Pith. sign in

REVIEW 2 major objections 6 minor 51 references

TS-SatFire: A Multi-Task Satellite Image Time-Series Dataset for Wildfire Detection and Prediction

T0 review · 2 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A multi-task VIIRS time-series dataset supports active fire detection, burned area mapping, and next-day spread prediction from one datacube.

desk verdict Useful multi-task wildfire dataset, but the active-fire test labels are thresholded from the same bands the models see, making that benchmark partially circular. read the letter →

arxiv 2412.11555 v1 pith:BTW4QDXD submitted 2024-12-16 cs.CV cs.AI

Yu Zhao , Sebastian Gerard , Yifang Ban This is my paper · ORCID
classification cs.CVcs.AI
keywords wildfiredetectionactivefireburnedareamappingnext-dayprogressionpredictionVIIRSsatelliteimagerymulti-taskdeeplearningremotesensingbenchmarkdataset
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

TS-SatFire is a remote-sensing dataset built from VIIRS satellite image time series of 179 wildfire events in the contiguous U.S. between 2017 and 2021, totaling 3,552 surface reflectance images and 71 GB of data when auxiliary weather, topography, land cover, and fuel layers are included. The paper's central claim is that this single dataset can support three related wildfire tasks at once: pixel-level active fire detection, daily burned area mapping, and next-day progression prediction, with manually quality-assured test labels for the first two. To back that claim, the authors run benchmark models spanning 1D pixel time-series, 2D image, and 3D spatial-temporal architectures and report that temporal and spatial-temporal models generally beat single-image models for detection. They also report that next-day prediction remains much harder than detection, with best F1 around 0.375. The significance is that a shared multi-task benchmark of this kind is a precondition for training deep learning models that monitor and anticipate fire behavior from the same sensor.

What carries the argument

The load-bearing object is the multi-temporal VIIRS datacube: six spectral bands (I1-I5 and M11) at 375-750 m resolution with day and night captures, aggregated over a sliding time window of length T, plus co-registered auxiliary channels for weather, weather forecast, topography, land cover, and fuel. Active fire labels come from the VIIRS AF product with manual visual inspection; burned area labels are the union of accumulated AF detections and NIFC perimeters; prediction labels are the daily difference of burned area masks. This design is what makes the three tasks share one input representation and lets the same model family be compared across detection and prediction.

What would settle it

Hold out a subset of test fire events with same-day high-resolution reference imagery (for example, 10-20 m Sentinel-2 scenes or airborne thermal data) and compare the active fire model's detections against that reference. If the model agrees with the manual I4/I5 threshold but not with the high-resolution fire perimeter, the active fire benchmark does not measure true detection accuracy.

Watch

Extended reading notes

Core claim

The central discovery is the dataset itself, together with the finding that its design supports three complementary tasks from one VIIRS datacube. The authors show that active fire detection benefits from temporal context, with the transformer-based T4Fire reaching an F1 of 0.802 on 6-day pixel time series and UNETR-3D reaching 0.811, while burned area mapping is best handled by SwinUNETR-3D with an F1 of 0.855. The progression prediction task is formulated so that the model predicts only the newly burned area, defined as the difference between consecutive burned area masks, to prevent a model from trivially copying the previous day's burn scar. On that task all tested segmentation models perform modestly, with best F1 about 0.375, which the authors interpret as showing that standard image segmentation architectures are not sufficient out of the box for forecasting fire spread. The paper thus claims the dataset provides a reusable foundation for developing and evaluating multi-task wildfire models, not a solved prediction system.

Load-bearing premise

The active fire test labels are produced by manually setting a threshold on the same I4/I5 spectral bands that the model receives as input, so the reported active-fire scores assume that this threshold is a valid reference for true fire locations.

Editorial extensions

If this is right

  • A single VIIRS datacube can be used to train models for both status monitoring and forecasting, removing the need to assemble separate datasets for detection and prediction.
  • Temporal context improves active fire detection: both the transformer T4Fire and the 3D UNETR outperform single-image spatial models on the active fire task.
  • Daily burned area mapping at 375-750 m is feasible with spatial-temporal models, a step beyond monthly burned area products.
  • The next-day progression task is not solved by standard segmentation models; the low auxiliary-feature importance suggests the bottleneck is likely model design and data fusion, not data availability.
  • The dataset's benchmark results provide a reproducible baseline against which future multi-task wildfire models, including foundation models, can be measured.

