Pith. sign in

REVIEW 4 major objections 5 minor 61 references

Variational Autoencoder Framework for Hyperspectral Retrievals (Hyper-VAE) of Phytoplankton Absorption and Chlorophyll a in Coastal Waters for NASA's EMIT and PACE Missions

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read VAE-based models retrieve phytoplankton absorption and chlorophyll a from hyperspectral ocean color with high precision and low bias, beating MDNs on PACE-scale inputs.

desk verdict VAE aphy retrieval is a plausible engineering result, but the paper's own data admission leaves the one-to-many motivation untested; worth a serious referee if the authors add uncertainty metrics and a genuine one-to-many test. read the letter →

arxiv 2504.13476 v1 pith:XENVN3Z7 submitted 2025-04-18 cs.LG cs.CVeess.IV

Jiadong Lou , Bingqing Liu , Yuanheng Xiong , Xiaodong Zhang , Xu Yuan This is my paper · ORCID
classification cs.LGcs.CVeess.IV
keywords HyperspectralremotesensingVariationalautoencoderPhytoplanktonabsorptionChlorophyllaCoastalwatersOne-to-manyinversionPACEEMIT
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

This paper claims that a Variational Autoencoder (VAE), repurposed from data generation to regression, retrieves phytoplankton absorption spectra and chlorophyll a from hyperspectral remote-sensing reflectance in optically complex coastal waters more accurately and stably than mixture density networks (MDNs). The authors tailor two VAE models to the spectral settings of the PACE and EMIT satellite instruments, and they report lower error and bias across eight metrics on held-out in-situ data. The largest gains appear at PACE's 141-band resolution, where the VAE's latent-space compression avoids the numerical instability that hampers full-covariance MDNs. The intended significance is a machine-learning inversion tool for current and upcoming hyperspectral ocean-color missions that can represent the fact that one reflectance spectrum may correspond to several valid water-constituent combinations.

What carries the argument

The load-bearing object is the stochastic latent vector $z=\mu(R_{rs})+\sigma(R_{rs})\odot\epsilon$, with $\epsilon\sim\mathcal{N}(0,I)$, sampled through the reparameterization trick during both training and inference. The encoder turns each $R_{rs}$ spectrum into means and variances for independent Gaussian latent components; the decoder turns a sampled $z$ into a predicted $a_{\mathrm{phy}}$ spectrum or a Chl-a value. The KL term in the loss, $L=\|a'_{\mathrm{phy}}-a_{\mathrm{phy}}\|_1+\lambda\,\mathrm{KL}(q(z|R_{rs})\|p(z))$, regularizes the latent space toward a standard normal prior. This structure lets the model distribute probability over several plausible outputs for one input while keeping individual predictions stable.

What would settle it

Construct or simulate a set of nearly identical $R_{rs}$ spectra whose true $a_{\mathrm{phy}}$ or Chl-a values differ substantially, for example by varying CDOM and non-algal particle concentrations jointly with phytoplankton in a radiative-transfer model. If repeated stochastic inferences from the trained VAE on those inputs do not yield multiple distinct, physically plausible outputs, or if its predictions equal those of a deterministic decoder with the latent noise fixed to zero, the one-to-many claim is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that VAE-based models are a suitable backbone for hyperspectral ocean-color inversion: VAE-aphy predicts full $a_{\mathrm{phy}}$ spectra from $R_{rs}$ at PACE and EMIT wavelengths, and VAE-Chl-a predicts log-transformed Chl-a, with both showing higher precision and lower bias than MDNs on held-out in-situ data. The authors argue the advantage is structural. MDNs trained on the compiled dataset are used deterministically, collapsing the multimodal target to a point estimate, while a sampled MDN is unstable; the VAE instead learns a regularized latent distribution whose stochastic draws produce stable, physically meaningful outputs. On the 141-band PACE setting in particular, the VAE keeps median bias within roughly $\pm10\%$ across 400-700 nm, whereas MDN deviations exceed $\pm50\%$ at several wavelengths.

Load-bearing premise

The paper's demonstration depends on the assumption that the compiled in-situ training set, which the paper itself says lacks true one-to-many cases, is enough to validate the VAE's claimed ability to handle ambiguous $R_{rs}$-to-IOP mappings; without such pairs, the stochastic sampling is never directly tested.

Editorial extensions

If this is right

  • At PACE spectral settings, the VAE yields higher-precision, lower-bias $a_{\mathrm{phy}}$ spectra than the MDN baseline, with the largest stability gains at wavelengths above 600 nm where MDN's median bias fluctuates widely.
  • The VAE-Chl-a model slightly outperforms MDN on Chl-a, but with smaller margins, consistent with the claim that the VAE's main advantage appears on high-dimensional spectral outputs.
  • The method generalizes to unseen optically complex waters with post-hurricane conditions, capturing fine spectral detail in the 450-500 nm and 480-620 nm ranges that matter for distinguishing phytoplankton groups.
  • Because the architecture maps high-dimensional inputs through a low-dimensional latent space, it avoids the quadratic growth of covariance parameters that makes full-covariance MDNs ill-conditioned at PACE resolution, and it trains roughly an order of magnitude faster.
  • The model configurations can be adapted to other hyperspectral missions' band settings, so the framework transfers to follow-on instruments.

