because the demoted candidates are dominated by the target lake’s own historical years, which reflect typi￾cal high-DO behavior. Embeddings encode statistically typical patterns from training data and fail to general￾ize when test conditions deviate from what the model has seen. The physics-aware stream promotes candi￾dates f… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

105 extracted references · 15 linked inside Pith

  1. [1]

    National water information system data available on the world wide web (usgs water data for the nation) , author=

  2. [2]

    Water quality data for national-scale aquatic research: The Water Quality Portal , volume =

    Read, Emily and Carr, Lindsay and De Cicco, Laura and Dugan, Hilary and Hanson, Paul and Hart, Julia and Kreft, James and Read, Jordan and Winslow, Luke , year =. Water quality data for national-scale aquatic research: The Water Quality Portal , volume =

  3. [3]

    Proceedings of the 2023 SIAM International Conference on Data Mining (SDM) , pages=

    Mini-Batch Learning Strategies for modeling long term temporal dependencies: a study in environmental applications , author=. Proceedings of the 2023 SIAM International Conference on Data Mining (SDM) , pages=. 2023 , organization=

    2023

  4. [4]

    Water Resources Research , volume=

    Stream temperature prediction in a shifting environment: Explaining the influence of deep learning architecture , author=. Water Resources Research , volume=. 2023 , publisher=

    2023

  5. [5]

    Water Resources Research , volume=

    Explore spatio-temporal learning of large sample hydrology using graph neural networks , author=. Water Resources Research , volume=. 2021 , publisher=

    2021

  6. [6]

    Hydrology and Earth System Sciences , volume=

    Towards learning universal, regional, and local hydrological behaviors via machine learning applied to large-sample datasets , author=. Hydrology and Earth System Sciences , volume=. 2019 , publisher=

    2019

  7. [7]

    Neural computation , volume=

    Long short-term memory , author=. Neural computation , volume=. 1997 , publisher=

    1997

  8. [8]

    Advances in neural information processing systems , volume=

    Attention is all you need , author=. Advances in neural information processing systems , volume=

  9. [9]

    arXiv preprint arXiv:1609.02907 , year=

    Semi-supervised classification with graph convolutional networks , author=. arXiv preprint arXiv:1609.02907 , year=

    Pith/arXiv arXiv

  10. [10]

    2015 , institution=

    PRMS-IV, the precipitation-runoff modeling system, version 4 , author=. 2015 , institution=

    2015

  11. [11]

    Philosophical Transactions of the Royal Society B: Biological Sciences , volume=

    Warming alters the metabolic balance of ecosystems , author=. Philosophical Transactions of the Royal Society B: Biological Sciences , volume=. 2010 , publisher=

    2010

  12. [12]

    Nature Climate Change , volume=

    Key ecological responses to nitrogen are altered by climate change , author=. Nature Climate Change , volume=. 2016 , publisher=

    2016

  13. [13]

    Hydrology and Earth System Sciences , volume=

    Influence of groundwater on distribution of dwarf wedgemussels (Alasmidonta heterodon) in the upper reaches of the Delaware River, northeastern USA , author=. Hydrology and Earth System Sciences , volume=. 2016 , publisher=

    2016

  14. [14]

    2021 IEEE International Conference on Data Mining (ICDM) , pages=

    Heterogeneous stream-reservoir graph networks with data assimilation , author=. 2021 IEEE International Conference on Data Mining (ICDM) , pages=. 2021 , organization=

    2021

  15. [15]

    Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences , volume=

    Big data need big theory too , author=. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences , volume=. 2016 , publisher=

    2016

  16. [16]

    Journal of Hydroinformatics , volume=

    Of data and models , author=. Journal of Hydroinformatics , volume=. 2003 , publisher=

    2003

  17. [17]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Return of frustratingly easy domain adaptation , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  18. [18]

    Proceedings of the IEEE , volume=

    A comprehensive survey on transfer learning , author=. Proceedings of the IEEE , volume=. 2020 , publisher=

    2020

  19. [19]

    Science , volume=

    Coping with the curse of freshwater variability , author=. Science , volume=. 2014 , publisher=

    2014

  20. [20]

    2020 , institution=

    Food security , author=. 2020 , institution=

    2020

  21. [21]