Reading between the lines

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

  • Inference: because the AF test labels are created by thresholding the same I4/I5 bands the models see, the reported F1 values likely measure agreement with that threshold rather than with independently verified fire locations; a high-resolution reference subset would quantify the gap.
  • Inference: the near-zero contribution of most auxiliary variables to the prediction task points to model architecture and fusion strategy, not data availability, as the main obstacle to next-day spread forecasting.
  • Inference: the training events are CONUS-only while the AF test events span multiple continents, so retraining or fine-tuning on global data is a natural next test of the dataset's generality.
  • Inference: the 'predict only newly burned area' label design could transfer to other hazard progression problems where copying yesterday's state is a trivial baseline.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 6 minor

Summary. The paper introduces TS-SatFire, a multi-task satellite image time-series dataset for wildfire research, covering 179 wildfire events in the contiguous U.S. from 2017 to 2021 (plus additional global test events for active fire detection), with VIIRS imagery and auxiliary weather, topography, land cover, and fuel data. The dataset supports three tasks: active fire (AF) detection, burned area (BA) mapping, and next-day fire progression prediction. The authors describe data sources, label generation (with claimed manual quality assurance of test labels), preprocessing, and benchmark baselines using spatial, temporal, and spatial-temporal deep learning models. Quantitative results are reported for all tasks, including seed variance and ablations on time-series length and feature importance.

Significance. If the label-quality concerns are resolved, TS-SatFire would be a valuable community resource: it is one of the few datasets that jointly supports detection and prediction tasks, it integrates multiple auxiliary modalities, it provides standardized baselines with reported seed variance, and the authors release both the dataset and processing/benchmark code. The multi-task framing is timely for Earth-observation foundation models. The explicit ablation studies (time-series length, feature importance) are a useful contribution. The main concern is that the active-fire test labels are not independent of the model input features, which weakens the benchmark claim; the burned-area test labels also rely on a subjective source-selection process. These issues are significant but appear addressable within the manuscript's scope.

major comments (2)
  1. [Labels / Active Fire Label; Table 1] The active-fire test labels are generated by 'manually set[ting] the threshold to Band I4/I5' (paragraph 2 of 'Labels / Active Fire Label'), while Bands I4 and I5 are listed as input channels in Table 1 and used directly in the active-fire detection input (per 'Preprocessing'). A model can therefore reproduce the test labels almost exactly by learning a threshold on input features, so the reported F1/IoU values for the AF task largely measure agreement with a hand-tuned threshold rather than independent fire-detection accuracy. The thresholds, the manual selection criteria, and the label-generation code for the AF test set are not reported or released, making the benchmark non-reproducible. Please report the threshold values, add a simple threshold baseline on I4/I5, and validate the AF test labels against an independent reference (e.g., higher-resolution fire perimeters or Landsat/Sentinel-2 burn-scar data) for at least a subset of events.
  2. [Labels / Burned Area Label] The burned-area test labels are chosen by visually inspecting which of two candidate sources (accumulated VIIRS AF detections plus NIFC perimeters versus AF accumulations only) 'better covers the burned area' (third paragraph of 'Labels / Burned Area Label'). No quantitative criteria, counts of how often each option was chosen, or inter-annotator agreement are given, and the selected labels still inherit potential errors from the VIIRS AF product. Because the BA task is one of the three headline tasks, this subjective and partly AF-derived labeling procedure should be justified with a quantitative evaluation on a sample of events against an independent burned-area reference, such as NIFC final perimeters or high-resolution imagery.
minor comments (6)
  1. [Table 1] The channel names contain typos: 'Min Tempreture', 'Max Tempreture', and 'Forecast Tempreture' should read 'Min Temperature', 'Max Temperature', and 'Forecast Temperature'.
  2. [Preprocessing] The text 'the pixel-wise median of GRIDNET weather data' should read 'GRIDMET'; please correct the dataset name.
  3. [Abstract / Methods / Spatial distribution of fires] The abstract states the dataset covers contiguous U.S. fires from January 2017 to October 2021, but the Methods section says the active-fire test set uses 17 events between 2018 and 2022 across multiple continents; please clarify the actual temporal and geographic coverage of the full dataset, including the 2022 events.
  4. [Preprocessing] The normalization applied to the VIIRS arrays before model input is not specified; please state the normalization procedure (e.g., min-max scaling per band or per image).
  5. [Table 3 / Technical Validation] The model is referred to as 'T4-Fire' in Table 3 but 'T4Fire' in the text; please use a consistent name.
  6. [Figure 2] The caption 'The legend in the test set does not cover any fires' is confusing; please rephrase to describe what the legend shows (or does not show).