Reading between the lines

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

  • The paper does not directly test the one-to-many mechanism, because its own reading of the training data finds no identical-$R_{rs}$ pairs; a controlled radiative-transfer simulation with deliberate ambiguity would separate the VAE's stochastic-mapping capability from its regularized-regression behavior.
  • Running the trained VAE many times on the same input yields an ensemble of $a_{\mathrm{phy}}$ spectra; that ensemble could be read as per-pixel retrieval uncertainty, an output the paper mentions conceptually but does not quantify.
  • A natural extension is a multi-decoder VAE that predicts the full IOP set from one shared latent representation; the paper flags this as future work, but simulated-data pre-training would be needed to realize it.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes VAE-based models (Hyper-VAE: VAE-aphy and VAE-Chl-a) to retrieve phytoplankton absorption spectra and chlorophyll a from hyperspectral remote sensing reflectance at NASA PACE (2.5 nm, 141 bands) and EMIT (7.4 nm, 41 bands) spectral settings. Using compiled GLORIA/SeaBASS data (2,114 Rrs-aphy pairs and 6,111 Rrs-Chl-a pairs after quality control), a 70/30 train/test split, and a retrained MDN baseline, the authors report improved error and bias metrics for VAE over MDN, greater wavelength-wise stability, and three qualitative example spectra on independent Galveston Bay data. The paper frames the VAE's stochastic latent sampling (Eq. 4) as addressing the one-to-many inversion problem and argues that VAE scales better to high-dimensional PACE data than MDN because it avoids full covariance modeling in the output space.

Significance. If the claimed advantage is robust, the framework is a practical contribution to hyperspectral ocean color inversion for EMIT/PACE and future SBG missions, and the detailed model configurations are useful to the community. Credit is due for retraining the MDN baseline under a common protocol rather than citing prior numbers, reporting eight metrics on a held-out test split, and including an independent Galveston Bay dataset, even if only qualitatively. However, the load-bearing one-to-many claim is not empirically tested, and the comparisons to MDN and M-MDN lack uncertainty quantification and distributional scoring; as presented, the central methodological novelty rests on assumptions that the manuscript itself states are absent from the training data.

major comments (4)
  1. [II-D and IV-B] The manuscript's central claim that VAE addresses the one-to-many inversion problem is not tested. Section II-D states that the compiled GLORIA/SeaBASS data 'does not contain one-to-many features—no different Chl-a values or aphy spectra correspond to the same or similar Rrs spectra,' and Section IV-B reports only point metrics on a random split. Under these conditions the L1 reconstruction term in Eq. (11) can be satisfied by a near-deterministic decoder or a collapsed posterior, so the stochastic latent sampling in Eq. (4) is never shown to be the active mechanism. I request a controlled experiment—for example, constructed or RTM-simulated Rrs spectra that share identical or near-identical input while paired with genuinely different aphy/Chl-a targets—together with a report of the learned posterior predictive variance per test input.
  2. [IV-B, Fig. 7] The VAE-vs-MDN comparison is asymmetric and the M-MDN comparison is anecdotal. The deterministic MDN is evaluated with point metrics on the test split, while the stochastic M-MDN is dismissed from a few example spectra in Fig. 7 without any distributional scoring; Section IV-B does not state how many epsilon draws per spectrum are used for the VAE, and no repeated-inference error bars or multiple-seed results are reported. A proper comparison should include strict proper scoring rules (e.g., negative log-likelihood or quantile coverage/CRPS) for all three models, with variance across seeds and across draws.
  3. [IV-D] The independent Galveston Bay generalization claim is supported only by three qualitative spectra in Figs. 9 and 10. No aggregate metrics, no comparison against deterministic MDN on this dataset, and no error bars are provided; the text says only that VAE 'closely align[s] well' while M-MDN shows zigzag patterns. I request quantitative metrics over the full Galveston Bay matchup set, with a paired comparison against MDN and M-MDN.
  4. [III-A, Eq. (6)/(11)] The KL-divergence weight λ and the latent dimensionality are treated as tuned but are never reported or varied, which matters because λ controls whether the KL term suppresses posterior variance. Without a sensitivity analysis or a report of the learned σ(x) magnitudes, one cannot distinguish the proposed VAE mechanism from a regularized deterministic autoencoder; this is load-bearing for the paper's central claim.
minor comments (5)
  1. [Throughout] The manuscript contains many typographical and formatting artifacts: 'V AE' spacing throughout, 'V AE-Chl-a' inconsistency, 'We first time tailor' in the abstract, and 'aphy (m(− 1))' in the Fig. 11 caption.
  2. [Fig. 4] The Figure 4 caption lists panels in a sequence that does not match the actual panel arrangement described in the text; the caption should be rewritten to match the layout.
  3. [II-B and IV-A] Details needed for reproducibility are missing: random seeds, whether a validation subset is used for early stopping, and the exact λ value used in Eqs. (11) and (12) are not stated.
  4. [Eq. (1)] Equation (1) has malformed parentheses: it should read c × (bbtotal/(atotal + bbtotal)) rather than 'c× ( bbtotal atotal +bbtotal )'.
  5. [V-A] The statement that VAE 'avoid[s] the uncertainty associated with the distribution nature' is confusing, since the decoder in Eq. (5) also defines a conditional Gaussian distribution; please clarify the intended contrast.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: Hyper-VAE's performance claims rest on held-out data and an external MDN baseline, not on its own fitted outputs; the one-to-many capability is a stated limitation, not a circular step.

