Dargana: fine-tuning EarthPT for dynamic tree canopy mapping from space

Freddie Kalaitzis; James Banister; James E. Geach; Luke Fleming; Michael J. Smith; Ryan J. Roberts

REVIEW 3 major objections 6 minor 14 references

Dargana: fine-tuning EarthPT for dynamic tree canopy mapping from space

T0 review · 3 major / 6 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 Fine-tuning a frozen EarthPT backbone on under 3% of its pre-training data yields a tree-canopy classifier that reaches ROC-AUC 0.98 on unseen satellite tiles.

desk verdict A useful and cheap specialisation of EarthPT for dynamic canopy mapping, but the headline sub-0.5 ha and temporal-change claims rest on contaminated NFI labels and a non-held-out case-study tile. read the letter →

arxiv 2504.17321 v1 pith:3OUDS3IK submitted 2025-04-24 physics.geo-ph cs.LG

Michael J. Smith , Luke Fleming , James E. Geach , Ryan J. Roberts , Freddie Kalaitzis , James Banister This is my paper · ORCID

classification physics.geo-phcs.LG

keywords earthobservationfoundationmodelfine-tuningtreecanopymappinglandcoverclassificationtemporalmonitoringNationalForestInventorySentinel-2deeplearning

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

The paper claims that a pre-trained Earth observation model can be specialised for a specific monitoring task at a fraction of the cost of training from scratch. Using a frozen EarthPT backbone and a small trainable MLP head, the fine-tuned model Dargana reaches pixel-level ROC-AUC 0.98 and PR-AUC 0.83 for tree versus not-tree classification on an unseen tile. It also separates conifer from broadleaved cover, picks out structures smaller than the training data's 0.5 hectare minimum unit, and shows increasing class probabilities as new woodland establishes. The significance is that large observational pre-training can be cheaply redirected toward granular, dynamic land-cover monitoring.

What carries the argument

The central mechanism is transfer from a frozen temporal backbone. EarthPT is a 300M-parameter transformer pre-trained on 30 billion 8x8 pixel patches of time-sequenced optical and SAR imagery; Dargana freezes its weights, attaches a small one-hidden-layer MLP (4096 units, GELU) to the central layer, and trains only that head with cross-entropy on the inventory labels. Because the backbone is frozen, fine-tuning uses under 3% of pre-training data and about 5% of pre-training compute, and the temporal structure inherited from EarthPT lets the model update class probabilities as new observations arrive.

What would settle it

Take a set of pixels that Dargana labels as tree but the inventory labels as not-tree, and check them against independent high-resolution reference data such as aerial imagery or lidar over several tiles; if most of these pixels are genuinely bare ground, the sub-0.5 ha extrapolation claim and the PR-AUC are inflated, and if most are genuine trees, the extrapolation claim is confirmed.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that large observation models can be specialised efficiently and productively: a 300M-parameter EarthPT model, with weights frozen and only a single-layer MLP attached to its central layer, becomes a dynamic tree-canopy classifier when fine-tuned on national forest inventory labels. Tested on the unseen SX36 tile, the model attains ROC-AUC 0.98 and PR-AUC 0.83 for the binary tree/not-tree task. The same model distinguishes conifer from broadleaved woodland within a single contiguous forest, detects hedgerows, coppice and isolated trees below the 0.5 ha minimum mapping unit of the training labels, and tracks the establishment of new planting over several years via rising per-pixel class probabilities.

Load-bearing premise

The reported scores and the below-limit detections assume the national forest inventory labels are correct enough at the pixel level to serve as ground truth, even though the inventory's 0.5-hectare minimum means some pixels labelled 'no tree' actually contain trees.

Editorial extensions

If this is right

County-scale dynamic tree cover maps can be produced at 10 m resolution and updated as new satellite observations arrive, enabling monitoring of planting and felling.
The same frozen backbone, with a swapped head, could be redirected to other land-cover vocabularies as soon as labelled data exist.
Because EarthPT is inherently scalable geographically, the approach should extend from county to national or continental coverage without architectural change.
Treating per-pixel class probabilities as covering fractions allows quantitative change detection, such as measuring the area of new woodland establishment over time.

