Uncertainty assessment and comparison of vegetation indices, surface emissivity models and split-window algorithms used to estimate surface temperature from satellite images
Bibliographic record
Abstract
Surface temperature is a key variable to understand and characterize biophysical processes in the soil-vegetation-atmosphere system. It needs to be estimated precisely at local and regional scales for many applications such as crop growth models, epidemiology of diseases, and monitoring of heat islands, which could be deeply influenced by microclimatic conditions. Several vegetation indices (VIs), surface emissivity models, and split-window algorithms are used to estimate surface temperature from satellite images, however their uncertainties are not always well known. This study aims to evaluate uncertainties and compare VIs, surface emissivity models, and split-window algorithms used to estimate surface temperature from satellite images. The Advanced Very High Resolution Radiometer (AVHRR) was used to acquire satellite images in the south of the province of Quebec, Canada. These images were used to estimate VIs like the modified soil adjusted vegetation index 2 (MSAVI2), the normalized difference vegetation index (NDVI), the soil-adjusted vegetation index (SAVI), and the optimized soil-adjusted vegetation index (OSAVI), and to estimate the percent vegetation cover (PVC), surface emissivity derived from the models of Van De Griend and Owe (1993) and Sobrino and Raissouni (2000), and surface temperature derived from the split-window algorithms of Price (1984), Becker and Li (1990), Coll et al. (1994), Ulivieri et al. (1994), and McClain et al. (1985). Split-window algorithms were estimated using scenarios combining different input parameters related to VIs, PVC, and surface emissivity models in order to assess their influence on the uncertainty of surface temperature. The assessment and comparison of VIs, PVC, surface emissivity models, and split-window algorithms were based on: their uncertainties (formulated using the law of uncertainty propagation), spatial correlation between models, surface temperature derived from airborne infrared thermography, in-situ meteorological data, and an overall performance coefficient (OPC). The MSAVI2 is the VI which recorded the lowest uncertainty values. MSAVI2 values on agricultural surfaces of the study area ranged from 0.36 to 0.82 (mean = 0.66, σ = ±0.064), with uncertainties from ±0.018 to ±0.033 (mean = ±0,023; σ = ±0.002). The PVC based on MSAVI2 is also the one that registered the lowest uncertainties (u) (0 ≤ u ≤ ±0,263; mean = ±0.182; σ = ±0.039). The surface emissivity model of Van De Griend and Owe (1993), which uses the MSAVI2 as input variable, is the one that recorded the lowest uncertainties (±0.001 ≤ u ≤ ±0.011; mean = ±0.002; σ = ±0.000) and had the highest correlations with all the other surface emissivity models. The differences between the surface temperature values from the various split-window algorithms reached 7.69 K. The maximum uncertainty of surface temperature ranged from ±0.120 K to ±4.346 K between split-window algorithms. The OPC of split-window algorithms varied between 51.43% and 91.43%. The split-window algorithm of Coll et al. (1994) using the surface emissivity model of Van De Griend and Owe (1993) and the MSAVI2 allowed the best estimation of surface temperature (OPC = 91.43%). According to this model, surface temperature was estimated with an average uncertainty of ±0.389 K; experimental uncertainty and bias in reference to the surface temperature derived from airborne infrared thermography were respectively ±0.454 K and 0.521 K. The high variation of uncertainties and OPC between models used to estimate surface temperature show the importance of the choice of split-window algorithms, surface emissivity models, and VIs in this estimation. The formulation of uncertainties and their validation, as well as the concept of OPC, are major contributions of this study.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".