full rationale

The paper's central result is an empirical comparison, not a derivation that reduces to its own inputs. VAE-aphy and VAE-Chl-a are trained on paired Rrs-aphy and Rrs-Chl-a data compiled by Pahlevan et al. and O'Shea et al., evaluated on a 30% holdout and on an independent Galveston Bay dataset; the MDN and M-MDN baselines are reproduced under the same preprocessing and compared on the same metrics. No fitted parameter is renamed as a prediction, and no quantity in Eqs. (7)-(12) is defined in terms of the predicted outputs by construction. The paper does cite work by an overlapping author (B. Liu) for Galveston Bay field data and hurricane-related context, but those citations are independent data sources rather than load-bearing arguments, so they do not constitute circularity. The one-to-many inversion claim is weakened by the paper's own admission in Section II-D that the compiled dataset 'does not contain one-to-many features' and by Section V-E's note that satellite match-up validation is ongoing; however, this is an untested assumption or limitation, not a self-referential reduction. The comparison to MDN is externally falsifiable from the reported metrics, so the paper is self-contained and receives a circularity score of 0.

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

The central claim rests mostly on data and architecture choices. No new physical entities or conserved quantities are introduced. The main free parameters are latent dimensions and the KL weight; the main axioms are the standard bio-optical model, representativeness of the training data, the assumed Gaussian latent, and the suitability of the independent Galveston data as an out-of-distribution test. The one-to-many phenomenon is cited to existing literature, but no data in the paper contains verified one-to-many pairs.

free parameters (4)
  • KL divergence weight lambda = not reported (hand tuned)
    Eqs. (6), (11)-(12) introduce lambda; Section III-A says it was "carefully tuned", but no value, search range, or ablation is given. The balance between reconstruction and latent regularization affects output diversity and accuracy.
  • VAE latent dimension, aphy model = 256
    Table I fixes a 256-dimensional latent vector for VAE-aphy; no ablation or justification is provided, and this dimension directly controls how much stochasticity the decoder sees.
  • VAE latent dimension, Chl-a model = 64
    Table II fixes a 64-dimensional latent vector for VAE-Chl-a; no ablation is provided.
  • MDN Gaussian component count and covariance structure = 5 components, diagonal covariance
    Section IV-A sets five Gaussian components and replaces the full covariance with a diagonal approximation to avoid numerical instability; this baseline choice shapes the VAE-vs-MDN comparison.
assumptions (5)
  • standard math Rrs is related to IOPs by the simplified radiative transfer equation Rrs = c * bb / (a + bb), and a is separable into water, CDOM, NAP and phytoplankton absorption (Eqs. 1-2).
    Used in Section I to justify why Rrs can be inverted for aphy; this is a standard ocean color approximation, but it is a modeling assumption and the proportionality constant c is not calibrated in this paper.
  • domain assumption The compiled SeaBASS/GLORIA dataset is representative of optically complex estuarine-coastal waters and the paired Rrs-aphy and Rrs-Chl-a measurements are accurate enough for training and testing.
    Section II-A adopts datasets from Pahlevan et al. 2021 and O'Shea et al. 2023; the paper does not re-validate the field measurements.
  • domain assumption The one-to-many inverse problem is a real feature of coastal waters, even though the training dataset lacks true one-to-many example pairs.
    Sections I and II-D invoke ambiguity from single Rrs to multiple IOP combinations as motivation; Section II-D admits GLORIA does not contain one-to-many features, so the motivation is not tested by the reported experiments.
  • ad hoc to paper Latent features of the Rrs-aphy relationship can be approximated as Gaussian, and the decoder with fixed-variance Gaussian output produces stable predictions without overfitting noise.
    Section III-A and V-B assume p(z) standard normal and p(x|z) fixed-variance Gaussian; no goodness-of-fit test or comparison to alternative latent distributions is provided.
  • domain assumption The independent Galveston Bay dataset, measured with different instruments and under post-hurricane conditions, is a valid out-of-distribution test comparable to the training radiometry.
    Section II-C describes GER 1500 measurements and QFT-derived aphy; Section IV-D uses it to claim generality, but only three example spectra are shown and no summary error metrics are reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Variational Autoencoder Framework for Hyperspectral Retrievals (Hyper-VAE) of Phytoplankton Absorption and Chlorophyll a in Coastal Waters for NASA's EMIT and PACE Missions." pith.science (2026). https://pith.science/paper/XENVN3Z7

@misc{pith2026250413476,
  author       = {Pith},
  title        = {Pith review of: Variational Autoencoder Framework for Hyperspectral Retrievals (Hyper-VAE) of Phytoplankton Absorption and Chlorophyll a in Coastal Waters for NASA's EMIT and PACE Missions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XENVN3Z7}},
  note         = {Machine review of arXiv:2504.13476}
}
read the original abstract