Reading between the lines

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

The per-pixel probability output could be calibrated against field or lidar measurements to produce continuous canopy-cover fractions rather than hard classes, a testable extension not examined in the paper.
A similar recipe—frozen EarthPT plus small head—could be applied to other dynamic surface phenomena such as flooding, snow cover, or crop type with comparable compute savings.
Because the backbone remains frozen, multiple specialised heads could in principle coexist on one model, serving several monitoring tasks from a single forward pass.
Label-noise modelling of the inventory's 'other' class could quantify how much the reported PR-AUC is affected by contaminated negatives, tightening the main claim.

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.

Desk Editor's Note

read the letter

Quick read of 2504.17321. The genuinely useful result is that you can freeze a 300M EarthPT backbone, train only a small MLP head, and get a 10-m, time-stepped tree/not-tree/conifer/broadleaved classifier with a held-out-tile ROC-AUC of 0.98 and PR-AUC 0.83, using a few percent of the pre-training data and compute. That is a real proof of concept for cheaply specialising a large EO model. The authors are also transparent about their main data caveat, which is more than many workshop papers do.

Now the soft spots, in order. The label contamination is not a footnote. They say NFI 'other' pixels can contain trees because the inventory is contiguous-area limited and misses hedgerows, small coppice, isolated trees. Those are exactly the features used to support the below-0.5-ha extrapolation. With that label source, the PR-AUC and the sub-limit claim are not independently verified; the model could be correctly detecting unlabelled trees, but the metric cannot tell us that. The visual case studies on SX25 (Trenant, Duloe) are not on a held-out tile: SX25 is not in the nine unseen tiles, so those plots may partly reflect fine-tuning labels rather than true generalisation. The evaluation is otherwise narrow: one test tile, no confidence intervals, pixel-level metrics over spatially autocorrelated NFI polygons, and no cheaper baseline. A simple model on raw Sentinel-2 might do surprisingly well on tree/not-tree; without that comparison, 0.98 ROC-AUC is harder to interpret.

There are also two concrete inconsistencies to fix. Fig 3a: a 45 m radius circle is about 0.64 ha, yet the text reports a ~20 ha increase and the axis goes to 40 ha. Table 2 lists 150M active parameters during fine-tuning and 23 output classes; a one-layer 4096-wide MLP head is ~5M parameters and the NFI task has six or seven classes. Those look like errors, and they undermine confidence in the numbers.

I don't think any of this kills the paper. The core demonstration — frozen LOM plus a tiny head gives dynamic canopy maps cheaply — is credible and useful. But the headline novelty claims (sub-limit detection and temporal tracking) are currently supported by visual inspection and a possibly contaminated ground truth. This deserves a serious referee: I'd send it with requests for multi-tile validation, an independent label source, a baseline, honest uncertainty, and fixes to Fig 3 and Table 2. If those land, it would be a solid paper.

For a reader: someone working on EO foundation models or natural-capital monitoring will get ideas and a clear warning about label quality. The citation pattern is fine; the EarthPT self-citations are appropriate because it is the model being specialised. I'd cite it as a proof of concept, with caveats.

Recommendation: peer review, yes; accept, not yet — the current result is conditional.

Referee Report

3 major / 6 minor

Summary. The paper presents Dargana, a fine-tuned variant of the EarthPT time-series foundation model for tree canopy mapping. The authors freeze EarthPT's backbone and attach a small trainable MLP head, training on National Forest Inventory (NFI) labels for Cornwall, UK, using Sentinel-2 ClearSky and Sentinel-1 imagery. They report pixel-level ROC-AUC 0.98 and PR-AUC 0.83 on a held-out tile (SX36) for a tree/not-tree binary task, and present case studies suggesting detection of hedgerows, coppice, and isolated trees below the NFI's 0.5 ha minimum mapping unit, as well as tracking of new woodland establishment. They claim the fine-tuning uses less than 3% of pre-training data volume and 5% of pre-training compute.