Circularity Check

1 steps flagged · score 6.0 of 10

AF test labels are thresholds on the same I4/I5 bands the models see, making the active-fire benchmark partially circular.

  1. fitted input called prediction [Labels / Active Fire Label; Preprocessing; Table 1]
    "For the test labels, we manually set the threshold to Band I4/I5 to ensure alignment with the bright spots observed in the images. ... For active fire detection, the arrays from the GeoTIFF file are directly used as input after the normalization."

    The AF test label is constructed by manually thresholding Bands I4/I5, and the same Bands I4 and I5 are input channels 4 and 5 in Table 1, with AF inputs used directly after normalization. Thus, the test target is, by construction, a deterministic function of the input features. A model's F1/IoU on the AF task largely measures how well it reproduces the manually set I4/I5 threshold rather than how well it detects fires against independent ground truth. The threshold values and manual selection criteria are not reported, and the quality-control inspection itself was performed on the same I3-I5/M11 bright-spot appearance, so no external validation is provided for the AF labels.

full rationale

The paper's central claim is that TS-SatFire provides accurate, reusable labels that support a meaningful benchmark for active fire detection. The AF test labels are created by manually thresholding the same I4/I5 bands that are the model inputs (Table 1, Preprocessing), so the benchmark target is derived from the input features by construction. This is a label-construction circularity that undercuts the AF task as an independent benchmark, since a model could approximate the test labels by learning a threshold on Bands I4/I5 alone. The burned-area labels also inherit VIIRS AF detections, further reducing label independence between tasks. The prediction task labels are defined as differences of the burned-area labels, which is a legitimate design choice to avoid trivial next-day overlap, but it inherits the same label-construction issues. No self-citation chain is load-bearing here; the cited WildfireSpreadTS and prior work provide independent baselines and dataset formats. However, the AF label circularity directly affects one of the three headline tasks, so the score is 6 rather than lower. The paper could reduce this by reporting threshold values, providing independent ground truth (e.g., high-resolution fire perimeters), or releasing label-generation code for verification.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new physical entities. Its main dependencies are on the correctness of external data sources and on label construction choices. The AF test label threshold is the most significant unquantified choice, and the BA label union inherits errors from its component sources.

free parameters (2)
  • AF test label threshold = unreported
    Manual threshold on Band I4/I5 used to create active fire test labels; value not specified in the paper, making the test labels non-reproducible and partially determined by this choice.
  • Prediction loss class weight = 446.7836
    Weight for the burned area class in DiceCELoss, based on class distribution in the training set. This is a manually chosen value used in the prediction benchmark.
assumptions (3)
  • domain assumption VIIRS L1B data and geolocation are accurate after co-registration.
    The dataset relies on the geometric accuracy of VNP02IMG and VNP03IMG products for pixel-level labels and inputs.
  • domain assumption NIFC perimeters and GlobFire event boundaries provide a reliable basis for burned area labels.
    The BA labels are the union of accumulated VIIRS AF detections and NIFC perimeters; the paper itself notes low accuracy at the beginning of fires and potential commission errors.
  • domain assumption Manual visual inspection is a valid quality control for AF and BA labels.
    The paper uses manual inspection to remove AF training labels and to choose between BA label generation methods on the test set, without providing inter-validator agreement or an objective protocol.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TS-SatFire: A Multi-Task Satellite Image Time-Series Dataset for Wildfire Detection and Prediction." pith.science (2026). https://pith.science/paper/BTW4QDXD