Phytoplankton absorb and scatter light in unique ways, subtly altering the color of water, changes that are often minor for human eyes to detect but can be captured by sensitive ocean color instruments onboard satellites from space. Hyperspectral sensors, paired with advanced algorithms, are expected to significantly enhance the characterization of phytoplankton community composition, especially in coastal waters where ocean color remote sensing applications have historically encountered significant challenges. This study presents novel machine learning-based solutions for NASA's hyperspectral missions, including EMIT and PACE, tackling high-fidelity retrievals of phytoplankton absorption coefficient and chlorophyll a from their hyperspectral remote sensing reflectance. Given that a single Rrs spectrum may correspond to varied combinations of inherent optical properties and associated concentrations, the Variational Autoencoder (VAE) is used as a backbone in this study to handle such multi-distribution prediction problems. We first time tailor the VAE model with innovative designs to achieve hyperspectral retrievals of aphy and of Chl-a from hyperspectral Rrs in optically complex estuarine-coastal waters. Validation with extensive experimental observation demonstrates superior performance of the VAE models with high precision and low bias. The in-depth analysis of VAE's advanced model structures and learning designs highlights the improvement and advantages of VAE-based solutions over the mixture density network (MDN) approach, particularly on high-dimensional data, such as PACE. Our study provides strong evidence that current EMIT and PACE hyperspectral data as well as the upcoming Surface Biology Geology mission will open new pathways toward a better understanding of phytoplankton community dynamics in aquatic ecosystems when integrated with AI technologies.

Figures

Figures reproduced from arXiv: 2504.13476 by the authors.

Figure 1
Figure 1. NASA’s PACE-OCI Level 2 AOP data obtained on May 15, 2024, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Spectral distribution of the Rrs-aphy dataset, with bars denoting the minimum and maximum values of (a) Rrs (sr−1 ) and (b) aphy (m−1 ) using EMIT spectral setting. C. In-situ Hyperspectral Measurements Additional hyperspectral Rrs-aphy field data are used to test the VAE’s robustness and generalizability. Phytoplankton absorption coefficient (aphy) was obtained from surface water samples collected during two survey… view at source ↗
Figure 3
Figure 3. The structure of VAE for Predicting aphy and Chl-a. Furthermore, the training objective of VAE is to minimize the loss calculated from two aspects, i.e., L = −Eq(z|x) [log p(x|z)] + λ · KL q(z|x)||p(z)  , (6) where λ is the weight to balance the two terms. The loss function aligns with the standard VAE loss introduced in [26] in their seminal work on VAEs and has been extensively used in many VAE-based studies [22]… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Scatter plots and evaluation metrics for VAE and MDN predictions of [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Performance of VAE in terms of RMSE, Log-Bias, and [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Performance of VAE in terms of RMSE, Log-Bias, and [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Comparison of actual and predicted aphy spectra in the 400–700 nm range: (a)–(c) show results for VAE, MDN, and M-MDN using PACE spectral setting, while (d)–(f) present corresponding results using EMIT wavelengths. pronounced. These results suggest that M-MDN struggles…
Figure 8
Figure 8. Figure 8: Evaluation metrics for VAE and MDN predictions of Chl-a using PACE and EMIT spectral settings. (a), (c) VAE for PACE and EMIT, and (b), (d) [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Comparison of actual and predicted aphy spectra at PACE wavelength setting in the range of 400–700 nm, using (a)–(c) VAE and (d)–(f) M-MDN.. difference in performance stems from the VAE’s learning structure, which does not directly fit the aphy distribution from the tr…
Figure 10
Figure 10. Figure 10: Comparison of actual and predicted aphy spectra at EMIT wavelength setting in the range of 400–700 nm, using (a)–(c) VAE and (d)–(f) M-MDN. the learned variable latent vectors and the derived prediction data, allowing VAEs to generate accurate predictions from the lat…
Figure 11
Figure 11. Figure 11: The aphy (m( − 1)) distribution at different wavelengths. missions. EMIT has a lower-dimensional input space (N=41) compared to PACE, which has 141 spectral bands, making aphy predictions the highest-dimensional to date. This further emphasizes the need for robust lea…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 52 canonical work pages

  1. [1]

    Mixture density networks for re-constructing historical ocean-color products over inland and coastal waters: demonstration and validation

    Sundarabalan V Balasubramanian, Ryan E O’Shea, Arun M Saranathan, Christopher C Begeman, Daniela Gurlin, Caren Binding, Claudia Gi- ardino, Michelle C Tomlinson, Krista Alikas, Kersti Kangro, et al. Mixture density networks for re-constructing historical ocean-color products over inland and coastal waters: demonstration and validation. Frontiers in Remote...

    work page 2025

  2. [2]

    Robust algorithm for estimating total suspended solids (tss) in inland and nearshore coastal waters

    Sundarabalan V Balasubramanian, Nima Pahlevan, Brandon Smith, Caren Binding, John Schalles, Hubert Loisel, Daniela Gurlin, Steven Greb, Krista Alikas, Mirjam Randla, et al. Robust algorithm for estimating total suspended solids (tss) in inland and nearshore coastal waters. Remote Sensing of Environment , 246:111768, 2020

    work page 2020

  3. [3]

    The science of hypoxia in the northern gulf of mexico: a review

    Thomas S Bianchi, Steven F DiMarco, JH Cowan Jr, Robert D Hetland, P Chapman, JW Day, and MA Allison. The science of hypoxia in the northern gulf of mexico: a review. Science of the Total Environment , 408(7):1471–1484, 2010

    work page 2010

  4. [4]