Significance. If substantiated, the result is a useful demonstration that a pre-trained Earth Observation foundation model can be cheaply specialized for dynamic, granular land-cover monitoring, with practical relevance for natural capital accounting. The strengths are the held-out test tile (SX36 was excluded from both pre-training and fine-tuning), the external anchoring to NFI labels, and the transparent reporting of fine-tuning cost. However, the quantitative claims rest on a single tile, on labels the authors acknowledge to be contaminated for the very objects of interest, and on a temporal-change calculation that is internally inconsistent. The paper does not report code, data, or uncertainty intervals, and the qualitative sub-0.5 ha claims are not validated against independent ground truth.

major comments (3)

[§2, 'Pre-training and fine-tuning datasets'] The authors state that pixels labelled 'other' (used as 'not-tree') will contain trees in some cases, since the NFI is contiguous-area limited and misses hedgerows, small coppice, and isolated trees. These missing features are precisely the objects that Section 3 and Figure 3b claim Dargana can detect below the 0.5 ha limit. Consequently, the reported PR-AUC of 0.83 on SX36 is computed with true tree pixels marked as negatives, so the metric conflates correct detections of unlabelled trees with false positives, and the sub-0.5 ha extrapolation claim has no quantitative support because the evaluation set lacks those features as positive labels. This issue is load-bearing for the headline metrics and the central novelty claim; independent pixel-level validation (e.g., from aerial imagery or LiDAR interpretation) on a held-out area is required.
[§3, Fig. 3a] The dashed circle has radius r = 45 m, giving an area of approximately 0.64 ha, yet the text reports a '~20 ha increase over five years' and the chart's vertical axis extends to 40 ha. Interpreting class probability as a covering fraction within the circular area of interest cannot yield an area larger than 0.64 ha, so the quantitative evidence for tracking new woodland establishment is internally inconsistent as reported. The authors should either correct the area calculation, specify a larger region of analysis, or explain the discrepancy; as it stands, this supporting evidence is invalid.
[§3 and Table 1] The performance evaluation on SX36 is a single tile with no confidence intervals, no per-year or per-class breakdown, and no comparison across the other held-out tiles. Pixel-level metrics computed on spatially autocorrelated NFI polygons and satellite imagery overstate the effective sample size, so the point estimates in Table 1 provide limited evidence for generalization. Reporting per-tile AUCs across the nine held-out tiles, with block-bootstrap or polygon-clustered confidence intervals, would materially strengthen the main performance claim.

minor comments (6)

[Table 2] The 'Active parameters' row lists 300M for pre-training and 150M for fine-tuning, which is inconsistent with the text's statement that the 300M base model is frozen and only the MLP head is trained; please clarify whether 150M refers to the trainable parameters or to a differently sized model.
[Table 2] The 'Output classes' entry for fine-tuning is 23, while Section 2 lists seven NFI class labels; please clarify the relationship between these numbers and the multi-class prediction setup.
[Fig. 6] The ROC and PR curves lack annotations for the operating threshold used in the accuracy calculations; adding threshold markers would help readers interpret the trade-off.
[§3] The description of combining 'all tree-like classes into a single class' does not specify how the multi-class probabilities are aggregated to compute the binary AUC; please state whether the sum or maximum over tree classes is used.
[Fig. 2 caption] The caption reads 'comprises of a pre-trained frozen EarthPT base model'; the correct phrasing is 'comprises a pre-trained...'.
[§3, Fig. 3a] The term 'annual aggregated probability' is not defined; please specify whether it is a mean, maximum, or other aggregation over observations within each calendar year.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the fine-tuned model is evaluated against external NFI ground truth on a tile held out from both pre-training and fine-tuning, so the reported performance is not forced by construction.