    Proceedings of the 1st ACM SIGCAS Conference on Computing and Sustainable Societies , pages=

    Deep transfer learning for crop yield prediction with remote sensing data , author=. Proceedings of the 1st ACM SIGCAS Conference on Computing and Sustainable Societies , pages=

  22. [22]

    Journal of Hydrology , volume=

    Applying transfer learning techniques to enhance the accuracy of streamflow prediction produced by long Short-term memory networks with data integration , author=. Journal of Hydrology , volume=. 2023 , publisher=

    2023

  23. [23]

    Water Resources Research , volume=

    Transferring hydrologic data across continents--leveraging data-rich regions to improve hydrologic prediction in data-sparse regions , author=. Water Resources Research , volume=. 2021 , publisher=

    2021

  24. [24]

    Water Resources Research , volume=

    Global-scale regionalization of hydrologic model parameters , author=. Water Resources Research , volume=. 2016 , publisher=

    2016

  25. [25]

    Environmental Modelling & Software , volume=

    The proper care and feeding of CAMELS: How limited training data affects streamflow prediction , author=. Environmental Modelling & Software , volume=. 2021 , publisher=

    2021

  26. [26]

    Sensors , volume=

    Transfer learning assisted classification and detection of Alzheimer’s disease stages using 3D MRI scans , author=. Sensors , volume=. 2019 , publisher=

    2019

  27. [27]

    Advances in neural information processing systems , volume=

    An empirical analysis of domain adaptation algorithms for genomic sequence analysis , author=. Advances in neural information processing systems , volume=

  28. [28]

    IEEE transactions on intelligent transportation systems , volume=

    Cross-domain traffic scene understanding: A dense correspondence-based transfer learning approach , author=. IEEE transactions on intelligent transportation systems , volume=. 2017 , publisher=

    2017

  29. [29]

    The world wide web conference , pages=

    Transfer meets hybrid: A synthetic approach for cross-domain collaborative filtering with text , author=. The world wide web conference , pages=

  30. [30]

    ISPRS Journal of Photogrammetry and Remote Sensing , volume=

    A deep transfer learning framework for mapping high spatiotemporal resolution LAI , author=. ISPRS Journal of Photogrammetry and Remote Sensing , volume=. 2023 , publisher=

    2023

  31. [31]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Transfer learning from deep features for remote sensing and poverty mapping , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  32. [32]

    IEEE Transactions on Geoscience and Remote Sensing , volume=

    Multisource maximum predictor discrepancy for unsupervised domain adaptation on corn yield prediction , author=. IEEE Transactions on Geoscience and Remote Sensing , volume=. 2023 , publisher=

    2023

  33. [33]

    Journal of Hydrology , volume=

    Can transfer learning improve hydrological predictions in the alpine regions? , author=. Journal of Hydrology , volume=. 2023 , publisher=

    2023

  34. [34]

    ACM Computing Surveys (CSUR) , volume=

    Spatio-temporal data mining: A survey of problems and methods , author=. ACM Computing Surveys (CSUR) , volume=. 2018 , publisher=

    2018

  35. [35]

    arXiv preprint arXiv:2403.15989 , year=

    Knowledge-guided Machine Learning: Current Trends and Future Prospects , author=. arXiv preprint arXiv:2403.15989 , year=

    Pith/arXiv arXiv

  36. [36]

    Water Resources Research , volume=

    Water for food: The global virtual water trade network , author=. Water Resources Research , volume=. 2011 , publisher=

    2011

  37. [37]

    arXiv preprint arXiv:1906.00121 , year=

    Graph wavenet for deep spatial-temporal graph modeling , author=. arXiv preprint arXiv:1906.00121 , year=

    Pith/arXiv arXiv 1906

  38. [38]

    IEEE transactions on intelligent transportation systems , volume=

    T-GCN: A temporal graph convolutional network for traffic prediction , author=. IEEE transactions on intelligent transportation systems , volume=. 2019 , publisher=

    2019

  39. [39]

    2022 IEEE 5th International Conference on Big Data and Artificial Intelligence (BDAI) , pages=

    Flooddan: Unsupervised flood forecasting based on adversarial domain adaptation , author=. 2022 IEEE 5th International Conference on Big Data and Artificial Intelligence (BDAI) , pages=. 2022 , organization=