@misc{pith2026241211555,
  author       = {Pith},
  title        = {Pith review of: TS-SatFire: A Multi-Task Satellite Image Time-Series Dataset for Wildfire Detection and Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BTW4QDXD}},
  note         = {Machine review of arXiv:2412.11555}
}
read the original abstract

Wildfire monitoring and prediction are essential for understanding wildfire behaviour. With extensive Earth observation data, these tasks can be integrated and enhanced through multi-task deep learning models. We present a comprehensive multi-temporal remote sensing dataset for active fire detection, daily wildfire monitoring, and next-day wildfire prediction. Covering wildfire events in the contiguous U.S. from January 2017 to October 2021, the dataset includes 3552 surface reflectance images and auxiliary data such as weather, topography, land cover, and fuel information, totalling 71 GB. The lifecycle of each wildfire is documented, with labels for active fires (AF) and burned areas (BA), supported by manual quality assurance of AF and BA test labels. The dataset supports three tasks: a) active fire detection, b) daily burned area mapping, and c) wildfire progression prediction. Detection tasks use pixel-wise classification of multi-spectral, multi-temporal images, while prediction tasks integrate satellite and auxiliary data to model fire dynamics. This dataset and its benchmarks provide a foundation for advancing wildfire research using deep learning.

Figures

Figures reproduced from arXiv: 2412.11555 by the authors.

Figure 1
Figure 1. Major components of the TS-SatFire dataset and three distinct tasks. products from these sensors offer a twice-daily temporal resolution, VIIRS holds an advantage in spatial resolution, providing 375 meters compared to the 1000 meters offered by MODIS and SLSTR17. VIIRS AF products are generated from VIIRS images using a multi-spectral contextual method that involves multi-criteria thresholds18. However, significant… view at source ↗
Figure 2
Figure 2. Overview of locations and land covers of fires in an example split into train/test/validation set. The legend in the test set does not cover any fires. wavelength of 3.7 µm, is particularly effective for detecting active fires due to its alignment with the black-body radiation curve peak of forest fires. Similarly, Long-wave Infrared Band I5 is used to detect higher-temperature fires. Both Bands I4 and I5 provide tw… view at source ↗
Figure 3
Figure 3. Percentage of missing values: Rates of missing values of each spectral band used in active fire detection, burned area mapping and fire progression prediction tasks. All the missing values are replaced with zeros during training and testing. October 2021. It comprises 3552 surface reflectance images along with auxiliary data such as weather, topography, land cover, and fuel information, amounting to a total size of … view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Feature Importance of the input bands for BA task (SwinUNETR-3D) and AF task (T4Fire). suggesting some advantage in leveraging limited temporal context. For the prediction task, both F1 and IoU Scores increase as the time-series length grows, likely because forecasting…
Figure 5
Figure 5. Figure 5: Feature importance for fire prediction task: We investigate a simple measure of feature importance, by setting one feature to zero in all inputs and measuring the resulting test F1 score. Removing important features should reduce the performance while removing unimport…
Figure 6
Figure 6. Figure 6: Results of UNETR-3D for active fire detection task. Fire ID: Currowan Fire, New South Wales, Australia. Burned Area Mapping For the burned area mapping task, the qualitative results are shown in [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Results of SwinUNETR for burned area mapping task. Fire ID: CA3658211879520210912. Fire prediction For the fire prediction task, the same study regions as the burned area mapping task are used. From [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Results of SwinUNETR for fire progression prediction task. Fire ID: NM3323810847220210520. dates as well as the region of interest, are organized in CSV files under the roi folder, with separate files for each year. Code Availability The python scripts used to process …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 44 canonical work pages

  1. [1]

    Andela, N. et al. A human-driven decline in global burned area. Science 356, 1356 – 1362 (2017). 13/16

    work page 2017

  2. [2]

    G., Slay, C

    Curtis, P. G., Slay, C. M., Harris, N. L., Tyukavina, A. & Hansen, M. C. Classifying drivers of global forest loss. Science 361, 1108 – 1111 (2018)

    work page 2018

  3. [3]

    Tyukavina, A. et al. Global trends of forest loss due to fire from 2001 to 2019. In Frontiers in Remote Sensing (2022)

    work page 2022

  4. [4]

