Leveraging Multi-Source Remote Sensing Images and Deep Learning to Map Caribou Lichens
Bibliographic record
Abstract
Given the importance of lichens for caribou during winters, disturbances to caribou lichens may affect caribou migration and distribution patterns, and possibly lead to their population declines. Using remote sensing (RS) data, it is possible to monitor forage lichen cover efficiently over large and inaccessible areas. In this thesis, three significantly challenging problems in lichen cover mapping (especially Cladonia spp.) in different regions of Canada (Newfoundland and Labrador, Quebec, and Northwest Territory) are addressed: 1) lichen mapping using limited ground-validation data; 2) deploying a single, universal model for lichen mapping in non-atmospherically corrected RS images; and 3) lichen fractional cover mapping over rocky landscapes where non-lichen features are spectrally similar to caribou lichens. To address these three challenges, we used a wide variety of RS images (micro-plot images, high-resolution aerial optical images, high-resolution satellite (HRS) images (WorldView-2 and -3), and airborne hyperspectral imagery (AVIRIS-NG)), and different advanced deep learning (DL) models. In the first phase of this research, we found that semi-supervised DL for lichen mapping can improve lichen mapping compared to fully supervised learning in the presence of limited training data. In the second phase, our experiments showed the high potential of Generative Adversarial Networks (GANs) for normalizing non-atmospherically corrected HRS images for lichen mapping using a single, universal lichen detector model under different atmospheric conditions. In the last phase of this research, we found that using AVIRIS-NG hyperspectral imagery, lichen fractional cover mapping was more accurate than using HRS imagery over a rocky landscape where non-lichen bright features resembled lichen species of interest (Cladonia spp.). The studies conducted in this thesis are significant as they open new insights into how the use of state-of-the-art DL solutions and multiple sources of RS data can improve the quality of lichen mapping under different challenging circumstances. One of the most important contributions of this study is that all the methods and results presented in this thesis can also be readily applied to other vegetation mapping applications using RS data as the methods were designed in such a way that they do not depend on a specific land cover type or application.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| 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".