Themeda-Based Spatiotemporal Deep Learning for Predicting Vegetation Dynamics and Fire-Induced Land Cover Change in Northern Australia Using ConvLSTM and Temporal U-Net Models

The tropical savannas of Northern Australia, dominated by the perennial C4 grass Themeda triandra (kangaroo grass), are among the most fire-prone and ecologically significant biomes in the Southern Hemisphere. These savannas cover approximately 1.9 million km² and experience annual fire frequencies affecting 20–40% of the total area. Understanding and predicting vegetation dynamics and fire-induced land cover change in these landscapes presents substantial challenges for land managers, conservation practitioners, and carbon accounting frameworks. This study presented a novel spatiotemporal deep learning framework integrating Convolutional Long Short-Term Memory (ConvLSTM) and Temporal U-Net architectures to simulate and map fire-driven land cover change across Northern Australian tropical savannas. The framework incorporated multi-source satellite time series spanning 2019–2022 MODIS NDVI composites (250 m), Landsat 8/9 OLI surface reflectance (30 m), North Australia Fire Information (NAFI) fire scar records, Bureau of Meteorology rainfall grids, and Sentinel-2 MSI data. A novel Themeda Vegetation Index (TVI) was developed from the spectral properties of curing C4 grass tissue and validated against 84 spatially independent field plots (Pearson r = 0.87, p < 0.001 on hold-out subset), outperforming NDVI (R² = 0.621) and EVI (R² = 0.648). Using a spatially blocked train/test partitioning design, 185 non-overlapping 100×100 km geographic tiles were used, ensuring no spatial autocorrelation between partitions. The ensemble model achieved an overall accuracy of 88.9%, Cohen's Kappa of 0.871, and a weighted F1-score of 0.884. NDVI prediction attained a root mean square error (RMSE) of 0.037 NDVI units and a mean absolute error (MAE) of 0.028 NDVI units. The ConvLSTM's forget gate mechanism accurately encoded abrupt fire-induced resets in vegetation states. The Temporal U-Net's learned temporal attention weights identified the dry-season months (June–September) as diagnostically critical for discriminating fire scars without explicit supervision. Post-fire recovery analysis confirmed a strong rainfall gradient: areas receiving>1,200 mm yr⁻¹ recovered to 93% of pre-fire NDVI within 24 months, compared to 57% in transitional low-rainfall zones. Critical limitations were noted. The 2019–2022 training period coincided with a sustained La Niña event (mean Southern Oscillation Index = +11.8 for 2020–2022; Niño 3.4 anomaly = −0.9°C). Recovery rates represented La Niña-period upper estimates, potentially 8–15 percentage points above climatological averages in low-rainfall zones. Model outputs were characterized as enabling spatial technology providing inputs relevant to the Savanna Burning Emissions Reduction (SBER) carbon accounting methodology, subject to project-scale validation by the Clean Energy Regulator. Sub-regional accuracy ranged from 84.1% (Barkly Tablelands) to 91.2% (Top End NT), confirming operationally viable performance across the heterogeneous study area.

Themeda triandra, ConvLSTM, Temporal U-Net, spatiotemporal deep learning, NDVI time series prediction, savanna fire ecology, Northern Australia remote sensing

1. de Queiroz, R. F. P., d’Oliveira, M. V. N., Rezende, A. V., & de Alencar, P. A. L. (2023). Estimation of aboveground biomass stock in tropical savannas using photogrammetric imaging. Drones, 7(7), 493. https://doi.org/10.3390/drones7080493 [Google Scholar] [Crossref]

2. Dubey, P., & Dubey, P. (2025). Bridging spatiotemporal wildfire prediction and decision modeling using transformer networks and fuzzy inference systems. MethodsX, 15, 103498. https://doi.org/10.1016/j.mex.2025.103498 [Google Scholar] [Crossref]

3. Edwards, A., Archer, R., De Bruyn, P., Evans, J., Lewis, B., Vigilante, T., Whyte, S., & Russell-Smith, J. (2021). Transforming fire management in northern Australia through successful implementation of savanna burning emissions reductions projects. Journal of Environmental Management, 290, 112568. https://doi.org/10.1016/j.jenvman.2021.112568 [Google Scholar] [Crossref]