full rationale

The paper's derivation chain is self-contained with respect to circularity. Dargana is constructed by freezing the self-cited EarthPT backbone and training only a small MLP head on Forestry Commission NFI labels. The headline metrics (ROC-AUC 0.98, PR-AUC 0.83) are computed on the unseen SX36 tile, which the paper states is 'kept unseen throughout the pre-training and fine-tuning process', and are compared directly against external NFI ground-truth labels rather than against any quantity fitted from that tile. The claim of detecting features below the NFI 0.5 ha minimum mapping unit is qualitative and supported visually; it is not derived by fitting to those features, so it is not circular, though it may be a validity concern. The paper explicitly acknowledges that NFI 'other' labels may contain trees ('pixels labelled other will contain trees in some cases, since the inventory is contiguous-area limited'), which is a data-quality limitation affecting the reliability of the AUC numbers, but this is not a case of the prediction being equivalent to its inputs. The self-citation to Smith et al. (2023) for EarthPT is a normal citation to prior work; the current evaluation provides independent evidence that the frozen representations are useful for the downstream task, so the self-citation is not load-bearing in a circular sense. The apparent inconsistency in the Fig. 3a area estimate (45 m radius circle vs. reported ~20 ha increase) is a quantitative error, not a circularity. No step in the paper reduces by definition or by fitted-parameter renaming to its own inputs.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard supervised fine-tuning of a frozen pre-trained backbone. The only fitted component is the MLP head, trained on NFI labels in the disclosed cross-entropy setup, which is the honest and standard fit. No hidden constants were introduced except the unstated class-prediction threshold. The domain assumptions concern label quality, regional representativeness, pixel independence in evaluation, and spectral discriminability, each of which is standard for remote sensing work but not independently verified.

free parameters (1)

class prediction threshold = not reported (implicitly 0.5)
Used to convert output probabilities into discrete class predictions for the maps and for Table 1's 'class predictions'; it does not affect the reported AUCs, which are rank-based.

assumptions (4)

domain assumption NFI polygons provide acceptable pixel-level labels of tree presence, with 'other' pixels treated as not-tree despite known contamination from sub-0.5 ha trees.
Acknowledged in §2 ('pixels labelled other will contain trees in some cases'); the §3 AUC evaluation and the below-limit extrapolation claims both depend on this label quality.
domain assumption Imagery from 2015-present over the South West of England is sufficiently representative for fine-tuning on Cornwall and testing on SX36, and temporal coverage spans the evaluation years.
Pre-training and fine-tuning data are regional (§2); generalization outside the region is not demonstrated, and the model is evaluated only on tiles in the same BNG region.
domain assumption Pixel-level ROC/PR statistics treat each pixel as an independent sample.
§3 reports pixel-level AUCs; contiguous NFI woodland polygons and 10m imagery create strong spatial autocorrelation, so the effective number of independent samples is far below the pixel count.
domain assumption Sentinel-2-equivalent optical and Sentinel-1 SAR temporal signatures suffice to separate conifer, broadleaved, and not-tree at 10m.
This is the usual remote-sensing assumption implicit in the choice of inputs (§2); the paper does not test whether simpler spectral features would perform similarly.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Dargana: fine-tuning EarthPT for dynamic tree canopy mapping from space." pith.science (2026). https://pith.science/paper/3OUDS3IK

@misc{pith2026250417321,
  author       = {Pith},
  title        = {Pith review of: Dargana: fine-tuning EarthPT for dynamic tree canopy mapping from space},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3OUDS3IK}},
  note         = {Machine review of arXiv:2504.17321}
}

read the original abstract

We present Dargana, a fine-tuned variant of the EarthPT time-series foundation model that achieves specialisation using <3% of its pre-training data volume and 5% of its pre-training compute. Dargana is fine-tuned to generate regularly updated classification of tree canopy cover at 10m resolution, distinguishing conifer and broadleaved tree types. Using Cornwall, UK, as a test case, the model achieves a pixel-level ROC-AUC of 0.98 and a PR-AUC of 0.83 on unseen satellite imagery. Dargana can identify fine structures like hedgerows and coppice below the training sample limit, and can track temporal changes to canopy cover such as new woodland establishment. Our results demonstrate how pre-trained Large Observation Models like EarthPT can be specialised for granular, dynamic land cover monitoring from space, providing a valuable, scalable tool for natural capital management and conservation.