    Wooster, M. J.et al. Satellite remote sensing of active fires: History and current status, applications and future requirements. Remote. Sens. Environ. (2021)

    work page 2021

  5. [5]

    V ., Roy, D

    Giglio, L., Loboda, T. V ., Roy, D. P., Quayle, B. & Justice, C. O. An active-fire based burned area mapping algorithm for the MODIS sensor. Remote. Sens. Environ. 113, 408–420 (2009)

    work page 2009

  6. [6]

    & Csiszar, I

    Schroeder, W., Oliva, P., Giglio, L. & Csiszar, I. The new VIIRS 375 m active fire detection data product: Algorithm description and initial assessment. Remote. Sens. Environ. (2014)

    work page 2014

  7. [7]

    Validation of GOES and MODIS active fire detection products using ASTER and ETM+ data

    Schroeder, W.et al. Validation of GOES and MODIS active fire detection products using ASTER and ETM+ data. Remote. Sens. Environ. (2008)

    work page 2008

  8. [8]

    & Justice, C

    Giglio, L., Schroeder, W. & Justice, C. The collection 6 MODIS active fire detection algorithm and fire products. Remote. sensing environment (2016)

    work page 2016

Show all 51 references
  1. [9]

    & Chuvieco, E

    Lizundia-Loiola, J., Otón, G., Ramo, R. & Chuvieco, E. A spatio-temporal active-fire clustering approach for global burned area mapping at 250 m from modis data. Remote. Sens. Environ. 236, 111493 (2020)

    2020

  2. [10]

    H., Fusioka, A

    de Almeida Pereira, G. H., Fusioka, A. M., Nassu, B. T. & Minetto, R. Active fire detection in Landsat-8 imagery: A large-scale dataset and a deep-learning study. ISPRS J. Photogramm. Remote. Sens. 178, 171–186 (2021)

    2021

  3. [11]

    Ban, Y ., Zhang, P., Nascetti, A., Bevington, A. R. & Wulder, M. Near real-time wildfire progression monitoring with Sentinel-1 SAR time series and deep learning. Sci. Reports 10 (2020)

    2020

  4. [12]

    & Sullivan, J

    Gerard, S., Zhao, Y . & Sullivan, J. Wildfirespreadts: A dataset of multi-modal time series for wildfire spread prediction. In Oh, A. et al. (eds.) Advances in Neural Information Processing Systems, vol. 36, 74515–74529 (Curran Associates, Inc., 2023)

    2023

  5. [14]

    Justice, C. O. et al. The modis fire products. Remote. Sens. Environ. 83, 244–262 (2002)

    2002

  6. [15]

    Giglio, L., Descloitres, J., Justice, C. O. & Kaufman, Y . J. An enhanced contextual fire detection algorithm for MODIS. Remote. Sens. Environ. 87, 273–282 (2003)

    2003

  7. [16]

    & Zhang, T

    Xu, W., Wooster, M., He, J. & Zhang, T. First study of Sentinel-3 SLSTR active fire detection and FRP retrieval: Night-time algorithm enhancements and global intercomparison to MODIS and VIIRS AF products. Remote. Sens. Environ. 248, 111947 (2020)

    2020

  8. [17]

    & Schroeder, W

    Oliva, P. & Schroeder, W. Assessment of VIIRS 375 m active fire detection product for direct burned area mapping. Remote. Sens. Environ. 160, 144–155 (2015)

    2015

  9. [18]

    & Csiszar, I

    Schroeder, W., Oliva, P., Giglio, L. & Csiszar, I. The new VIIRS 375 m active fire detection data product: Algorithm description and initial assessment. Remote. Sens. Environ. 143, 85–96 (2014)

    2014

  10. [19]

    Zhu, X. et al. Deep learning in remote sensing: a review. ArXiv abs/1710.03959 (2017)

    2017 arXiv

  11. [20]

    & Ban, Y

    Zhao, Y . & Ban, Y . GOES-R time series for early detection of wildfires with deep GRU-network.Remote. Sens. 14, 4347 (2022)

    2022

  12. [21]

