Distance Transform-Based Spatiotemporal Model for Approximating Missing NDVI from Satellite Data
Bibliographic record
Abstract
One widely used method for analyzing vegetation growth from satellite imagery is the Normalized Difference Vegetation Index (NDVI), a key metric for assessing vegetation dynamics. NDVI varies not only spatially but also temporally, which is essential for analyzing vegetation health and growth patterns over time. High-resolution, cloud-free satellite images, particularly from publicly available sources like Sentinel, are ideal for this analysis. However, such images are not always available due to cloud and shadow contamination. To address this limitation, we propose a model that integrates both the temporal and spatial aspects of the data to approximate the missing or contaminated regions. In this method, we separately approximate NDVI using spatial and temporal components of the time-varying satellite data. Spatial approximation near the boundary of the missing data is expected to be more accurate, while temporal approximation becomes more reliable for regions further from the boundary. Therefore, we propose a model that leverages the distance transform to combine these two methods into a single, weighted model, which is more accurate than either method alone. We introduce a new decay function to control this transition. We evaluate our spatiotemporal model for approximating NDVI across 16 farm fields in Western Canada from 2018 to 2023. We empirically determined the best parameters for the decay function and distance-transform-based model. The results show a significant improvement compared to using only spatial or temporal approximations alone (up to a 263% improvement as measured by RMSE relative to the baseline). Furthermore, our model demonstrates a notable improvement compared to simple combination (up to 51% improvement as measured by RMSE) and Spatiotemporal Kriging (up to 28% improvement as measured by RMSE). Finally, we apply our spatiotemporal model in a case study related to improving the specification of the peak green day for numerous fields.
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 | 0.001 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.003 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| 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".