Identifying native grasslands and key phenological stages using time series Sentinel-2 data and deep learning models
Bibliographic record
Abstract
Canadian prairies are among the world’s most endangered ecosystems, and the identification of native grasslands is crucial for supporting grassland management and wildlife conservation. However, the current amount, distribution, and dynamic changes of native grassland remain uncertain, partly due to the difficulty of separating native grassland with other land cover types, especially tame grassland in landcover classification products. This study aims to identify native grasslands in the Mixed Grasslands ecoregion of Saskatchewan using Sentinel-2 Normalized Difference Vegetation Index (NDVI) time series data through deep learning and multi-temporal approaches. Three classifiers, one-dimensional convolutional (Conv1D), Attention Long Short-Term Memory (At-LSTM), and Random Forest (RF), were applied to distinguish the native and tame grasslands, analyzing the key phenological stages. The results show that the Conv1D-based model outperformed the others in both the South (accuracy: 0.88; F1 score: 0.87) and North (accuracy: 0.78; F1 score: 0.77) sub-regions. In contrast, the At-LSTM and RF models performed worse, particularly in the North sub-region, with F1 scores of 0.64 and 0.68, respectively. The early July period (Day of Year 170 to 200) was critical for distinguishing native and tame grassland landcover, especially in the South sub-region. Additionally higher resolution data (5-day intervals) generally improved model accuracy compared to 10-day and 30-day intervals. Overall, the study demonstrates the effectiveness of time series NDVI data and deep learning approaches for identifying native grasslands, offering valuable information to assist in grassland ecosystem management and conservation strategies in the Canadian prairies.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".