    & Sullivan, J

    Zhao, Y ., Ban, Y . & Sullivan, J. Tokenized time-series in satellite image segmentation with transformer network for active fire detection. IEEE Transactions on Geosci. Remote. Sens. 61, 1–13 (2023)

    2023

  13. [22]

    P., Lewis, P

    Roy, D. P., Lewis, P. & Justice, C. O. Burned area mapping using multi-temporal moderate spatial resolution data—a bi-directional reflectance model-based expectation approach. Remote. Sens. Environ. 83, 263–286 (2002)

    2002

  14. [23]

    Roy, D. P. Multi-temporal active-fire based burn scar detection algorithm. Int. J. Remote. Sens. 20, 1031–1038 (1999)

    1999

  15. [24]

    & Jin, Y

    Scaduto, E., Chen, B. & Jin, Y . Satellite-based fire progression mapping: A comprehensive assessment for large fires in northern california. IEEE J. Sel. Top. Appl. Earth Obs. Remote. Sens. 13, 5102–5114 (2020)

    2020

  16. [25]

    Chuvieco, E. et al. Generation and analysis of a new global burned area product based on modis 250 m reflectance bands and thermal anomalies. Earth Syst. Sci. Data (2018)

    2018

  17. [26]

    & Nascetti, A

    Zhang, P., Ban, Y . & Nascetti, A. Total-variation regularized u-net for wildfire burned area mapping based on sentinel-1 c-band sar backscattering data. ISPRS J. Photogramm. Remote. Sens. (2023). 14/16

    2023

  18. [27]

    & Ban, Y

    Zhao, Y . & Ban, Y . Global scale burned area mapping using bi-temporal alos-2 palsar-2 l-band data.IGARSS 2022 - 2022 IEEE Int. Geosci. Remote. Sens. Symp. 695–698 (2022)

    2022

  19. [28]

    & Martinis, S

    Knopp, L., Wieland, M., Rättich, M. & Martinis, S. A deep learning approach for burned area segmentation with Sentinel-2 data. Remote. Sens. 12, 2422 (2020)

    2020

  20. [29]

    M., Nassu, B

    Pereira, G., Fusioka, A. M., Nassu, B. T. & Minetto, R. Active fire detection in Landsat-8 imagery: a large-scale dataset and a deep-learning study. ArXiv abs/2101.03409 (2021)

    2021 arXiv

  21. [30]

    & Martín, M

    Gómez, I. & Martín, M. P. Prototyping an artificial neural network for burned area mapping on a regional scale in mediterranean areas using modis images. Int. J. Appl. Earth Obs. Geoinformation 13, 741–752 (2011)

    2011

  22. [31]

    M., Libonati, R., Trigo, R

    Pinto, M. M., Libonati, R., Trigo, R. M., Trigo, I. F. & DaCamara, C. C. A deep learning approach for mapping and dating burned areas using temporal sequences of satellite images. ISPRS J. Photogramm. Remote. Sens. 160, 260–274, DOI: https://doi.org/10.1016/j.isprsjprs.2019.12...

    2020 doi

  23. [32]

    & Papoutsis, I

    Kondylatos, S., Prapas, I., Camps-Valls, G. & Papoutsis, I. Mesogeos: A multi-purpose dataset for data-driven wildfire modeling in the mediterranean. ArXiv abs/2306.05144 (2023)

    2023 arXiv

  24. [33]

    Finney, M. A. FARSITE: Fire area simulator-model development and evaluation. Res. Pap. RMRS-RP-4, Revis. 2004. Ogden, UT: U.S. Dep. Agric. For. Serv. Rocky Mountain Res. Station. 47 p.4, DOI: 10.2737/RMRS-RP-4 (1998)

    1998 doi

  25. [34]

    W., Wotton, B

    Tymstra, C., Bryce, R. W., Wotton, B. M., Taylor, S. W. & Armitage, O. B.Development and structure of Prometheus: the Canadian Wildland Fire Growth Simulation Model., vol. 417 (Natural Resources Canada, 2010). ISSN: 0831-8247

    2010

  26. [35]

    & Ellsworth, D

    Radke, D., Hessler, A. & Ellsworth, D. Firecast: Leveraging deep learning to predict wildfire spread. In International Joint Conference on Artificial Intelligence (2019)

    2019

  27. [36]