    2022

  40. [40]

    AGU Fall Meeting Abstracts , volume=

    Structure Learning and Transfer Learning for Streamflow Prediction across Ungauged Basins , author=. AGU Fall Meeting Abstracts , volume=

  41. [41]

    Proceedings of the 2024 SIAM International Conference on Data Mining (SDM) , pages=

    Spatial-Temporal Augmented Adaptation via Cycle-Consistent Adversarial Network: An Application in Streamflow Prediction , author=. Proceedings of the 2024 SIAM International Conference on Data Mining (SDM) , pages=. 2024 , organization=

    2024

  42. [42]

    IEEE transactions on knowledge and data engineering , volume=

    Generalizing to unseen domains: A survey on domain generalization , author=. IEEE transactions on knowledge and data engineering , volume=. 2022 , publisher=

    2022

  43. [43]

    arXiv preprint arXiv:2505.06266 , year=

    Knowledge Guided Encoder-Decoder Framework: Integrating Multiple Physical Models for Agricultural Ecosystem Modeling , author=. arXiv preprint arXiv:2505.06266 , year=

    Pith/arXiv arXiv

  44. [44]

    Environmental Data Science , volume=

    Time series predictions in unmonitored sites: a survey of machine learning techniques in water resources , author=. Environmental Data Science , volume=. 2025 , publisher=

    2025

  45. [45]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Physics-guided foundation model for scientific discovery: An application to aquatic science , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  46. [46]

    ACM Computing Surveys , volume=

    Integrating scientific knowledge with machine learning for engineering and environmental systems , author=. ACM Computing Surveys , volume=. 2022 , publisher=

    2022

  47. [47]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    DroughtSet: Understanding Drought Through Spatial-Temporal Learning , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  48. [48]

    International Conference on Machine Learning , pages=

    Automatic shortcut removal for self-supervised representation learning , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  49. [49]

    Science , volume=

    Cultural water and Indigenous water science , author=. Science , volume=. 2023 , publisher=

    2023

  50. [50]

    Science , volume=

    Global water use efficiency saturation due to increased vapor pressure deficit , author=. Science , volume=. 2023 , publisher=

    2023

  51. [51]

    Nature , volume=

    Rapid groundwater decline and some cases of recovery in aquifers globally , author=. Nature , volume=. 2024 , publisher=

    2024

  52. [52]

    Limnology and Oceanography , volume=

    Time-varying responses of lake metabolism to light and temperature , author=. Limnology and Oceanography , volume=. 2020 , publisher=

    2020

  53. [53]

    ACM Transactions on Intelligent Systems and Technology (TIST) , volume=

    A survey of unsupervised deep domain adaptation , author=. ACM Transactions on Intelligent Systems and Technology (TIST) , volume=. 2020 , publisher=

    2020

  54. [54]

    Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    Domain generalization with adversarial feature learning , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

  55. [55]

    Nature Reviews Earth & Environment , volume=

    Differentiable modelling to unify machine learning and physical models for geosciences , author=. Nature Reviews Earth & Environment , volume=. 2023 , publisher=

    2023

  56. [56]

    arXiv preprint arXiv:2110.11328 , year=

    A fine-grained analysis on distribution shift , author=. arXiv preprint arXiv:2110.11328 , year=

    Pith/arXiv arXiv

  57. [57]

    Advances in Neural Information Processing Systems , volume=

    Wild-time: A benchmark of in-the-wild distribution shift over time , author=. Advances in Neural Information Processing Systems , volume=

  58. [58]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Domain generalization by solving jigsaw puzzles , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  59. [59]

    IEEE transactions on pattern analysis and machine intelligence , volume=

    Domain generalization: A survey , author=. IEEE transactions on pattern analysis and machine intelligence , volume=. 2022 , publisher=

    2022

  60. [60]

    Artificial Intelligence Review , volume=

    Domain generalization through meta-learning: a survey , author=. Artificial Intelligence Review , volume=. 2024 , publisher=

    2024

  61. [61]

    arXiv preprint arXiv:2104.02008 , year=

    Domain generalization with mixstyle , author=. arXiv preprint arXiv:2104.02008 , year=