    Variational inference: A review for statisticians

    David M Blei, Alp Kucukelbir, and Jon D McAuliffe. Variational inference: A review for statisticians. Journal of the American statistical Association, 112(518):859–877, 2017

    2017

  5. [5]

    Natural variability of phytoplanktonic absorption in oceanic waters: In- fluence of the size structure of algal populations

    Annick Bricaud, Herv ´e Claustre, Josephine Ras, and Kadija Oubelkheir. Natural variability of phytoplanktonic absorption in oceanic waters: In- fluence of the size structure of algal populations. Journal of Geophysical Research: Oceans, 109(C11), 2004

    work page 2004

  6. [6]

    Importance weighted autoencoders

    Yuri Burda, Roger Grosse, and Ruslan Salakhutdinov. Importance weighted autoencoders. arXiv preprint arXiv:1509.00519 , 2015

    arXiv 2015

  7. [7]

    Information content of absorption spectra and implications for ocean color inversion

    BB Cael, Alison Chase, and Emmanuel Boss. Information content of absorption spectra and implications for ocean color inversion. Applied Optics, 59(13):3971–3984, 2020

    work page 2020

  8. [8]

    Phytoplankton composition from space: Requirements, opportunities, and challenges

    Ivona Cetini ´c, Cecile S Rousseaux, Ian T Carroll, Alison P Chase, Sasha J Kramer, P Jeremy Werdell, David A Siegel, Heidi M Dierssen, Dylan Catlett, Aimee Neeley, et al. Phytoplankton composition from space: Requirements, opportunities, and challenges. Remote Sensing of Environment, 302:113964, 2024

    work page 2024

Show all 61 references
  1. [9]

    Evaluation of diagnostic pigments to estimate phytoplankton size classes

    Alison P Chase, Sasha J Kramer, Nils Ha ¨entjens, Emmanuel S Boss, Lee Karp-Boss, Mimi Edmondson, and Jason R Graff. Evaluation of diagnostic pigments to estimate phytoplankton size classes. Limnology and Oceanography: Methods , 18(10):570–584, 2020

    2020

  2. [10]

    Variational lossy autoencoder

    Xi Chen, Diederik P Kingma, Tim Salimans, Yan Duan, Prafulla Dhariwal, John Schulman, Ilya Sutskever, and Pieter Abbeel. Variational lossy autoencoder. arXiv preprint arXiv:1611.02731 , 2016

    2016 arXiv

  3. [11]

    Assessment of the relationships between dominant cell size in natural phytoplankton com- munities and the spectral shape of the absorption coefficient

    Aurea M Ciotti, Marlon R Lewis, and John J Cullen. Assessment of the relationships between dominant cell size in natural phytoplankton com- munities and the spectral shape of the absorption coefficient. Limnology and Oceanography, 47(2):404–417, 2002. 15

    2002

  4. [12]

    How ambiguous is the inverse problem of ocean color in coastal waters? Journal of Geophysical Research: Oceans, 112(C3), 2007

    Michael Defoin-Platel and Malik Chami. How ambiguous is the inverse problem of ocean color in coastal waters? Journal of Geophysical Research: Oceans, 112(C3), 2007

    2007

  5. [13]

    Living up to the hype of hyperspectral aquatic remote sensing: science, resources and outlook

    Heidi M Dierssen, Steven G Ackleson, Karen E Joyce, Erin L Hestir, Alexandre Castagna, Samantha Lavender, and Margaret A McManus. Living up to the hype of hyperspectral aquatic remote sensing: science, resources and outlook. Frontiers in Environmental Science , 9:649528, 2021

    2021

  6. [14]

    Tutorial on variational autoencoders

    Carl Doersch. Tutorial on variational autoencoders. arXiv preprint arXiv:1606.05908, 2016

    2016 arXiv

  7. [15]

    Galveston bay and coastal ocean optical-geochemical response to hurricane harvey from viirs ocean color

    Eurico J D’Sa, Ishan Joshi, and Bingqing Liu. Galveston bay and coastal ocean optical-geochemical response to hurricane harvey from viirs ocean color. Geophysical Research Letters , 45(19):10–579, 2018

    2018

  8. [16]

    Biogeochemical response of apalachicola bay and the shelf waters to hurricane michael using ocean color semi-analytic/inversion and hydrodynamic models

    Eurico J D’Sa, Ishan D Joshi, Bingqing Liu, Dong S Ko, Christo- pher L Osburn, and Thomas S Bianchi. Biogeochemical response of apalachicola bay and the shelf waters to hurricane michael using ocean color semi-analytic/inversion and hydrodynamic models. Frontiers in Marine Sci...