    Singla, S. et al. WildfireDB: An open-source dataset connecting wildfire occurrence with relevant determinants. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2)(2021)

    2021

  28. [37]

    Huot, F. et al. Next day wildfire spread: A machine learning dataset to predict wildfire spreading from remote-sensing data. IEEE Transactions on Geosci. Remote. Sens. 60, 1–13, DOI: 10.1109/TGRS.2022.3192974 (2022). Conference Name: IEEE Transactions on Geoscience and Remote Sensing

    2022

  29. [38]

    & Sullivan, J

    Gerard, S., Zhao, Y . & Sullivan, J. WildfireSpreadTS: A dataset of multi-modal time series for wildfire spread prediction. In Neural Information Processing Systems (2023)

    2023

  30. [39]

    Gorelick, N. et al. Google earth engine: Planetary-scale geospatial analysis for everyone. Remote. Sens. Environ. (2017)

    2017

  31. [40]

    Abatzoglou, J. T. Development of gridded surface meteorological data for ecological applications and modelling. Int. J. Climatol. 33 (2013)

    2013

  32. [41]

    Clough, S. A. et al. Atmospheric radiative transfer modeling: a summary of the AER codes. J. Quant. Spectrosc. Radiat. Transf. 91, 233–244, DOI: 10.1016/j.jqsrt.2004.05.058 (2005)

    2005 doi

  33. [42]

    S., Peterson, D

    Littell, J. S., Peterson, D. L., Riley, K. L., Liu, Y . & Luce, C. H. A review of the relationships between drought and forest fire in the united states. Glob. Chang. Biol. 22 (2016)

    2016

  34. [43]

    Brown, P. T. et al. Climate warming increases extreme daily wildfire growth risk in california. Nature 621, 760–766 (2023)

    2023

  35. [44]

    A., Hessburg, P

    Povak, N. A., Hessburg, P. F. & Salter, R. B. Evidence for scale-dependent topographic controls on wildfire spread. Ecosphere (2018)

    2018

  36. [45]

    NASA shuttle radar topography mission global 1 arc second

    JPL, N. NASA shuttle radar topography mission global 1 arc second. NASA EOSDIS Land Process. DAAC DOI: 10.5067/MEaSUREs/SRTM/SRTMGL1.003 (2013)

    2013 doi

  37. [46]

    M., Abercrombie, S

    Sulla-Menashe, D., Gray, J. M., Abercrombie, S. P. & Friedl, M. A. Hierarchical mapping of annual global land cover 2001 to present: The MODIS collection 6 land cover product. Remote. Sens. Environ. 222, 183–194 (2019)

    2019

  38. [47]

    Friedl, M. A. & Sulla-Menashe, D. MODIS/terra+aqua land cover type yearly l3 global 500m SIN grid v061. NASA EOSDIS Land Process. DAAC DOI: 10.5067/MODIS/MCD12Q1.061 (2022)

    2022 doi

  39. [48]

    Ts-satfire, DOI: 10.34740/KAGGLE/DSV/8675553 (2024)

    Zhao, Y . Ts-satfire, DOI: 10.34740/KAGGLE/DSV/8675553 (2024)

    2024

  40. [49]

    & Brox, T

    Ronneberger, O., Fischer, P. & Brox, T. U-Net: Convolutional networks for biomedical image segmentation. In MICCAI (2015)

    2015

  41. [50]

    Hatamizadeh, A. et al. Unetr: Transformers for 3d medical image segmentation (2021). 2103.10504. 15/16

    2021 arXiv

  42. [51]

    Self-supervised pre-training of swin transformers for 3d medical image analysis

    Tang, Y .et al. Self-supervised pre-training of swin transformers for 3d medical image analysis. 2022 IEEE/CVF Conf. on Comput. Vis. Pattern Recognit. (CVPR)20698–20708 (2021)

    2021

  43. [52]

    Liu, Z. et al. Swin transformer: Hierarchical vision transformer using shifted windows. 2021 IEEE/CVF Int. Conf. on Comput. Vis. (ICCV) 9992–10002 (2021). Acknowledgements The research is part of the project ‘Sentinel4Wildfire’ funded by Formas, the Swedish research council fo...

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.