4. Garg, P., Roche, T., Eden, M., Matz, J., Oakes, J. M., Bellini, C., & Gollner, M. J. (2021). Effect of moisture content and fuel type on emissions from vegetation using a steady state combustion apparatus. International Journal of Wildland Fire, 30, WF20118. https://doi.org/10.1071/WF20118 [Google Scholar] [Crossref]

5. Gargiulo, M., Dell’Aglio, D. A. G., Iodice, A., Riccio, D., & Ruello, G. (2020). Integration of Sentinel-1 and Sentinel-2 Data for Land Cover Mapping Using W-Net. Sensors, 20(10), 2969. https://doi.org/10.3390/s20102969 [Google Scholar] [Crossref]

6. Garnot, V., & Landrieu, L. (2021). Panoptic segmentation of satellite image time series with convolutional temporal attention networks. LASTIG, Université Gustave Eiffel, ENSG, IGN. [Google Scholar] [Crossref]

7. Hemraj Bhattarai, Maria Val Martin, Stephen Sitch, & David H. Y. Yung. (2025, March). Global patterns and drivers of climate-driven fires in a warming world. EGUsphere. https://doi.org/10.5194/egusphere-2025-804 [Google Scholar] [Crossref]

8. Luck, L., Hutley, L. B., Calders, K., & Levick, S. R. (2020). Exploring the Variability of Tropical Savanna Tree Structural Allometry with Terrestrial Laser Scanning. Remote Sensing, 12(23), 3893. https://doi.org/10.3390/rs12233893 [Google Scholar] [Crossref]

9. Nolan, R. H., Boer, M. M., Collins, L., Resco de Dios, V., Clarke, H., Jenkins, M., Kenny, B., & Bradstock, R. A. (2020). Causes and consequences of eastern Australia’s 2019–20 season of mega-fires. Global Change Biology, 26(3), 1039–1041. https://doi.org/10.1111/gcb.14987 [Google Scholar] [Crossref]

10. Pickering, B. J., Kultaev, D., Holyland, B., Ababei, D., & Penman, T. D. (2025). The changing risk of fire to human and environmental assets under climate induced altered fire regimes in south-east Australia. FLARE Wildfire Research, School of Agriculture, Food and Ecosystem Sciences, The University of Melbourne. Published online June 27, 2025. [Google Scholar] [Crossref]

11. Veiga, R. M. da, von Randow, C., Burton, C., Kelley, D. I., Cardoso, M., & Morelli, F. (2025). Review article: Fire emissions in the Brazilian Cerrado – dynamics, estimates, management, and their role in the global carbon budget. Biogeosciences. Published online September 23, 2025. [Google Scholar] [Crossref]

12. Wang, S., Guan, K., Zhang, C., Lee, D., Margenot, A. J., Ge, Y., & Ainsworth, E. A. (2022). Using soil library hyperspectral reflectance and machine learning to predict soil organic carbon: Assessing potential of airborne and spaceborne optical soil sensing. Remote Sensing of Environment, 271, Article 112914. https://doi.org/10.1016/j.rse.2022.112914 [Google Scholar] [Crossref]

13. Zhang, Z., Song, Q., Duan, M., Liu, H., Huo, J., & Han, C. (2025). Deep Learning Model for Precipitation Nowcasting Based on Residual and Attention Mechanisms. Remote Sensing, 17(7), 1123. https://doi.org/10.3390/rs17071123 [Google Scholar] [Crossref]

• What the Desert Fathers Teach Data Scientists: Ancient Ascetic Principles for Ethical Machine-Learning Practice
• Comparative Analysis of Some Machine Learning Algorithms for the Classification of Ransomware
• Comparative Performance Analysis of Some Priority Queue Variants in Dijkstra’s Algorithm
• Transfer Learning in Detecting E-Assessment Malpractice from a Proctored Video Recordings.
• Dual-Modal Detection of Parkinson’s Disease: A Clinical Framework and Deep Learning Approach Using NeuroParkNet