    2019

  9. [17]

    The peak near 700 nm on radiance spectra of algae and water: relationships of its magnitude and position with chlorophyll concentration

    Anatoly Gitelson. The peak near 700 nm on radiance spectra of algae and water: relationships of its magnitude and position with chlorophyll concentration. International Journal of Remote Sensing , 13(17):3367– 3373, 1992

    1992

  10. [18]

    A chlorophyll-retrieval algorithm for satellite imagery (medium resolution imaging spectrometer) of inland and coastal waters

    Herman J Gons, Machteld Rijkeboer, and Kevin G Ruddick. A chlorophyll-retrieval algorithm for satellite imagery (medium resolution imaging spectrometer) of inland and coastal waters. Journal of Plankton Research, 24(9):947–951, 2002

    2002

  11. [19]

    The nasa earth surface mineral dust source investigation

    Robert Green. The nasa earth surface mineral dust source investigation. 43rd COSPAR Scientific Assembly. Held 28 January-4 February , 43:37, 2021

    2021

  12. [20]

    Remote estimation of chl-a concentration in turbid productive waters—return to a simple two-band nir-red model? Remote Sensing of Environment , 115(12):3479–3490, 2011

    Daniela Gurlin, Anatoly A Gitelson, and Wesley J Moses. Remote estimation of chl-a concentration in turbid productive waters—return to a simple two-band nir-red model? Remote Sensing of Environment , 115(12):3479–3490, 2011

    2011

  13. [21]

    Future phytoplankton diversity in a changing climate

    Stephanie A Henson, BB Cael, Stephanie R Allen, and Stephanie Dutkiewicz. Future phytoplankton diversity in a changing climate. Nature communications, 12(1):5372, 2021

    2021

  14. [22]

    beta-vae: Learning basic visual concepts with a constrained variational framework

    Irina Higgins, Loic Matthey, Arka Pal, Christopher Burgess, Xavier Glorot, Matthew Botvinick, Shakir Mohamed, and Alexander Lerchner. beta-vae: Learning basic visual concepts with a constrained variational framework. In International conference on learning representations , 2017

    2017

  15. [23]

    Habnet: Machine learning, remote sensing-based detection of harmful algal blooms

    Paul R Hill, Anurag Kumar, Marouane Temimi, and David R Bull. Habnet: Machine learning, remote sensing-based detection of harmful algal blooms. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 13:3229–3239, 2020

    2020

  16. [24]

    Alia L Khan, Heidi M Dierssen, Ted A Scambos, Juan H ¨ofer, and Raul R Cordero. Spectral characterization, radiative forcing and pigment content of coastal antarctic snow algae: approaches to spectrally discriminate red and green communities and their impact on snowmelt. The C...

    2021

  17. [25]

    Tyler King, Carl Legleiter, Natalie Hall, Adam Mumford, Kurt Car- penter, and E Slonecker. Spectral mixture analysis for surveillance of harmful algal blooms (smash): A hybrid laboratory-and satellite-based approach to identify algal taxa in inland waterbodies via remote sensi...

    2021

  18. [26]

    Auto-encoding variational bayes, 2013

    Diederik P Kingma, Max Welling, et al. Auto-encoding variational bayes, 2013

    2013

  19. [27]

    Improved variational inference with inverse autoregressive flow

    Durk P Kingma, Tim Salimans, Rafal Jozefowicz, Xi Chen, Ilya Sutskever, and Max Welling. Improved variational inference with inverse autoregressive flow. Advances in neural information processing systems, 29, 2016

    2016

  20. [28]

    Simulated inherent optical properties of aquatic particles using the equivalent algal populations (eap) model

    Lisl Robertson Lain, Jeremy Kravitz, Mark Matthews, and Stewart Bernard. Simulated inherent optical properties of aquatic particles using the equivalent algal populations (eap) model. Scientific Data, 10(1):412, 2023

    2023

  21. [29]

    Deriving inherent optical properties from water color: a multiband quasi-analytical algorithm for optically deep waters

    ZhongPing Lee, Kendall L Carder, and Robert A Arnone. Deriving inherent optical properties from water color: a multiband quasi-analytical algorithm for optically deep waters. Applied optics, 41(27):5755–5772, 2002

    2002

  22. [30]

    Gloria-a globally representative hyperspectral in situ dataset for optical sensing of water quality

    Moritz K Lehmann, Daniela Gurlin, Nima Pahlevan, Krista Alikas, Ted Conroy, Janet Anstee, Sundarabalan V Balasubramanian, Cl ´audio CF Barbosa, Caren Binding, Astrid Bracher, et al. Gloria-a globally representative hyperspectral in situ dataset for optical sensing of water qua...

    2023

  23. [31]

    Estimation of chlorophyll a concentration using nir/red bands of meris and classification procedure in inland turbid water

    Yunmei Li, Qiao Wang, Chuanqing Wu, Shaohua Zhao, Xing Xu, Yanfei Wang, and Changchun Huang. Estimation of chlorophyll a concentration using nir/red bands of meris and classification procedure in inland turbid water. IEEE Transactions on Geoscience and Remote Sensing , 50(3):9...

    2011

  24. [32]

    Multi-decadal trends and influences on dissolved organic carbon distribution in the barataria basin, louisiana from in-situ and landsat/modis observations

    Bingqing Liu, Eurico J D’Sa, and Ishan Joshi. Multi-decadal trends and influences on dissolved organic carbon distribution in the barataria basin, louisiana from in-situ and landsat/modis observations. Remote Sensing of Environment, 228:183–202, 2019

    2019

  25. [33]

    Bingqing Liu, Eurico J D’Sa, and Ishan D Joshi. Floodwater impact on galveston bay phytoplankton taxonomy, pigment composition and photo- physiological state following hurricane harvey from field and ocean color (sentinel-3a olci) observations. Biogeosciences, 16(9):1975–2001, 2019

    1975

  26. [34]

    Bingqing Liu, Eurico J D’Sa, Kanchan Maiti, Victor H Rivera-Monroy, and Zuo Xue. Biogeographical trends in phytoplankton community size structure using adaptive sentinel 3-olci chlorophyll a and spectral empirical orthogonal functions in the estuarine-shelf waters of the north...