Figures

Figures reproduced from arXiv: 2504.17321 by the authors.

**Figure 2.** Dargana comprises of a pre-trained frozen EarthPT base model (Φ) and a trainable MLP head (Ψ) that is appropriate for the downstream task at hand. In this study that task is tree canopy type classification. Pre-training and fine-tuning datasets. Our pre-training data consists of temporal sequences of ClearSky2 and Sentinel-1 C-band imagery of the South West of England, from 2015 to the present day. The imagery is … view at source ↗

**Figure 3.** Case studies showing (a): evidence of Dargana’s ability to track change and (b): informa [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: A larger copy of Fig. 1 [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

**Figure 5.** Figure 5: Class probabilities for conifer and broadleaved tree types for SX25. (Top) pixels are [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: Left: ROC curve for the tree/not-tree binary classification task on SX36 described in §3. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 4 canonical work pages

[5]

doi: 10.48550/arXiv.2304.07193. M. J. Smith and J. E. Geach. Astronomia ex machina: a history, primer and outlook on neural networks in astronomy. Royal Society Open Science , 10(5),

arXiv doi:10.48550/arxiv.2304.07193
[6]

ISSN 2054-5703. doi:

work page
[7]

doi: 10.48550/arXiv.2309.07207. M. J. Smith, R. J. Roberts, E. Angeloudi, and M. Huertas-Company. AstroPT: Scaling Large Obser- vation Models for Astronomy. ArXiv e-prints,

arXiv doi:10.48550/arxiv.2309.07207
[9]

doi: 10.1016/j.rse.2023.113888

ISSN 0034-4257. doi: 10.1016/j.rse.2023.113888. H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozi `ere, N. Goyal, E. Hambro, F. Azhar, A. Rodriguez, A. Joulin, E. Grave, and G. Lample. LLaMA: Open and Efficient Foundation Language Models. ArXiv e-prints,

arXiv 2023
[10]

doi: 10.48550/arXiv.2302.13971. P. Villalobos, A. Ho, J. Sevilla, T. Besiroglu, L. Heim, and M. Hobbhahn. Will we run out of data? Limits of LLM scaling based on human-generated data. ArXiv e-prints,

arXiv doi:10.48550/arxiv.2302.13971
[12]

Curran Associates Inc. 6 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 A H YPERPARAMETERS Table 2: Hyperparameters used in pre-training EarthPT and fine-tuning Dargana. The underlying EarthPT model is trained on approximately 30B observation tokens across 640,000 ha of the South West of England. Fine-tuning is...

work page 2025
[13]

1 showing regions classified as dom- inated by trees, as predicted by Dar- gana across Cornwall, UK on 31st December

Output classes – 23 7 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 B A DDITIONAL RESULTS Figure 4: A larger copy of Fig. 1 showing regions classified as dom- inated by trees, as predicted by Dar- gana across Cornwall, UK on 31st December

work page 2025
[14]

8 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 Figure 5: Class probabilities for conifer and broadleaved tree types for SX25. (Top) pixels are coloured according to class probability on a logarithmic scaling to emphasise fine structure; (mid- dle) pixels are coloured green where the probability was the maximu...

work page 2025

Show all 14 references

[2016]

doi: 10.48550/arXiv.1606.08415. J. Jakubik, S. Roy, C. E. Phillips, P. Fraccaro, D. Godwin, B. Zadrozny, D. Szwarcman, C. Gomes, G. Nyirjesy, B. Edwards, D. Kimura, N. Simumba, L. Chu, S. K. Mukkavilli, D. Lambhate, K. Das, R. Bangalore, D. Oliveira, M. Muszynski, K. Ankur, M....