    Pith/arXiv arXiv

  62. [62]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Metaaugment: Sample-aware data augmentation policy learning , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  63. [63]

    Journal of machine learning research , volume=

    Domain-adversarial training of neural networks , author=. Journal of machine learning research , volume=

  64. [64]

    2015 , institution=

    Summary of hydrologic modeling for the delaware river basin using the water availability tool for environmental resources (water) , author=. 2015 , institution=

    2015

  65. [65]

    Agricultural Economics , volume=

    Bringing automated, remote-sensed, machine learning methods to monitoring crop landscapes at scale , author=. Agricultural Economics , volume=. 2019 , publisher=

    2019

  66. [66]

    Geoscientific Model Development , volume=

    KGML-ag: a modeling framework of knowledge-guided machine learning to simulate agroecosystems: a case study of estimating N 2 O emission using data from mesocosm experiments , author=. Geoscientific Model Development , volume=. 2022 , publisher=

    2022

  67. [67]

    Proceedings of the 2021 SIAM International Conference on Data Mining (SDM) , pages=

    Physics-Guided Recurrent Graph Model for Predicting Flow and Temperature in River Networks , author=. Proceedings of the 2021 SIAM International Conference on Data Mining (SDM) , pages=. 2021 , organization=

    2021

  68. [68]

    2021 IEEE International Conference on Data Mining (ICDM) , pages=

    A statistically-guided deep network transformation and moderation framework for data with spatial heterogeneity , author=. 2021 IEEE International Conference on Data Mining (ICDM) , pages=. 2021 , organization=

    2021

  69. [69]

    Water Resources Research , volume=

    Predicting Water Temperature Dynamics of Unmonitored Lakes With Meta-Transfer Learning , author=. Water Resources Research , volume=. 2021 , publisher=

    2021

  70. [70]

    Deep Learning for the Earth Sciences: A Comprehensive Approach to Remote Sensing, Climate Science, and Geosciences , pages=

    Applications of deep learning in hydrology , author=. Deep Learning for the Earth Sciences: A Comprehensive Approach to Remote Sensing, Climate Science, and Geosciences , pages=. 2021 , publisher=

    2021

  71. [71]

    International conference on machine learning , pages=

    Sharp-maml: Sharpness-aware model-agnostic meta learning , author=. International conference on machine learning , pages=. 2022 , organization=

    2022

  72. [72]

    arXiv preprint arXiv:2405.00749 , year=

    More is better: Deep domain adaptation with multiple sources , author=. arXiv preprint arXiv:2405.00749 , year=

    Pith/arXiv arXiv

  73. [73]

    International Conference on Machine Learning , pages=

    f-domain adversarial learning: Theory and algorithms , author=. International Conference on Machine Learning , pages=. 2021 , organization=

    2021

  74. [74]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Informer: Beyond efficient transformer for long sequence time-series forecasting , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  75. [75]

    arXiv preprint arXiv:2405.14616 , year=

    Timemixer: Decomposable multiscale mixing for time series forecasting , author=. arXiv preprint arXiv:2405.14616 , year=

    Pith/arXiv arXiv

  76. [76]

    arXiv preprint arXiv:2210.02186 , year=

    Timesnet: Temporal 2d-variation modeling for general time series analysis , author=. arXiv preprint arXiv:2210.02186 , year=

    Pith/arXiv arXiv

  77. [77]

    International conference on machine learning , pages=

    Axiomatic attribution for deep networks , author=. International conference on machine learning , pages=. 2017 , organization=

    2017

  78. [78]

    arXiv preprint arXiv:2509.14563 , year=

    Learning to Retrieve for Environmental Knowledge Discovery: An Augmentation-Adaptive Self-Supervised Learning Framework , author=. arXiv preprint arXiv:2509.14563 , year=

    arXiv

  79. [79]

    arXiv preprint arXiv:2505.04163 , year=

    Retrieval augmented time series forecasting , author=. arXiv preprint arXiv:2505.04163 , year=

    Pith/arXiv arXiv

  80. [80]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    GREAT: Generalizable Representation Enhancement via Auxiliary Transformations for Zero-Shot Environmental Prediction , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

Showing first 80 references.