    2021

  27. [35]

    Bingqing Liu, Eurico J D’Sa, Francesca Messina, Melissa M Baustian, Kanchan Maiti, Victor H Rivera-Monroy, Wei Huang, and Ioannis Y Georgiou. Dissolved organic carbon dynamics and fluxes in mississippi- atchafalaya deltaic system impacted by an extreme flood event and hurrican...

    2023

  28. [36]

    Phytoplankton optical fingerprint libraries for development of phyto- plankton ocean color satellite products

    Michael W Lomas, Aimee R Neeley, Ryan Vandermeulen, Antonio Mannino, Crystal Thomas, Michael G Novak, and Scott A Freeman. Phytoplankton optical fingerprint libraries for development of phyto- plankton ocean color satellite products. Scientific Data, 11(1):168, 2024

    2024

  29. [37]

    Adversarial autoencoders

    Alireza Makhzani, Jonathon Shlens, Navdeep Jaitly, Ian Goodfel- low, and Brendan Frey. Adversarial autoencoders. arXiv preprint arXiv:1511.05644, 2015

    2015 arXiv

  30. [38]

    Light and water: radiative transfer in natural waters

    Curtis D Mobley. Light and water: radiative transfer in natural waters. (No Title), 1994

    1994

  31. [39]

    Ocean optics protocols for satellite ocean color sensor validation, revision 4

    James L Mueller, A Morel, R Frouin, C Davis, R Arnone, K Carder, ZP Lee, RG Steward, S Hooker, CD Mobley, et al. Ocean optics protocols for satellite ocean color sensor validation, revision 4. volume iii: Radiometric measurements and data analysis protocols. 2003

    2003

  32. [40]

    A hyperspectral inversion framework for estimating absorbing inherent optical properties and biogeochemical parameters in inland and coastal waters

    Ryan E O’Shea, Nima Pahlevan, Brandon Smith, Emmanuel Boss, Daniela Gurlin, Krista Alikas, Kersti Kangro, Raphael M Kudela, and Diana Vai ˇci¯ut˙e. A hyperspectral inversion framework for estimating absorbing inherent optical properties and biogeochemical parameters in inland ...

    2023

  33. [41]

    Hyperspectral retrievals of phytoplankton absorption and chlorophyll-a in inland and nearshore coastal waters

    Nima Pahlevan, Brandon Smith, Caren Binding, Daniela Gurlin, Lin Li, Mariano Bresciani, and Claudia Giardino. Hyperspectral retrievals of phytoplankton absorption and chlorophyll-a in inland and nearshore coastal waters. Remote Sensing of Environment , 253:112200, 2021

    2021

  34. [42]

    Seamless retrievals of chlorophyll-a from sentinel-2 (msi) and sentinel-3 (olci) in inland and coastal waters: A machine- learning approach

    Nima Pahlevan, Brandon Smith, John Schalles, Caren Binding, Zhi- gang Cao, Ronghua Ma, Krista Alikas, Kersti Kangro, Daniela Gurlin, Nguyen H, et al. Seamless retrievals of chlorophyll-a from sentinel-2 (msi) and sentinel-3 (olci) in inland and coastal waters: A machine- learn...

    2020

  35. [43]

    Hypoxia in the gulf of mexico

    Nancy N Rabalais, R Eugene Turner, and William J Wiseman Jr. Hypoxia in the gulf of mexico. Journal of environmental quality , 30(2):320–329, 2001

    2001

  36. [44]

    Stochas- tic backpropagation and approximate inference in deep generative mod- els

    Danilo Jimenez Rezende, Shakir Mohamed, and Daan Wierstra. Stochas- tic backpropagation and approximate inference in deep generative mod- els. In International conference on machine learning, pages 1278–1286. PMLR, 2014

    2014

  37. [45]

    Spectrophotometric measurements of particulate absorption using filter pads

    Collin Roesler, Dariusz Stramski, Eurico D’Sa, R ¨udiger R ¨ottgers, and Rick A Reynolds. Spectrophotometric measurements of particulate absorption using filter pads. Washington, DC: IOCCG, 2018

    2018

  38. [46]

    Spectral beam attenuation coefficient retrieved from ocean color inversion

    Collin S Roesler and Emmanuel Boss. Spectral beam attenuation coefficient retrieved from ocean color inversion. Geophysical Research Letters, 30(9), 2003

    2003

  39. [47]

    Noise regularization for conditional density estimation

    Jonas Rothfuss, Fabio Ferreira, Simon Boehm, Simon Walther, Maxim Ulrich, Tamim Asfour, and Andreas Krause. Noise regularization for conditional density estimation. arXiv preprint arXiv:1907.08982 , 2019

    1907 arXiv

  40. [48]