doi arXiv
[2017]

doi: 10.48550/arXiv.1711.05101. M. Oquab, T. Darcet, T. Moutakanni, H. V o, M. Szafraniec, V . Khalidov, P. Fernandez, D. Haziza, F. Massa, A. El-Nouby, M. Assran, N. Ballas, W. Galuba, R. Howes, P.-Y . Huang, S.-W. Li, I. Misra, M. Rabbat, V . Sharma, G. Synnaeve, H. Xu, H. J...

doi arXiv
[2022]

5 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 F

doi: 10.48550/ arXiv.2211.04325. 5 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 F. Xue, Y . Fu, W. Zhou, Z. Zheng, and Y . You. To repeat or not to repeat: insights from scaling llm under token-crisis. In Proceedings of the 37th I...

2025 doi arXiv
[2023]

doi: 10.48550/arXiv.2310.18660. I. Loshchilov and F. Hutter. Decoupled Weight Decay Regularization. ArXiv e-prints,

doi arXiv
[2024]

doi: 10.48550/arXiv.2405.14930. J. Tolan, H.-I. Yang, B. Nosarzewski, G. Couairon, H. V . V o, J. Brandt, J. Spore, S. Majumdar, D. Haziza, J. Vamaraju, T. Moutakanni, P. Bojanowski, T. Johns, B. White, T. Tiecke, and C. Cou- prie. Very high resolution canopy height maps from ...

doi arXiv
[2025]

doi: 10.48550/arXiv.2501.12948. D. Hendrycks and K. Gimpel. Gaussian Error Linear Units (GELUs). ArXiv e-prints,

doi arXiv

Pith tools

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

[1] [5]

doi: 10.48550/arXiv.2304.07193. M. J. Smith and J. E. Geach. Astronomia ex machina: a history, primer and outlook on neural networks in astronomy. Royal Society Open Science , 10(5),

arXiv doi:10.48550/arxiv.2304.07193

[2] [6]

ISSN 2054-5703. doi:

work page

[3] [7]

doi: 10.48550/arXiv.2309.07207. M. J. Smith, R. J. Roberts, E. Angeloudi, and M. Huertas-Company. AstroPT: Scaling Large Obser- vation Models for Astronomy. ArXiv e-prints,

arXiv doi:10.48550/arxiv.2309.07207

[4] [9]

doi: 10.1016/j.rse.2023.113888

ISSN 0034-4257. doi: 10.1016/j.rse.2023.113888. H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozi `ere, N. Goyal, E. Hambro, F. Azhar, A. Rodriguez, A. Joulin, E. Grave, and G. Lample. LLaMA: Open and Efficient Foundation Language Models. ArXiv e-prints,

arXiv 2023

[5] [10]

doi: 10.48550/arXiv.2302.13971. P. Villalobos, A. Ho, J. Sevilla, T. Besiroglu, L. Heim, and M. Hobbhahn. Will we run out of data? Limits of LLM scaling based on human-generated data. ArXiv e-prints,

arXiv doi:10.48550/arxiv.2302.13971

[6] [12]

Curran Associates Inc. 6 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 A H YPERPARAMETERS Table 2: Hyperparameters used in pre-training EarthPT and fine-tuning Dargana. The underlying EarthPT model is trained on approximately 30B observation tokens across 640,000 ha of the South West of England. Fine-tuning is...

work page 2025

[7] [13]

1 showing regions classified as dom- inated by trees, as predicted by Dar- gana across Cornwall, UK on 31st December

Output classes – 23 7 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 B A DDITIONAL RESULTS Figure 4: A larger copy of Fig. 1 showing regions classified as dom- inated by trees, as predicted by Dar- gana across Cornwall, UK on 31st December

work page 2025

[8] [14]

8 Published as a workshop paper at ‘Tackling Climate Change with Machine Learning’, ICLR 2025 Figure 5: Class probabilities for conifer and broadleaved tree types for SX25. (Top) pixels are coloured according to class probability on a logarithmic scaling to emphasise fine structure; (mid- dle) pixels are coloured green where the probability was the maximu...

work page 2025