    Learn- ing representations by back-propagating errors

    David E Rumelhart, Geoffrey E Hinton, and Ronald J Williams. Learn- ing representations by back-propagating errors. nature, 323(6088):533– 536, 1986

    1986

  41. [49]

    Understanding variation in trophic status of lakes on the boreal plain: A 20 year retrospective using landsat tm imagery

    GZ Sass, IF Creed, SE Bayley, and KJ Devito. Understanding variation in trophic status of lakes on the boreal plain: A 20 year retrospective using landsat tm imagery. Remote Sensing of Environment, 109(2):127– 141, 2007

    2007

  42. [50]

    Retrieval of atmospheric and oceanic properties from meris measurements: A new case-2 water pro- 16 cessor for beam

    Th Schroeder, M Schaale, and J Fischer. Retrieval of atmospheric and oceanic properties from meris measurements: A new case-2 water pro- 16 cessor for beam. International Journal of Remote Sensing, 28(24):5627– 5632, 2007

    2007

  43. [51]

    A chlorophyll-a algorithm for landsat- 8 based on mixture density networks

    Brandon Smith, Nima Pahlevan, John Schalles, Steve Ruberg, Reagan Errera, Ronghua Ma, Claudia Giardino, Mariano Bresciani, Claudio Barbosa, Tim Moore, et al. A chlorophyll-a algorithm for landsat- 8 based on mixture density networks. Frontiers in Remote Sensing , 1:623678, 2021

    2021

  44. [52]

    Ladder variational autoencoders

    Casper Kaae Sønderby, Tapani Raiko, Lars Maaløe, Søren Kaae Sønderby, and Ole Winther. Ladder variational autoencoders. Advances in neural information processing systems , 29, 2016

    2016

  45. [53]

    Topological generality and spectral dimensionality in the earth mineral dust source investigation (emit) using joint characterization and the spectral mixture residual

    Daniel Sousa and Christopher Small. Topological generality and spectral dimensionality in the earth mineral dust source investigation (emit) using joint characterization and the spectral mixture residual. Remote Sensing, 15(9):2295, 2023

    2023

  46. [54]

    Hyperspectral remote sensing of the pigment c-phycocyanin in turbid inland waters, based on optical classification

    Deyong Sun, Yunmei Li, Qiao Wang, Jay Gao, Chengfeng Le, Changchun Huang, and Shaoqi Gong. Hyperspectral remote sensing of the pigment c-phycocyanin in turbid inland waters, based on optical classification. IEEE Transactions on Geoscience and Remote Sensing , 51(7):3871–3884, 2013

    2013

  47. [55]

    Uniqueness in remote sensing of the inherent optical properties of ocean water

    Michael Sydor, Richard W Gould, Robert A Arnone, Vladimir I Haltrin, and Wesley Goode. Uniqueness in remote sensing of the inherent optical properties of ocean water. Applied optics, 43(10):2156–2162, 2004

    2004

  48. [56]

    An evaluation of models for the satellite-estimation of phytoplankton absorption coeffi- cients in coastal/oceanic waters

    Surya Prakash Tiwari and Palanisamy Shanmugam. An evaluation of models for the satellite-estimation of phytoplankton absorption coeffi- cients in coastal/oceanic waters. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 7(1):364–371, 2013

    2013

  49. [57]

    A ground-based remote sensing system for high-frequency and real-time monitoring of phytoplankton blooms

    Weijia Wang, Kun Shi, Yibo Zhang, Na Li, Xiao Sun, Dong Zhang, Yunlin Zhang, Boqiang Qin, and Guangwei Zhu. A ground-based remote sensing system for high-frequency and real-time monitoring of phytoplankton blooms. Journal of Hazardous Materials , 439:129623, 2022

    2022

  50. [58]

    Generalized ocean color inversion model for retrieving marine inherent optical properties

    P Jeremy Werdell, Bryan A Franz, Sean W Bailey, Gene C Feldman, Emmanuel Boss, Vittorio E Brando, Mark Dowell, Takafumi Hirata, Samantha J Lavender, ZhongPing Lee, et al. Generalized ocean color inversion model for retrieving marine inherent optical properties. Applied optics,...

    2019

  51. [59]

    Phytoplankton group identification using simulated and in situ hyper- spectral remote sensing reflectance

    Hongyan Xi, Martin Hieronymi, Hajo Krasemann, and R ¨udiger R¨ottgers. Phytoplankton group identification using simulated and in situ hyper- spectral remote sensing reflectance. Frontiers in Marine Science, 4:272, 2017

    2017

  52. [60]

    Hyperspectral differentiation of phytoplankton tax- onomic groups: a comparison between using remote sensing reflectance and absorption spectra

    Hongyan Xi, Martin Hieronymi, R ¨udiger R ¨ottgers, Hajo Krasemann, and Zhongfeng Qiu. Hyperspectral differentiation of phytoplankton tax- onomic groups: a comparison between using remote sensing reflectance and absorption spectra. Remote Sensing, 7(11):14781–14805, 2015

    2015

  53. [61]

    Infovae: Information maximizing variational autoencoders

    Shengjia Zhao, Jiaming Song, and Stefano Ermon. Infovae: Information maximizing variational autoencoders. arXiv preprint arXiv:1706.02262, 2017

    2017 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.